Legal and ethical implications ofapplications based on agreementtechnologies: the case of auction-based roadintersections

Abstract

Agreement technologies refer to a novel paradigm for the construction of distributed intelligent systems, where autonomous software agents negotiate to reach agreements on behalf of their human users. Smart Cities are a key application domain for agreement technologies. While several proofs of concept and prototypes exist, such systems are still far from ready for being deployed in the real-world. In this paper we focus on a novel method for managing elements of smart road infrastructures of the future, namely the case of auction-based road intersections. We show that, even though the key technological elements for such methods are already available, there are multiple non-technical issues that need to be tackled before they can be applied in practice. For this purpose, we analyse legal and ethical implications of auctionbased road intersections in the context of international regulations and from the standpoint of the Spanish legislation. From this exercise, we extract a set of required modifcations, of both technical and legal nature, which need to be addressed so as to pave the way for the potential real-world deployment of such systems in a future that may not be too far away.

Full text

1 23
Artificial Intelligence and Law
ISSN 0924-8463
Volume 28
Number 4
Artif Intell Law (2020) 28:385-414
DOI 10.1007/s10506-019-09259-8
Legal and ethical implications of
applications based on agreement
technologies: the case of auction-based road
intersections
José-Antonio Santos, Alberto Fernández,
Mar Moreno-Rebato, Holger Billhardt,
José-A.Rodríguez-García & Sascha
Ossowski

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Nature B.V.. This e-offprint is for personal
use only and shall not be self-archived
in electronic repositories. If you wish to
self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.

Vol.:(0123456789)
Artificial Intelligence and Law (2020) 28:385–414
https://doi.org/10.1007/s10506-019-09259-8
1 3
Legal andethical implications ofapplications based
onagreement technologies: thecase ofauction‑based road
intersections
José‑AntonioSantos1 · AlbertoFernández2 · MarMoreno‑Rebato1 ·
HolgerBillhardt2 · José‑A.Rodríguez‑García1 · SaschaOssowski2
Published online: 30 September 2019
© Springer Nature B.V. 2019
Abstract
Agreement technologies refer to a novel paradigm for the construction of distributed
intelligent systems, where autonomous software agents negotiate to reach agree-
ments on behalf of their human users. Smart Cities are a key application domain for
agreement technologies. While several proofs of concept and prototypes exist, such
systems are still far from ready for being deployed in the real-world. In this paper
we focus on a novel method for managing elements of smart road infrastructures of
the future, namely the case of auction-based road intersections. We show that, even
though the key technological elements for such methods are already available, there
are multiple non-technical issues that need to be tackled before they can be applied
in practice. For this purpose, we analyse legal and ethical implications of auction-
based road intersections in the context of international regulations and from the
standpoint of the Spanish legislation. From this exercise, we extract a set of required
modifications, of both technical and legal nature, which need to be addressed so as
to pave the way for the potential real-world deployment of such systems in a future
that may not be too far away.
Keywords Agreement technologies· Autonomous vehicles· Intersection
management· Intelligent transportation systems· Ethical and legal aspects· Spanish
law
* José-Antonio Santos
joseantonio.santos@urjc.es
1 Faculty ofLaw andSocial Sciences, University Rey Juan Carlos, Madrid, Spain
2 CETINIA, University Rey Juan Carlos, Móstoles,Madrid, Spain
Author's personal copy

386
J.-A.Santos et al.
1 3
1 Introduction
The transactions and interactions among people in modern societies are increas-
ingly mediated by computers. From email, over social networks, to virtual worlds,
the way people work and enjoy their free time is changing dramatically. The
resulting networks are usually large in scale, involving huge numbers of interac-
tions, and are open for the interacting entities to join or leave at will. People are
often supported by software components of different complexity to which some
of the corresponding tasks can be delegated. In practice, such systems cannot be
built and managed based on rigid, centralised client–server architectures, but call
for more flexible and decentralised means of interaction. Accordingly, Brynjolfs-
son and McAfee (2016, p 96) say: “We can’t predict exactly what new insights,
products, and solutions will arrive in the coming years, but we are fully confi-
dent that they’ll be impressive. The second machine age will be characterized by
countless instances of machine intelligence and billions of interconnected brains
working together to better understand and improve our world. It will make mock-
ery out of all that came before.” A new era that is in its infancy and that moves at
a vertiginous pace towards an uncertain future.
In this context, machines with ability to show intelligent behaviour have turned
what was strange to humans into an artificial naturalness. That is, these increasingly
autonomous systems are becoming part of our daily routine and life is not under-
stood without them. As Billhardt et al. (2015) argue, Artificial Intelligence (AI)
challenges human work and tasks will be increasingly carried out in close collabora-
tions between humans and machines creating so called human-agent teams.
From the technical point of view, the emerging field of agreement technologies
(AT) (Ossowski etal. 2013) aims at supporting this vision. It aims at next-gener-
ation open distributed systems, where interactions between software components
are based on the concept of agreement, and which enact two key mechanisms: a
means to specify the space of agreements that the agents can possibly reach, and
an interaction model by means of which agreements can be effectively reached.
Autonomy, interaction, mobility and openness are key characteristics that are
tackled from a theoretical and practical perspective.
There is a broad variety of domains where the potential of AT becomes apparent
(see Part VII of Ossowski etal. 2013). In these domains, the choices and actions
of a large number of autonomous stakeholders need to be coordinated, and interac-
tions can be regulated, by some sort of intelligent computing infrastructure (Omicini
etal. 2004), through some sort of institutions and institutional agents (Fornara etal.
2013), or simply by strategically providing information in an environment with a
significant level of uncertainty (Centeno etal. 2009). The advent of intelligent road
infrastructures, with support for vehicle-to-vehicle and vehicle-to-infrastructure
communications, make smart transportation a challenging field of application for
AT, as it allows for a decentralised coordination of individually rational commut-
ers. The work by Vasirani and Ossowski (2012) on networks of auction-based road
intersections, which account for preferential use of road infrastructures depending
on drivers’ profiles and preferences, is an excellent example for this.
Author's personal copy

387
1 3
Legal andethical implications ofapplications based on…
In this context, interaction between human beings and machines can enable
an individualised and at the same time fair access to public resources, which will
increase the efficiency of such smart infrastructures and thus has the potential to
improve both individual and social welfare. Still, such a scenario can be considered
as yet another case of a cultural revolution fuelled by technological progress, with
potentially ambivalent consequences. On the one hand, it shows a wide range of
possibilities for action and possibilities of choice that can help improve the level of
welfare of drivers. For example, in semi-autonomous driving systems, the passenger
would be legally authorised to be distracted by phone calls or use social networks.
Autonomous vehicles may also allow for carrying goods in less time by maintaining
a more constant speed and without the obligatory breaks required for drivers. On the
other hand, all sorts of driving-related jobs (taxi, bus, lorry, etc.) would be lost. In
addition, decreased traffic accidents would probably mean fewer insurance claims
and fewer repairs to the garages. In any case, it is important to notice that, even
though the technology is (almost) ready for those visions to become reality, there
are numerous ethical and legal issues that need to be addressed before experimenta-
tion with those systems in the real world will become an option. In this matter, it is
necessary to remember that no decision made in the field of law and ethics is trivial:
decisions and their consequences need to be analysed within the force field among
law, technology and ethics.
As Casanovas (2013) outlines, especially novel applications based on AT have a
wide range of legal question and problems, which must be addressed if such systems
are ever meant to be applied to the real-world. In this paper we focus on AT applied
to smart intersection management, so as to illustrate this claim through a case study.
In particular, setting out from the auction-based intersection managers introduced by
Vasirani and Ossowski (2012), we identify and analyse legal and ethical problems
related to the potential real-world deployment of this AT-based technology. Fur-
thermore, we design and discuss solutions to overcome these problems from both
perspectives: what modifications would be needed in the smart traffic infrastructure
itself, and what changes in legal regulations would be required, if we were to deploy
auction-based intersections in our society in a properly regulated manner.
This approach is related but intrinsically different to the inverse process of creat-
ing systems or applications that comply with legal requirements and which is usually
treated in the field of legal requirements engineering (e.g. Boella etal. 2014; Rabi-
nia and Ghanavati 2018). In legal requirements engineering research the focus is
usually on specifying languages and methods to specify legal regulations, in order to
identify their implications on a new system. Here, we want to specify those aspects
of a given system that may need modifications or adaptations in the existing regula-
tions. We think this is a common problem, which is often not treated in a systematic
manner. Especially in academic research, new solution approaches are usually pro-
posed without analysing their viability from legal or ethical points of view. Thus, a
posterior analysis in this direction should be or has to be carried out before transfer-
ring such approaches to commercial systems.
Our paper is structured as follows: Sect.2 describes the technological founda-
tions of an envisioned future road traffic infrastructure, where intersections are man-
aged by smart controllers while drivers can reserve space/time slots at intersections
Author's personal copy

388
J.-A.Santos et al.
1 3
according to their needs, and highlights those aspects of the approach that could
affect or could be affected by legal issues. Section3 analyses legal and ethical impli-
cations of such auction-based road intersections in the context of international regu-
lations and from the standpoint of the Spanish legislation. From this exercise, in
Sect.4 we identify aspects from the proposal presented in Sect.2 that do not comply
with the current regulations and discuss possible solutions, of both technical and
ethical/legal nature, which would need to be implemented in order to pave the way
for the potential real-world deployment of such systems in a future that may not be
too far away. Section5 summarises our findings and points to future work.
2 Coordination oftrac ows throughintelligent intersections
Removing the human driver from the control loop through the use of autonomous
vehicles integrated with an intelligent road infrastructure can be considered as the
ultimate, long-term goal of the set of systems and technologies grouped under the
name of Intelligent Transportation Systems (ITS). The advantages of such an inte-
gration span from improved road safety to a more efficient operational use of the
transportation network. For instance, vehicles can exchange critical safety informa-
tion with the infrastructure, so as to recognise high-risk situations in advance and
therefore to alert drivers. Furthermore, traffic signal systems can communicate sig-
nal phase and timing information to vehicles to enhance the use of the transportation
network.
In this regard, many authors have recently paid attention to the potential of a
tighter integration of autonomous vehicles with the infrastructure for intersection
management (Namazi etal. 2019). Dresner and Stone (2008) introduce the reser-
vation-based control system, where an intersection is regulated by an intelligent
software component, called intersection manager, which assigns reservations of
space and time to each autonomous vehicle intending to cross the intersection. Each
vehicle is operated by another software agent, called driver agent. When a vehicle
approaches an intersection, the driver agent asks the intersection manager to reserve
the necessary space–time slots to safely cross the intersection. The latter simulates
the vehicle´s trajectory inside the intersection and informs the driver agent whether
its request is in conflict with the already confirmed reservations. If such a conflict
does not exist, the driver agent stores the reservation details and tries to meet them;
otherwise it may try again at a later time. The authors show through simulations that
insituations of balanced traffic, if all vehicles are autonomous, their delays at the
intersection are drastically reduced compared to traditional traffic lights.
Vasirani and Ossowski (2012) extend the above model for reservation-based
intersection control along two major lines. Firstly, for single intersections, they
elaborate an auction-based policy for the allocation of reservations to vehicles that
grants preferential access to vehicles based on the drivers´ different attitudes regard-
ing their travel times. Secondly, for network of intersections, they set up a compu-
tational market where intersection managers compete with each other as suppliers
of reservations, and selfishly adapt the reserve prices of the auctions that they run,
so as to match the actual demand. In this manner, traffic flows are better distributed
Author's personal copy

389
1 3
Legal andethical implications ofapplications based on…
among the road network, and the social cost of applying a preferential assignment
strategy at intersection-level is reduced. In the following we summarise the dynam-
ics and functionalities of this approach, as it sets the ground for the ethical and legal
questions analysed in the rest of this article.
2.1 Mechanism forsingle intersection
For a single reservation-based intersection, the problem that an intersection manager
has to solve comes down to allocating reservations among a set of drivers in such
a way that a specific objective is maximised. This objective can be, for instance,
minimising the average delay caused by the presence of the regulated intersection.
In this case, the simplest policy to adopt is allocating a reservation to the first agent
that requests it, as occurs with the first-come first-served (FCFS) policy proposed
by Dresner and Stone in their original work. Another work in line with this objec-
tive takes inspiration from adversarial queuing theory for the definition of several
alternative control policies that aim at minimising the average delay (Vasirani and
Ossowski 2009).
However, these policies ignore the fact that in the real world, depending on the
context and their personal situation, people value the importance of travel times and
delays quite differently. Since processing the incoming requests to grant the associ-
ated reservations can be considered as a process of assigning resources to agents that
request them, intersection managers are supposed to allocate the disputed resources
to the agents that value them the most. In line with findings from mechanism design,
it is assumed that the more a human driver is willing to pay for the desired set of
space–time slots, the more they value the good. Thus, the policy for the allocation of
resources relies on combinatorial auctions.
The goods assigned through these combinatorial auctions are the right to use cer-
tain space inside the intersection at a given time. An intersection is modelled as a
discrete matrix of space slots. Let S be the set of the intersection space slots, and T
the set of future time-steps, then the set of items that a bidder can bid for is I = S × T.
Due to the nature of the problem, a bidder is only interested in bundles of items over
the set I. As Fig.1 illustrates, in the absence of acceleration in the intersection, a res-
ervation request implicitly defines which space slots at which time the driver needs
in order to pass through the intersection. A bid over a bundle of items is implicitly
defined by the reservation request. Given the parameters arrival time, arrival speed,
lane and type of turn, the auctioneer (i.e., the intersection manager) is able to deter-
mine which space slots are needed at which time. In addition, the value of its bid, i.e.
the amount of money that the driver is willing to pay for the requested reservation,
must be included into a vehicle’s reservation request.
Figure2 shows the interaction protocol used to regulate the combinatorial auc-
tion. It starts with the auctioneer waiting for bids for a certain amount of time.
Once the new bids are collected, they constitute the bid set. Then, the auction-
eer executes an anytime approximation algorithm to solve the winner determina-
tion problem (WDP), thus determining the winner set with the bids whose res-
ervation requests have been accepted. During the WDP algorithm execution, the
Author's personal copy

390
J.-A.Santos et al.
1 3
auctioneer still accepts incoming bids, but they will only be included in the bid
set of the next round. The auctioneer sends a CONFIRMATION message to all
bidders that submitted the bids contained in the winner set, while a REJECTION
message is sent to the bidders that submitted the remaining bids. Then a new
Fig. 1 Bundle of items for a reservation request (from Vasirani and Ossowski 2012)
Fig. 2 Auction policy proposed by Vasirani and Ossowski (2012)
Author's personal copy

391
1 3
Legal andethical implications ofapplications based on…
round begins, and the auctioneer collects new incoming bids for a certain amount
of time.1
Through a set of simulated experiments, Vasirani and Ossowski (2012) confirmed
that the policy based on combinatorial auctions (CA) enforces an inverse relation
between the amounts spent by the bidders and their delay (the increase in travel time
due to the presence of the intersection). That is, the more money a driver is will-
ing to spend for crossing the intersection, the faster will be its transit through it.
They noticed that even with a theoretically infinite amount of money, a driver cannot
experience zero delay when approaching an intersection, as the travel time is influ-
enced by slower potentially poorer vehicles in front of it. They also confirmed that
the CA policy has a higher average delay compared to FCFS strategy, since it grants
a reservation to the driver that values it the most, rather than maximising the number
of granted requests. This social cost of CA is the bigger, the higher the load of the
intersection, i.e. the more drivers compete for reservations. The extension of the CA
mechanism to multiple intersections described in the sequel aims at reducing this
social cost of giving preference to drivers with a high valuation of time.
2.2 Mechanisms formultiple intersections
In case of a single intersection, a driver agent simply needs to decide on the pre-
ferred value for the bid that it submits to the auctioneer. The decision space of a
driver agent in an urban road network with multiple intersections is much broader:
complex and mutually dependent decisions must be taken such as route choice and
departure time selection. Therefore, this scenario opens up new possibilities for
intersection managers to affect the behaviour of drivers. For example, an intersec-
tion manager may be interested in influencing the collective route choice performed
by the drivers, using variable message signs, information broadcast, or individual
route guidance systems, so as to evenly distribute the traffic over the network. This
problem is called traffic assignment.
In the Competitive Traffic Assignment/Combinatorial Auction strategy (CTA-
CA), each intersection manager runs the CAs described in the previous section, but
includes a reserve price, i.e. it announces a minimum price that it charges for the
reservations that it sells. At each time t, there may be different reserve prices pt(l)
for each of the incoming link l of the intersection. The reserve prices of the inter-
sections are made available to driver agents in real time, which then choose their
routes according to their personal preferences about travel times as well as the cur-
rent monetary costs.
Each intersection manager competes with all others for the supply of the reserva-
tions that are traded. It applies a simple reserve price update rule that aims at attract-
ing drivers to the intersection when traffic demand is low, and at deterring them
1 The protocol also allows vehicles to cancel a reservation that they already have acquired for a certain
time t. This may happen, for instance, when a driver realises that, due to changing traffic conditions, it is
likely to arrive at the intersection at some time t’ > t. The auction protocol includes additional constraints
that keep agents from strategizing based on this option.
Author's personal copy

392
J.-A.Santos et al.
1 3
from choosing routes that transit through the intersection when it is oversaturated.
At time t + 1, the reserve price on link l is computed as follows:
In this formula, the constant s(l) represents the optimal amount of vehicles that
can cross the intersection coming from link l. The excess demand zt(l) expresses the
difference between total demand at time t and supply s(l) on link l (notice that zt(l)
becomes negative when the supply on link l exceeds the demand). By dynamically
adjusting the reserve prices of the network’s intersection managers, traffic flows are
better distributed among the network, thus avoiding an over-saturation of bottleneck
intersections.
Vasirani and Ossowski (2012) evaluated the approach using simulations upon a
topology that resembled the high capacity road network of the city of Madrid in
Spain. Among others, they compared CTA-CA with a FCFS strategy, where each
intersection manager assigns reservations in isolation on a first-come first-served
basis. Notice that differently from CTA-CA, on the one hand, in FCFS there is no
notion of social cost as all drivers are treated equally, but on the other hand FCFS
cannot benefit from the traffic assignment effects of the reserve prices of CTA-CA.
In the experiments, with the FCFS strategy, the drivers´ route choice model simply
selects the route with minimum expected travel time at free flow, since there is no
notion of price. For the evaluation of the CTA-CA strategy, it is assumed that driv-
ers choose the most preferred route they can afford. The experiments measure the
moving average of the travel time, that is, how the average travel time of the entire
population of drivers, computed over all the origin–destination pairs, evolves during
the simulation.
The results show that CTA-CA outperforms FCFS, while maintaining the inverse
relation between the amounts spent by driver agents and their vehicle’s delay. The
performance gain of CTA-CA is the bigger, the higher the load at the intersections.
These results indicate that preferential assignment methods such as CTA-CA can
provide an individualised access to the public traffic infrastructure that is tailored to
a driver’s preferences and needs, while their negative effect on social welfare (i.e.
increased average travel time) can be mitigated by adequately designing the assign-
ment mechanisms.
2.3 Description ofsystem operation
While the previous sections summarised the technical feasibility of auction-based
intersections and studied their physical performance from both individual and social
perspectives, we now aim at analysing the compliance of the proposed intersection
management system with current legal regulations. The objective is to identify nec-
essary and possible modifications, not only in the legal regulations but also in the
smart infrastructure itself, that would allow to properly implement the system in
society.
p
t+1(l)=pt(l)+pt(l)
zt(l)
s
(
l
)
Author's personal copy

393
1 3
Legal andethical implications ofapplications based on…
In order to allow for a legal analysis of a technical solution (as presented in this
paper) essentially three steps have to be carried out:
1. System operation specification Identification and specification of all aspects of the
proposed solution that could affect or could be affected by legal issues. On the one
hand, such a specification should include the high-level application domain(s) of
the system as well as all affected entities. And on the other hand, it should specify
the principal operations (that may affect legal regulations) but hiding away techni-
cal details. The aim is to allow legal experts to analyse compliance with current
regulations.
2. Compliance analysis Legal experts should analyse whether the different aspects
identified in the previous step match current regulations. This step is often not
straightforward and may require a more detailed analysis of different interpreta-
tions of existing norms. If a violation of existing norms is detected, the violated
norms or legal principles should be identified and associated to the corresponding
aspect of the system.
3. Modification proposals Based on the analysis and discussion in step 2, possible
solutions should be identified that would make the proposed system compliant.
Such solutions may be proposals of modifications of the existing regulations,
modifications that have to be introduced in the proposed system, or hybrid solu-
tions. Here, the effects or consequences of modifying each part (regulation or
system) have to be pondered so as to decide which one should be adapted. For
example, in the case of a detected incompliance with fundamental rights adapting
the regulation may be unfeasible, thus the system modification would be the only
option to make it compliant.
Even though formal languages with rigorous semantics and proof theory could be
used to specify the operation principles of the proposed system, for the aforemen-
tioned purpose in step 1 this is not really necessary. We rather require that the speci-
fication should be “understandable” by legal experts in order to allow for the analy-
sis and the discussions in steps 2 and 3, so a structured natural language description
seems more appropriate. Upon this background, in the sequel we summarise key
aspects of the operation of autonomous intersections and networks presented in
Sects.2.1 and 2.2 as follows:
• Domain: Traffic.
• Subdomain: regulation of paths of vehicles through intersections.
• Affected entities: vehicles, intersection manager.
• Definitions:
• Intersection Manager: virtual entity that regulates the paths of vehicles
through a specific intersection. Different intersections are regulated by differ-
ent intersection managers.
• Vehicles: vehicles that want to pass through an intersection.
• Space–time slot: a space slot that is part of a route through an intersection
from one incoming road to an outgoing road.
Author's personal copy

394
J.-A.Santos et al.
1 3
• Operational facts:
• Fact 1: A vehicle can only pass through an intersection using the space–time
slots assigned to it previously by the intersection manager and using the
assigned space slots at the specified time.
• Fact 2: The intersection manager assigns at most one vehicle to each space
slot at each particular time.
• Fact 3: In order to claim a path through an intersection, any vehicle has to
request space–time slots from the corresponding intersection manager.
• Fact 4: If more than one vehicle requests the same space–time slots, the
assignment is based on an auction between the vehicles. The type of auction
employed is one-shot, sealed bid. The winner of an auction pays the bid to the
intersection manager and the intersection manager assigns the corresponding
space–time slots to the winner.
• Fact 5: The intersection manager makes benefit.
• Fact 6: Each intersection has a reserve price which is published by the inter-
section manager. In order to participate in an auction, each vehicle’s bid has
to be higher than the current reserve price of the intersection.
• Fact 7: The intersection manager aims at improving the distribution of the
traffic by specifying the reserve price. The intersection manager establishes
this price locally, incrementing it when the demand grows. There is no maxi-
mum price.
• Fact 8: Each intersection manager takes its decisions locally, without explicit
coordination with other managers in the neighbourhood. Thus, in scenarios
with high general traffic density in a particular area and limited number of
alternative paths, reserve prices at all intersections in that area may grow.
• Assumptions:
• Assumption 1: All vehicles that want to cross an intersection need to be capa-
ble to communicate with the intersection manager and have to have the soft-
ware system installed.
• Affected general issues:
• Responsibility (in case of failure, or illicit behaviour).
• Data protection.
3 Legal issues andconcerns
The proposal described in the preceding pages of this article regarding smart inter-
sections that grant reservations of space and time to drivers based on electronic
online auctions gives rise to numerous legal-administrative problems within the
framework of the current European and Spanish legislation, which stems from the
Vienna Convention on Road Signs and Signals (1968). In the following we will
carry out a legal science fiction exercise to point out some of the many legislative
changes that would need to be introduced into these legal systems to enable the
Author's personal copy

395
1 3
Legal andethical implications ofapplications based on…
deployment of intelligent road infrastructure of such characteristics,2 beginning with
the modification of traffic regulations (von Unger-Sternberg 2018).
3.1 Auction‑based intersections andadministrative law inSpain
The technology is prepared, or almost ready, to deploy auction-based intersections,
but the law not yet. Still, we know that the law always goes one step behind the
reality that it aims to regulate. The United States is the pioneer, in terms of legisla-
tion, regarding the incipient regulation of autonomous vehicles and intelligent road
infrastructures.3 For examples, just consider the state laws of Nevada,4 California,5
or Michigan.6
European Union Law defines Intelligent Transport Systems (ITS) as “advanced
applications which without embodying intelligence as such aim to provide innova-
tive services relating to different modes of transport and traffic management and
enable various users to be better informed and make safer, more coordinated and
‘smarter’ use of transport networks.” The European Union identifies the following
priority areas for the development and use of ITS7:
• Optimal use of road, traffic and travel data,
• Continuity of traffic and freight management ITS services,
• ITS road safety and security applications,
• Linking the vehicle with the transport infrastructure.
In Spain, an intelligent road infrastructure that uses auction-based intersections
would be subject, in the first place, to the road legislation. In order for this road
infrastructure to be installed and used, such as the smart intersection based on auc-
tions, in Madrid (Spain), as stated in this paper (Sect.2), it is necessary to know that
Spain is a politically decentralised country where political power resides in three
2 The Directorate-General of Traffic of Spain has signed a cooperation agreement with the company
Mobileye to start testing autonomous vehicles on Spanish roads, https ://newsr oom.intel .com/news-relea
ses/mobil eye-spain s-road-safet y-autho rity-dgt-colla borat e-enhan ce-road-safet y/. Vision Zero technol-
ogy of this company is important when developing autonomously managed intersections especially those
where bicycles and pedestrians circulate,… vid. Towards Vision Zero: Intelligent Intersection Infrastruc-
ture to enhance safe operations of (self-driving) cars, https ://its.berke ley.edu/node/13121 .
3 National Conference of States Legislatures, http://www.ncsl.org/resea rch/trans porta tion/auton omous
-vehic les-self-drivi ng-vehic les-enact ed-legis latio n.aspx and Self-Drive Act, Sec. 5, https ://www.congr ess.
gov/bill/115th -congr ess/house -bill/3388/text.
4 Nevada Revised Statutes § 482A, https ://www.leg.state .nv.us/NRS/NRS-482A.html.
5 California Vehicle Code § 38750, https ://legin fo.legis latur e.ca.gov/faces /codes _displ aySec tion.xhtml
?lawCo de=VEH&secti onNum =38750 .
6 Michigan Vehicle Code § 257.2.b.; http://www.legis latur e.mi.gov/(S(z4ly2 5pwxo hcwib ffe5s zntb))/
mileg .aspx?page=GetOb ject&objec tname =mcl-257-2b.
7 Directive 2010/40/EU of the European Parliament and of the Council, of 7 July 2010, on the frame-
work for the deployment of Intelligent Transport Systems in the field of road transport and for interfaces
with other modes of transport. In Spain, Royal Decree 662/2012, of April 13, on the framework for the
deployment of Intelligent Transport Systems in the field of road transport and for interfaces with other
modes of transport.
Author's personal copy

396
J.-A.Santos et al.
1 3
territorial levels: The Spanish State, the Autonomous Communities and the local
governments. In Spain, these three territorial spheres have competence on roads.
Consequently, this autonomous intersection can be used on a highway whose owner
is the State, the Autonomous Community of Madrid or the municipality of Madrid.
In the proposal of a single intersection (Sect.2.1) that collects this paper it is neces-
sary to determine to which public administration the road belongs because each ter-
ritorial administration has elaborated its own regulations on roads, hence the impor-
tance of knowing the existing territorial division in Spain.
The State Highway Law promotes research and development of intelligent infra-
structures of this type. In this sense, article 1 of the Spanish Law 37/2015, of Sep-
tember 29, of Roads has the following purposes:
“(b) Offer the necessary infrastructure for the transport of people or goods.
(d) To obtain an offer of road infrastructure and services associated to them of
quality, safe and efficient, with an adequate allocation of resources.
(e) Promote research, development and technological innovation, as well as its
dissemination.
(g) Encourage the development of advanced services to mobility and road
transport.”
In any case, intelligent road infrastructures would need important incentives and
investments, for example, the current Smart Cities and Islands Plan of Spain should
take these investments into account and prioritise them.8
However, also in this area, not only the competences on roads matter, but other
State competences are important as well, such as:
• “public works of general interest or whose implementation affects more than one
Autonomous Community” (Judgement of the Spanish Constitutional Court No.
65/1998);
• the general communications regime;
• the road system that passes beyond the country borders;
In matters of “traffic and circulation of motor vehicles” (article 149.1.21 of the
Spanish Constitution), “are not encompassed only by the conditions related to traffic
8 On Spanish Smart Cities Plan, 2015, vid. http://www.minet ad.gob.es/turis mo/es-ES/Noved ades/
Docum ents/Plan_Nacio nal_de_Ciuda des_Intel igent es.pdf; Estrategia española de i + d + i en inteligencia
artificial, 2019, vid. http://www.cienc ia.gob.es/stfls /MICIN N/Cienc ia/Fiche ros/Estra tegia _Intel igenc ia_
Artifi cial _IDI.pdf. On AENOR Norms in relation to smart cities, vid. http://www.agend adigi tal.gob.es/
plane s-actua cione s/Bibli oteca ciuda desin telig entes /Mater ial%20com pleme ntari o/norma s_ciuda des_intel
igent es.pdf. Also, there are projects financed by the European Union in relation to intelligent road infra-
structure, vid. the Inframix R & D initiative (Road Infrastructure ready for mixed vehicle traffic flows)
https ://cordi s.europ a.eu/proje ct/rcn/21013 1_en.html and Digital Europe programme for the period 2021–
2027, https ://eur-lex.europ a.eu/legal -conte nt/EN/TXT/HTML/?uri=CELEX :52018 PC043 4&from=EN.
For its part, the New York State Law encourages the establishment of communication technology that
allows wireless-enabled infrastructure to share information electronically with vehicles and thereby facil-
itate the progress of the implementation of autonomous vehicles, vid. An Act to emend the transportation
law, in relation to establishing a pilot program for vehicle-to-infrastructure technology, of the State of
New York, January 30, 2018.
Author's personal copy

397
1 3
Legal andethical implications ofapplications based on…
(for example, traffic signals, speed limitations, etc.), but the conditions to be met
by vehicles that circulate by roads”, because the scope of this subsection of article
149.1.21a reaches any guarantees of safety in circulation (Judgments of the Span-
ish Constitutional Court No. 59/1985 and, subsequently, No. 181/1992). Therefore,
when projecting this precept on the road matter, the State is also competent to estab-
lish all those rules of a technical nature that directly affect safety in traffic. This
solution is supported, as indicated in the aforementioned Judgments, “in the fact that
the guarantees of safety in circulation, according to the will of the Constitution…,
must be uniform throughout the national territory” and, consequently, the Spanish
State must comply with international treaties on these matters, especially interna-
tional conventions on road traffic. A vehicle can only pass through an intersection
using the space–time slots assigned to it previously by the intersection manager and
using the assigned space slots at the specified time. Traffic regulations, Spanish and
international, should be modified to contemplate this possibility (facts 1 and 2).
The road legislation, both state and autonomous communities, considers that
roads are ways of public domain and of public use, built and marked mainly for the
circulation of motor vehicles (article 2.2 of the State Road Law and article 3 of the
Law of Madrid of roads). Therefore, the proposal made in this paper (Sect.2.3 and
fact 1) should bear in mind that roads and their intersections or crossroads are public
goods properties, with all the legal consequences and limitations that this implies
for an eventual smart intersection based on auctions, where drivers need to pay to
reserve their time and space for transiting safely through it. Although, in principle,
the proposed intersection would not contradict the consideration as public domain
property of the roads, except for some points that will be discussed in this paper
(fact 1).
For the legal-administrative regulation of these auction-based intersections we
must rely on: the regulation on public goods properties, on the possible rever-
sal of the public domain,9 the possible financing of their construction (public,
private, types of contracts for their tendering,…), public management (direct or
through a company or public entity); private management (concessionaire, pay-
ment of a fee,.…).10 Also, a public law entity could be created and it would be
subject to private law, a State Agency or a special company for the construction
and exploitation of this intelligent infrastructure and, in particular, the autono-
mously managed intersections, which guarantees cooperation between the differ-
ent public administrations, a fact that does not guarantee the proposal included
9 Public domain goods cannot be privately owned; they are outside the private legal trade. They are pub-
lic property. The decision to declare a category of goods as public goods properties belongs to the leg-
islator through the Law (article 132 of the Spanish Constitution: 1. The legal system governing public
domain and community property shall be regulated by law, on the principle that they shall be inalienable
and imprescriptible and not subject to attachment or encumbrance. 2. The property of the State public
domain shall be established by law and shall, in any case, include coastal area, beaches, territorial waters
and natural resources of the economic zone and the continental shelf). Some examples of public goods
properties, in Spain, are roads, surface and underground waters, ports, coasts, … (coasts because the
Spanish Constitution says it). The reversal would suppose that the good is outside the public domain.
The reversal of roads of the public domain would need a Law of reversal. It is a decision of the legislator.
10 Automated vehicles. Do we know which road to take? Infrastructure Partnerships Australia, 2017.
Author's personal copy

398
J.-A.Santos et al.
1 3
in Sect.2.2 of this paper, multiple intersections, Fact 8. Finally, we could con-
sider the complete privatization of the road network, although the legal doctrine
is unanimous in considering that the common use, whose guarantee corresponds
to the public authorities, would be very seriously compromised with the delivery
of the road network to the private sector (Bobes Sánchez 2007, pp 127–128).
The goods that the legislator configures as a public domain are susceptible to
three types of uses: 1. The common or general use that any citizen can make, pro-
vided that others are not excluded, it is free and does not require any formal legal
base. 2. The use especially intense, dangerous, profitable, or preferably in cases
of scarcity that requires an administrative authorization or a concession contract,
if its use exceeds 4years and, 3. the private use that implies occupation of a por-
tion of the public domain, excludes others and requires the granting of an admin-
istrative concession contract (articles 84, 85 and 86 of the Spanish Law 33/2003,
of November 3, on the Patrimony of Public Administrations).
In this regard, it has been said that “the common use of the public domain
corresponds to the exercise of public freedom. Thus the freedom to circulate on
public roads is a manifestation of freedom of movement” (Bobes Sánchez 2007,
p 125). In other words, roads are part of the public domain (article 5 of the Span-
ish Law 33/2003, of November 3, the Patrimony of Public Administrations) and
the circulation of vehicles is the common use of roads (Santamaría-Pastor 1985,
p 390) which is the manifestation of freedom of movement in a motorised world
“and it corresponds equally and indiscriminately to all citizens, so that the use by
some does not impede the other interested ones,” as says article 85 of the State
Spanish Law on the Patrimony of Public Administrations. It would also be pos-
sible, as we said earlier, the special use and even the exclusive use of roads, as
public domain goods (at least in some of their segments that include alternative
routes).
In relation to the regulation of roads (including intersections), the principle of
equality, in the use of the public domain, in our case the auction-based intersec-
tion, should be included its future regulation (Fact 1). However, both case-law and
legal doctrine in relation to this principle of equality in the use of the public domain
of roads have placed special emphasis on distinguishing identical factual assump-
tions that can support a legitimate difference in the use of the corresponding public
domain as, for example, we can distinguish and, therefore, limit their use to neigh-
bours and residents of certain roads of other users; also, we can differentiate depend-
ing on the tonnage of vehicles to prohibit the use of roads in certain hours and days.
Other reasons of public interest are: the reduction of pollution or the regulation of
traffic density to facilitate circulation, also allow introducing variables that can une-
quivocally suggest some type of discrimination, whose differentiation has a reason-
ableandobjective justification (Bobes Sánchez 2007, p 126). These issues related to
the legal classification of roads and their intersections, as well as smart intersections
such as the one included in this paper, as public domain goods, must guarantee the
freedom of movement and the principle of equality. This does not preclude a special
use that distinguishes, for example, among residents, among vehicles according to
their tonnage, restricts or even prohibits the circulation of certain vehicles due to
their degree of pollution or which can only circulate through roads fully autonomous
Author's personal copy

399
1 3
Legal andethical implications ofapplications based on…
vehicles to facilitate the circulation or avoid accidents. However, all these forecasts
are not included in the algorithm of Sect.2.3 of this paper.
Does this intersection guarantee freedom of movement and equality? Does it
allow a vehicle to cross the intersection without paying because there is no alter-
native route? In principle, the algorithm described in Sect. 2 does not allow any
vehicle to cross the intersection for free. This road infrastructure of autonomous
intersection based on auctions, proposed in Sect.2, would violate the freedom of
movement because the proposed algorithm does not guarantee that every vehicle
without a driver can cross the intersection for free (fact 6). Therefore, the algorithm
should be modified so that the user is informed of the waiting time to be able to pass
for free. In this way, the user will assess their individual welfare, as explained in
Sect.2: “the more money a driver is willing to spend to cross the intersection, the
faster the transit will be through it”. If the user is not willing to pay, the waiting time
will be longer but the pass through the intersection will be not prohibited. Another
option would be the construction or existence of an alternative route, for free, to the
autonomous intersection.
Among these restrictions on the common use of the road public domain the regu-
lations on traffic and road safety are included. This traffic regulation protects the
public interest that involves the development of traffic in conditions of safety and flu-
ency. This autonomous infrastructure, as indicated in this paper, has as main objec-
tives to guarantee road safety and traffic flow. The proposed algorithm prioritises
the traffic flow through the payment of a price determined by the auction (fact 7).
The Spanish legislation on traffic has its main rule in the Royal Legislative Decree
6/2015, of October 30, which approves the revised text of the Law on Traffic, Cir-
culation of Motor Vehicles and Road Safety. The current traffic regulations and road
safety do not contemplate an infrastructure like the proposal of this paper and do not
allow the use of roads by fully autonomous vehicles (assumption 1) (On Dutch traf-
fic law, Prakken 2017).11 Promote the development and use of the autonomous vehi-
cle by developing specific legislation and classifying the possible legal gap posed by
the introduction of the autonomous vehicle in circulation”, (Official Gazette of the
Spanish Parliament, No. 204, September 8, 2017, pp 32–33).
Finally, unlike other proposals (for example, Dresner and Stone 2008), the smart
intersections proposed in this paper require payments for space/time reservations at
the intersection, which would be adjusted through an auction mechanism. The type
of auction proposed in this paper fits with the electronic auction, described in article
143 of the Spanish Law 9/2017, of November 8, on Contracts of the Public Sector
(fact 4).
The proposal of autonomous intersection with payment for its use or space/time
reservation implies obtaining benefits (fact 5). Although we have already indicated
that the construction and management of this type of infrastructure can be carried
out in different ways (public management/private management) we now focus on the
issue of road tolls. The assumption would be as follows: the public Administration
11 The non-law Proposition of the Spanish Popular Party states: “The Congress of Deputies urges the
Government to establish an adequate legal framework that allows:…
Author's personal copy

400
J.-A.Santos et al.
1 3
in charge of a highway awards the construction and operation of auction-based inter-
sections to a private company that would be the concessionaire. The concessionaire,
in exchange for building the infrastructure, acquires the exploitation of the intelli-
gent infrastructure as compensation and imposes the payment of a road toll;12 that
is, the payment of the users for the use of this infrastructure. Although the public
administration is competent to determine the maximum price, the algorithm pro-
posed in Sect.2 must have a maximum limit because currently it does not contem-
plate it (fact 7). The payment could be justified by making sure that the common
use of roads was not compromised, the fundamental freedom of movement, in this
case that the only possibility of the exercise of freedom of movement was through
toll roads. The algorithm must contemplate that it can be crossed free of charge and
that there is a maximum price to cross the intersection (facts 6 and 7). The existence
of the road toll could violate the principle of equality but not the “free of charge”
principle, which, as such, does not integrate the concept of public domain (Bobes
Sánchez 2007, p 128).
The tolls for the use of road infrastructures, as the Spanish Supreme Court has
indicated, are payments that users pay to the concessionaire of a service, they are not
fees or fiscal obligations or payments for services rendered to the public within the
meaning of article 31. 3 of the Spanish Constitution (“Personal or property contribu-
tions for public purposes may only be imposed in accordance with the law”) but they
are trade-offs for the service provided by the concessionaire that the concessionaire
makes its own by private law; that is, through the private management of this road
infrastructure benefits are obtained, without prejudice to the intervention that the
granting Administration may have in setting it up in exercise of the corresponding
the tariff regulation, different from taxing powers (Judgment of the Court Spanish
Supreme Court of April 30, 2001, among others). The public Administration (state,
autonomous or local) must establish a maximum price to pass through the inter-
section and, it can demand that the crossing be for free, in certain conditions like
the nonexistence of alternative route (facts 6 and 7). It is not contrary to article 14
of the Spanish Constitution (principle of equality) [Spaniards are equal before the
law and may not in any way be discriminated against on account of birth, race, sex,
religion, opinion or any other personal or social condition or circumstance.], accord-
ing to the Spanish Supreme Court, the difference in the payment of tolls at the time
12 Directive 2004/52/EU of the European Parliament and of the Council, of 29 April 2004, on the inter-
operability of electronic road toll systems in the Community. This Directive lays down the conditions
necessary to ensure the interoperability of electronic road toll systems in the Community. It applies to the
electronic collection of all types of road fees, on the entire Community road network, urban and inter-
urban, motorways, major and minor roads, and various structures such as tunnels, bridges and ferries.
Systems of electronic toll collection which are put in place in the Member States should meet the follow-
ing fundamental criteria: the system should be amenable to ready incorporation of future technological
and systems improvements and developments without costly redundancy of older models and methods,
the costs of its adoption by commercial and private road users should be insignificant compared with
the benefits to those road users as well as to society as a whole, and its implementation in any Member
State should be non-discriminatory in all respects between domestic road users and road users from other
Member States. In Spain, Royal Decree 94/2006, of February 3, on the interoperability of electronic road
toll systems in State roads.
Author's personal copy

401
1 3
Legal andethical implications ofapplications based on…
that some must satisfy it while other users are not required to do so to the same
extent (Judgement of the Spanish Supreme Court of April 29, 2002, for example).
Following this case-law of the Spanish Supreme Court,13 the design of the auction
algorithms for crossing the smart intersection should include differentials in the pay-
ment of the road toll, which may have a surcharge, or a bonus, or even the total
exemption depending on the type of vehicle. In this regard, Vasirani and Ossowski
(2011, p 214) argue that the computational market could be enriched by allowing
“intersection managers to offer a range of different pricing instruments (e.g., off-
peak pricing, daily subscriptions, and fares based on usage of the intersection)”.
However, the proposal contained in Sect.2 does not do it. For example, total exemp-
tion vehicles might include traffic police, government police and other law enforce-
ment and judicial authorities, as well as ambulances and fire services when they had
to carry out some mission on the highway.14 But the algorithm described in Sect.2
does not differentiate between types of vehicles, whether they are emergency or not.
The algorithm must allow the crossing of these emergency vehicles first, before any
other vehicle, and free of charge. Following the example of the Massachusetts, Bill
S. no. 1945, 2017 (which allows the use of roads of that State of the United States by
autonomous vehicles and a fee is established for their use).15 Also, the existence of
discounts in the payment of these road tolls could be considered: depending on the
number of passengers carried by the autonomous vehicle, encouraging public trans-
port or collaborative transport (door to door); depending on transit geographic areas
where there are few public transport options available; based on income or personal
income, establishing a higher rate depending on the tonnage of the heavy vehicle,
etc. The proposed algorithm would increase the price in case of traffic congestion
to infinity; that is to say, as we have indicated, it does not contemplate a maximum
price that the Spanish public administrations should set. Finally, the proposed algo-
rithm does not include any of these promotion measures for vehicles with high occu-
pancy or less polluting; based on family income; etc.
On the other hand, in the proposal of multiple intersections in a network of urban
roads (Sect.2.2 of this paper), as in the case of Madrid, for it to work it is essen-
tial that the three public administrations (State, Autonomous Community of Madrid
or the municipality of Madrid) must coordinate, collaborate and cooperate (article
103 of the Spanish Constitution, article 3 of Law 40/2015, of October 1, the legal
regime of the public sector). A single intersection manager of the Madrid road net-
work should be created with the aim of improving traffic flow. However, in the pro-
posal included in Sect.2.3 (fact 8) each intersection depends on a coordinator who
makes decisions locally; that is, it does not coordinate with the other intersection
managers. This situation goes against the principles of coordination, cooperation
15 Its title is: “An Act to promote the safe integration of autonomous vehicles into the transportation sys-
tem of the Commonwealth”, 2017, Massachusetts (https ://maleg islat ure.gov/Bills /190/S1945 ).
13 Judgment of the Spanish Supreme Court of February 15, 1996; Judgment of the Spanish Supreme
Court of February 21, 1997, …
14 Spanish Decree 215/1973, of 25 January, by which approves the content of the administrative specifi-
cation sheets for the construction, conservation and operation of highways in concession contract.
Author's personal copy

402
J.-A.Santos et al.
1 3
and collaboration between public administrations and violates the principle of effi-
ciency in the allocation and use of public resources recognised in article 7 of the
Organic Law 2/2012, of April 27, on Budgetary Stability and Financial Sustainabil-
ity. Therefore, in a next work we will look for the formula of coordinating intersec-
tion networks to guarantee these principles and above all the traffic flow.
Therefore, the use of auction-based intersections does not violate freedom of
movement and to circulate, principles of equality and non-discrimination. It even
contemplates the possibility of crossing these intersections for free. EU law has
shown that the establishment of tolls throughout the road infrastructure network
does not violate this law when it is based on the “user pays” and “polluter pays”
principles and the national legislation is not discriminatory (e.g., Directive 2011/76/
EU of the European Parliament and of the Council, of 27 September 2011, amending
Directive 1999/62/CE on the charging of heavy goods vehicles for the use of certain
infrastructures and Judgment of the Court of Justice of the European Union [Grand
Chamber] June 18, 2019 Case C 591/17, Republic of Austria v. Federal Republic of
Germany). The EU directives do not guarantee to EU citizens that, when exercising
their right to free movement, it must be carried out free of charge. The economic
policy of each EU member country must be able to determine whether such road
infrastructure is financed by taxpayers or users. In short, it is an economic policy
decision that corresponds to the Member States, therefore such road infrastructure
can be financed by taxes or tolls (Opinion of Advocate General Wahl, delivered on
6 February 2019, Case C 591/17 Republic of Austria v. Federal Republic of Ger-
many). In this way the tolls by auction add more price flexibility is adjusting supply
and demand, establishing maximum and minimum prices.
3.2 Human rights andlegal principles related toauction‑based intersections
The invention of the traffic light and its installation in the streets of the cities caused
the modification of the laws, especially, of the traffic regulations. Until the regula-
tion was not reformed, traffic lights could not regulate the circulation of vehicles.
Intelligent road infrastructures, such as the intersection proposed in this paper for the
circulation of fully autonomous vehicles, will imply a clear legal revolution of many
regulations so that the installation and building of this auction-based intersection be
allowed on roads and streets. The construction of these intelligent infrastructures
will eliminate traffic lights and other traffic signals, and as numerous works have
shown, these intelligent intersections reduce traffic congestion and prevent accidents
(Dresner and Stone 2008, pp 627–628; Beiker 2012; Shubbak 2013; Surden and
Williams 2016; Chen etal. 2017; Kockelman etal. 2018; among others). It would be
necessary to unify the legal and technical solutions on the autonomous intersections
from the international law as it happens with the traffic signals and the traffic lights
that would come to replace. In the previous section we have outlined some regula-
tory changes so that these infrastructures can work, but also this type of infrastruc-
tures based on artificial intelligence in their source code must bear in mind a series
of legal principles and values, as well as fundamental rights. For example, the Euro-
pean Parliament resolution of 16 February 2017, which includes recommendations
Author's personal copy

403
1 3
Legal andethical implications ofapplications based on…
to the Commission on civil law rules on robotics, suggests that research activities on
robotics must respect fundamental rights. In this sense, the Spanish Constitution in
article 18.4 provides: “The law shall limit the use of the computer science in order
to guarantee the honour and personal and family privacy of citizens and the full
exercise of their rights.” In other words, the use of computer science cannot curtail
the exercise of citizens’ rights; constitutional limitation that must be very present in
the regulation of artificial intelligence, also in these intelligent infrastructures. Bear-
ing in mind this constitutional limitation, the auction-based intersections proposed
in this paper must respect the right of free movement of citizens, as has been shown
in Sect.3.1.
Liberty and Equality are closely linked. All citizens have the right to use pub-
lic domain goods to guarantee freedom of movement. Provided there is no alterna-
tive route, the proposed algorithm must guarantee that the autonomous vehicle can
cross the intersection free of charge. Fact that is not done currently (fact 6). And,
in this regard, the proposed algorithm in this paper is not discriminatory and does
not violate the article 14 of Spanish Constitution nor the anti-discrimination rules
(Perez 2016; Pearson 2016; Levin 2016; Council of Europe 2018). The algorithm
that these auction-based intersections use is neutral and refrain from the elaboration
of user profiles;16 that is, this algorithm does not correspond to what has been called
“opaque algorithms” (Navas Navarro 2017). In other words, we keep in mind in the
design of these algorithms what is called “embedded values” (Surden 2017). These
values are widely used in the technological design choices made by engineers, for
this reason the algorithm may end up having the effect of promoting or prioritiz-
ing certain social values over others or giving advantages or disadvantages to some
social groups over others, especially in the process of auctioning the right and tim-
ing of passing through such autonomously managed intersections (for example, pri-
oritizing the circulation of vehicles from some neighbourhood against others). Thus,
for example, in the next regulation of autonomous vehicles of Massachusetts (An
Act to promote the safe integration of autonomous vehicles into the transportation
system of the Commonwealth, Bill S. no. 1945, 2017) it is said in the regulation of
these vehicles that it will seek to protect the most affected and disadvantaged com-
munities of the State and will ensure equal protection and the equitable distribution
of the benefits and costs associated with the introduction of autonomous vehicles. In
the proposal of this paper, the algorithm, in its initial formulation, is neutral because
it does not incorporate promotional or discriminatory measures. We do not know
what will happen when the algorithm starts to learn. (Caliskan etal. 2017; Packin
and Lev-Aretz 2018). It is necessary to ensure that this algorithm is transparent, for
example, through open code and guarantees that this algorithm can be audited.
16 Regulation (EU) 2016/679 of the European Parliament and of the Council; of 27 April 2016; on the
protection of natural persons with regard to the processing of personal data and on the free movement
of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) defines ‘profiling’
as “means any form of automated processing of personal data consisting of the use of personal data to
evaluate certain personal aspects relating to a natural person, in particular to analyse or predict aspects
concerning that natural person’s performance at work, economic situation, health, personal preferences,
interests, reliability, behaviour, location or movements.”.
Author's personal copy

404
J.-A.Santos et al.
1 3
When we translate a law (legal texts) to a formal rule (computer system) we can
mask a series of subjective and debatable decisions about the meaning and scope
of the formal rule (Pagallo and Durante 2016), as we can choose between different
legal options (Sect.3.1) in relation to issues such as public or private administration
of the auction-based intersection manager, the free access during periods of time, the
reversal of public domain goods or the crossing priority of some vehicles over oth-
ers, etc. That is, the algorithms, being programmed by people, might have a strong
ideological component that will have to be monitored and audited. The algorithm
proposed in Sect.2 is subject to such monitoring and auditing. Finally, to avoid dis-
crimination and algorithmic manipulation, it is necessary to audit the algorithms,
when it includes the concepts of “algorithmic responsibility” and the “right to expla-
nation (Kaminski 2019); that is to say that there must be transparency and acces-
sibility of the code (Goodman and Flaxman 2017). The algorithm proposed in this
paper is transparent and accessible. In summary, references to autonomous decision-
making by artificial intelligence systems, including this autonomous infrastructure,
cannot exempt the creators, owners and managers of these systems from responsibil-
ity for human rights violations, of the principle of non-discrimination and of other
regulations, which could be committed with the use of these intelligent intersection
systems (Etzioni and Etzioni 2016; Bench-Capon and Modgil 2017). Consequently,
this proposed algorithm must keep all information about the allocation of crossings,
bids, etc. during a certain period of time (approximately 5years) for facilitates the
audit of the same and to verify that this algorithm responds to legality, it is accessi-
ble for a third party (human); it’s transparent; recognise the right to judicial recourse
to automated decisions (e-auctions) that do not violate human rights and do not dis-
criminate for any reason.
As it has been revealed, the proposed algorithm must be designed and imple-
mented to protect the personal data it uses, as required by Community law (Commu-
nication from the European Commission to the European Parliament, the Council,
the European economic and social Committee and the Committee of the Regions,
2019). This proposed infrastructure would use databases, e.g. to store the already
granted space/time reservations at the intersection managers, to keep track of the
current reserve prices at the different intersections, etc. For this reason, its legal reg-
ulation is necessary in order to protect the privacy of data (article 10 of the Directive
2010/40/EU of the European Parliament and of the Council, of 7 July 2010, on the
framework for the deployment of Intelligent Transport Systems in the field of road
transport and for interfaces with other modes of transport). The Spanish Constitu-
tional Court has indicated, in the Judgment no. 254/1993, that all citizens have the
“right to control the use of the data inserted in a computer program”. The constitu-
tional case-law, and especially, the Judgment no. 292/2000, of November 30, has
come to specify the scope, content and limits of this fundamental right to the protec-
tion of personal data. The Regulation (European Union) 2016/679 of the European
Parliament and of the Council; of 27 April 2016; on the protection of natural persons
with regard to the processing of personal data and on the free movement of such
data, and repealing Directive 95/46/EC (General Data Protection Regulation) is the
most important legal text in this matter. This European Regulation, which entered
into force on May 25, 2018, indicates that necessary and proportionate measures
Author's personal copy

405
1 3
Legal andethical implications ofapplications based on…
must be contemplated to ensure the security of the network and of the information;
that is, the ability of a network or information system to resist, at a certain level of
trust, accidental events or illicit or malicious actions that compromise the availabil-
ity, authenticity, integrity and confidentiality of personal data stored or transmitted,
and the security of the services offered (Bartolini etal. 2018). This could include,
for example, preventing unauthorised access to electronic communications networks
and the malicious distribution of codes, and curbing “denial of service” attacks
and damage to computer systems and electronic communications. The administra-
tors of these auction-based intersection managers must guarantee the data security
and protection, as well as having systems that avoid the hacking of the intelligent
infrastructure (Taeihagh and Lim 2019), as well as avoiding possible damages to the
vehicles and to the people who use this infrastructure (auction-based intersection)
with the objective that its use is completely safe.17 It would be obligatory that intel-
ligent infrastructures of this type have insurance, which entails that the insurance
law regulates the clauses of this type of insurance contracts that cover the hypo-
thetical economic liability derived from any failure of this intelligent intersection
(Hernáez Esteban 2018). There could be several assumptions of responsibility for
the operation of the autonomous intersection (Lohmann 2016; Marchant and Lin-
dor 2012): (a) if there is an accident due to failure of the coordinator or manager
of intersections, for example, by giving simultaneous permission and two vehicles
occupy the same space–time slot. The algorithm described in Sect.2 is designed so
that this does not happen. The responsibility would correspond, in principle, to the
intersection manager. (b) The assumption that an accident occurs due to failure to
comply with the conditions of crossing (space or time), the responsibility could be
attributed to the constructor of the autonomous vehicle, (c) if a breakdown occurs in
the intersection, the intersection manager must avoid the accident and the responsi-
bility could fall on the designer of the autonomous vehicle in accordance with the
responsibility for damages caused by defective products. In any case, the algorithm
proposed in Sect.2 must be implemented with the requirements of security, encryp-
tion and privacy required by the European Union regulations on data protection and
cybersecurity.
This proposal of autonomously managed intersections based on auctions for
reserving time and space must bear in mind that the outcome of these auctions
would correspond to what has been called “smart contracts” (e-auctions) (Surden
2012). Smart contracts have four characteristics: (1) have a computer program, (2)
be based on the blockchain technology (Millard 2018; De Filippi and Wright 2018;
Górriz López 2017), (3) be self-executing and, (4) be autonomous in the sense that
human intervention is not necessary. The proposed algorithm can perfectly imple-
ment these four notes.
17 Directive (EU) 2016/1148 of the European parliament and of the Council of 6 July 2016 concern-
ing measures for a high common level of security of network and information systems across the Union
and Commission Recommendation (EU) 2019/534 of 26 March 2019 Cybersecurity of 5G networks,
C/2019/2335, DO L 88 de 29.3.2019, p. 42/47.
Author's personal copy

406
J.-A.Santos et al.
1 3
Differently from early proposals for autonomous intersection management in
which the first vehicle that arrived was the first to cross the intersection (first in time,
greater in right or First Come, First Served [Dresner and Stone 2008; Vasirani and
Ossowski 2011, p 197; Parker and Nitschke 2017]), in this proposal the intersec-
tion manager set prices for space/time reservations within a computational market
that must be subject to legal principles of contractual regulations such as freedom of
access to the reservation purchase system, publicity and transparency of electronic
contracting procedures (public e-auctions), and non-discrimination and equal treat-
ment between the buyers, among others. The algorithm proposed in Sect.2 must
take into account all these legal requirements, avoiding discriminatory or dynamic
prices according to user profiles (Zuiderveen Borgesius and Poort 2017).
In the case proposed in this paper, it would be an automated contracting of both
negotiating parties; that is to say, the contracting between intelligent agents (autono-
mous vehicle and auction-based intersection manager). The problems in this case
are placed in relation to the risk distribution criteria in the event that there is an
error in contracting or occurrence of fortuitous case as those described above (von
Unger-Sternberg 2018; Casanovas etal. 2018). The traditional category of vices of
consent will have to be questioned because it is inadequate for a computational mar-
ket whose actors are intelligent agents. This type of contracts between two intelli-
gent agents raises many questions about the willingness of the parties, in accordance
withthe current contractual regulation. Therefore, a reform of contract laws would
be compulsory (Navas Navarro 2017).
4 Challenges forreal‑world‑deployment
In this section we analyse certain important technical aspects that do not comply
with the proposed system from the standpoint of Spanish legislation and in the con-
text of international regulations, proposing solutions from the ethical and legal and/
or technical scopes to guarantee its viability. We are aware that our proposed system
is only a real approximation of as many others as possible in order to pave the way
for the deployment in the real world of such systems in the near future.
As we have argued in the previous section, numerous legal problems need to
be addressed in order to give legal support a smart traffic infrastructure including
auction-based intersections. An exciting subject that exceeds the current regulation,
where many sectors of the legal activity are intermingled: the own institution of the
roads like goods of public domain and the use of the same one; competencies among
public administrations; the public or private management of such infrastructures,
freedoms and rights involved; the public contracts; vehicle passing preference and a
long etcetera.
As we know, the intelligent infrastructure pursues intersection crossing assigning
reserves to each vehicle. This operating system affects vehicles and traffic. Thus,
from the point of view of the law, the regulation to cross the intersection in Spain
must be carried out by the corresponding authority (state, autonomous or local
Administrations), because there are different levels of competence.
Author's personal copy

407
1 3
Legal andethical implications ofapplications based on…
In the sequel, we examine the different operational facts presented in Sect. 2.3
and analyse their compliance with the current legal regulation. If needed, we suggest
how the current law or technical solution should be adapted before such a system
could be deployed in a real setting (Table1 summarises this discussion):
• Fact 1: A vehicle can only pass through an intersection using space–time slots
previously assigned by the intersection manager and using space quotas allocated
at the specific time. From a legal point of view, this implies taking into account
that roads are public domain goods, so implementations of intersection managers
have to comply with traffic legislation. This is currently not the case for the sys-
tem proposed in Sect.2. In Spain, for instance, the Law on Traffic, Circulation
of Motor Vehicles and Road Safety would need to be modified in order to allow
the use of autonomous vehicles in their different modalities. In the realization
of this reform would be intended to avoid violations that affect the freedom of
movement and the principle of equality. It is a fact that the reserve system must
be sustainable and not imply strong imbalances in terms of equality, so the indis-
criminate use of reserves can be limited to a number of times per month or year.
The purpose is to guarantee the access of the greatest number of users to the
reservation system, provided they are autonomous vehicles (driver-less vehicle
or with safety drivers).
• Fact 2: The intersection manager assigns a maximum of one vehicle to each
space in a specific moment of time. This fact does not comply with the legis-
lation, because the Spanish Law on Traffic, Circulation of Motor Vehicles and
Road Safety does not include the possibility that there are managers of intersec-
tions. Again, in order to assure the legal viability of autonomous intersections in
line with the proposal of Sect.2, both norms would need to be modified.
• Fact 3: In order to claim a pass through an intersection, the vehicle must request
space–time slots at a given time from the manager of the corresponding intersec-
tion. This fact does comply with the legislation so no modifications would be
necessary.
• Fact 4: The allocation of space–time slots between vehicles is based on sealed-
bid one shot auction. From the perspective of law, this constitutes an electronic
auction regulated by the Spanish Law 33/2003 of November 3, on the Patrimony
of Public Administrations and the Spanish Law of 9/2017 of November 8, Con-
tracts of the Public Sector. Again, as to this aspect, the mechanism proposed in
Sect.2 would comply with Spanish legislation.
• Fact 5: In the proposed system, the intersection manager obtains benefits. From
a legal point of view, the implementation of the intersection based on auctions,
as described in Sect. 2, can be established in two ways: either through public
management without benefits, or by means of private management with benefits
through a concession. Both alternatives would comply with the current legisla-
tion in Spain.
• Fact 6: To participate in an auction, each intersection has a reserve price that has
been published by the intersection manager. However, the offer of each vehicle
must be higher than the price of the current reservation of the intersection. Here,
the proposal of Sect. 2 does not comply with the legislation, roads are public
Author's personal copy

408
J.-A.Santos et al.
1 3
Table 1 Summary of legal implications and possible solutions to the operational facts of the auction-based intersection system analysed in this paper
Fact Description Legal implications Proposed solution
1 Vehicle can only cross an intersection using the
space–time slots assigned by the intersection
manager
Traffic law Law modification: allow autonomous vehicles
2 The intersection manager assigns at most one vehicle
to each space slot at each particular time
Law on Traffic, Circulation and Road Safety Law modification: allow autonomous intersection
coordinators
3 To cross an intersection, any vehicle has to request
space–time slots from the intersection manager
Complies with current legislation Not needed
4 Space–time slots assignment is based on a one-shot,
sealed bid auction
Laws of Patrimony of Public Administrations and
Contracts of the Public Sector
Not needed
5 The intersection manager obtains benefit Public/private management Not needed
6 Each vehicle’s bid has to be higher than the current
reserve price of the intersection
Free use of public goods Technical: free access to drivers who have been wait-
ing certain time
7 The intersection manager aims at improving traffic
distribution by modifying the reserve price. This is
done locally, without maximum price
Public Administration is competent to set maximum
price
Technical: fixing a maximum reserve price
8 Intersection managers do not explicitly coordinate
with each other. Thus, with high traffic density,
reserve prices at all intersections may grow
Efficiency of the infrastructure and cooperation
between public administrations
Technical: fixing a maximum reserve price
Author's personal copy

409
1 3
Legal andethical implications ofapplications based on…
domain goods and drivers’ need to be given an option to use them free of charge.
Here, we would favour a technical solution to comply with this legal require-
ment. The auction protocol could simply be modified such that drivers who have
been waiting for a certain amount of time would be granted access to the inter-
section for free.
• Fact 7: The intersection managers introduced in Sect.2 set reserve prices primar-
ily to improve the distribution of traffic, rather than with the aim of obtaining
benefits. Nonetheless, from the legal point of view, the price of the reservation
is not limited and depends on the traffic demand. Still, the public Administra-
tion is competent to determine a maximum price and the algorithm introduced
in Sect.2 should have such a limit. Still, this is not contemplated in the proposal
outlined in Sect.2, so it currently does not comply with Spanish legislation. An
additional drawback pertains to the possibility of reserve price increasing con-
tinuously, if many vehicles wanted to use that intersection and there are no, or
only very few, alternative routes. In this case drivers should be deterred from
using the intersection. A higher price would certainty serve for this purpose but
may not be socially and is definitely not legally acceptable. A solution to both
aforementioned problems would consist in fixing a maximum reserve price for
intersection and that, once this maximum price is reached at some intersection, it
could be temporarily closed for drivers and would thus not accept new reserva-
tions for a certain amount of time.
• Fact 8: Each intersection manager makes decisions at the local level without
explicit coordination with other managers. As outlined above, in cases with high
traffic density reserve prices at all intersections in a certain area can grow. This
behaviour does not comply with the legislation as, from the legal standpoint, the
principles of efficiency of the infrastructure and cooperation between the differ-
ent entities involved need to be respected. As above, setting a maximum price
for intersection would be the easiest solution to put limits to such an “arms race”
between different intersection managers. Furthermore, even though the mecha-
nism outlined in Sect.2 does not employ explicit communication between inter-
section managers, their reserve prices are coordinated implicitly through the
environment.
Next, we analyse five general assumptions that affect the proposed system and
have an impact on the applicable law. The first assumption (assumption 1) deter-
mines that all vehicles that wish to cross an intersection must be able to commu-
nicate with the intersection manager and must have the software system installed.
The regulation could, perfectly, include such an assumption, which would imply that
vehicles cannot circulate without having the software system installed. However,
socially this may not be an acceptable solution. From an ethical-legal standpoint
and in order to avoid short-term inequalities, it would be better to establish a transi-
tion period where intersections can be used by both, vehicles that are able to use
the proposed technology and those that are not. Technically, the proposed system
could be adapted in this sense. The solution here consists in establishing two types
of reservation time intervals that are alternated—one for the auction-based assign-
ment and one for “traditional” vehicles. In fact, it would also be possible to change
Author's personal copy

410
J.-A.Santos et al.
1 3
both intervals dynamically and adapt them to the current demand in each moment by
monitoring the vehicles that want to pass through an intersection. Such monitoring
could be done by combining communication with the vehicles with standard means
of determining the number of arriving vehicles (e.g., through sensors). Dresner and
Stone (2008) already analysed different techniques to allow for the coexistence of
traditional and automatic vehicles in reservation-based infrastructures.
We have highlighted other general assumptions such as responsibility (assump-
tion 2) and data protection (assumption 3). With regards to the former, in case of
failure by the intersection manager (e.g., in the case that the manager has granted
simultaneous permission to two vehicles to which a behaviour can be reported as
compliance with some of the conditions of crossing) the fault will be yours when
imposing civil and/or criminal liabilities. The assignment algorithm is designed such
that it is not possible that two vehicles have the right to use the same space–time-
slots, e.g., that two vehicles are authorised to cross the intersection at the same time.
Another problem would be to regulate an alternative modus operandi in case an
intersection manager is not working because of any failure. Such a case could be
regulated in a similar way as it is done for standard traffic lights.
With regards to assumption 3, it is necessary to apply General Data Protection
Regulation (EU) 2016/679 in order to improve cybersecurity and protection of data.
The algorithms used within the computational mechanisms involving auction-based
intersections must be effectively audited. The establishment of an Agency for Robot-
ics and Artificial Intelligence, in Spain, would be desirable to this respect, following
the recommendation of the European Parliament (Walker-Osborn and Chan 2017;
Kroll etal. 2017; Tutt 2017). This Agency should dedicate itself to the supervision
and audit of the algorithms (García-Herrero 2017) to, on the one hand, verify that
the algorithms are a faithful translation of the regulations (including respect for
human rights and the principle of non-discrimination) and, on the other, this algo-
rithm audit must include the need to protect personal data. In addition, a certifica-
tion mechanism must be used to prove compliance with these obligations and, con-
sequently, a new work emerges: the algorithm inspector. In short, the algorithms
must respond to the legality, the algorithms must be accessible so that a third party
(human) can audit them, and the algorithms must be transparent. The right to apply
to the courts must be recognised in the face of automated decisions that violate
human rights, or are discriminatory or, in general, breach the legal system. There-
fore, we suggest that the information of bids, assignments, etc. of each intersection
manager be saved for a certain amount of time, which would need to be established
at a normative level (usually 5years).
Finally, we analyse two assumptions not included in the proposed system, but that
must be taken into account from an ethical and legal point of view. In this context,
the underlying concept of morality plays an important role as limit to the legal and
ethical decisions of auction-based road intersections. According to Floridi and Sand-
ers, an “agent is morally good if its actions all respect” the “threshold”, that they call
morality; “and it is morally evil if some action violates it. That view is particularly
informative when the agent constitutes a software or digital system, and the observa-
bles are numerical” (2004, p 349). Here, we consider two especial cases. The first
case (assumption 4), emergency vehicles (e.g. ambulances, fire trucks, police
Author's personal copy

411
1 3
Legal andethical implications ofapplications based on…
vehicles, etc.) that should have priority in the crossing, especially in emergency situ-
ations. In the second case (assumption 5), discounts/incentives can be applied for
vehicles with special characteristics (e.g. low-emission cars, high-occupancy vehi-
cle, vehicles by disabled drivers with different degrees of disability, etc.). Conse-
quently, the legislator must establish promotion policies as a systemof tax benefits
for users who meet some of those characteristics.
Again, the work of Dresner and Stone (2008) already pointed to different techni-
cal option for implementing such priorities in an implementation of a reservation-
based intersection. In general, the aforementioned two aspects could be solved in a
common way by assigning priority levels to different types of vehicles (e.g., ambu-
lances, police cars, etc. but also low-emission cars or cars owned by disabled). Such
priority levels, for example, could establish discounts or weights on the reservation
cost and the bidding price (in form of multipliers) or could even exempt certain vehi-
cles from the task of bidding (e.g., for ambulances or other emergency vehicles). In
fact, an assignment of mobility resources (like intersections), based on economic
principles, is much more suitable for individualising the usage of such resources in
order to meet certain global utility criteria.
5 Conclusions
The emerging field of AT aims at supporting large-scale open distributed systems,
made up of smart software components that negotiate on behalf of their users. A
key field of application are Smart Cities, and smart transportation in particular.
Still, while the sandbox of AT provides models, methods, and tools to enable such
applications from a technical point of view, several legal and ethical issues need to
be addressed before the potential real-world deployment of such systems in a near
future. In this paper, we used the approach of auction-based smart intersections as a
case study to illustrate this matter. We argued that technically it is possible to sup-
port networks of such intersections, and to palliate the negative effect of providing
preferential access to the infrastructure on social welfare through the dynamic adap-
tation of reserve prices. Then, we have analysed what would be the legal and ethical
implications if such smart intersections were to be put into reality in the future. In
particular, we have analysed what modifications would be needed in the smart traffic
infrastructure itself, and what changes in legal regulations would be required before
a potential real-world deployment could be considered.
The exercise we carried out is related but different to traditional legal require-
ments engineering where the objective is to determine the requirements a new sys-
tem has to fulfil in order to comply with current regulations. Our objective, instead,
was rather to determine the modification that should be introduced, both at the tech-
nical and the legal level, in order to put innovative solutions into practice. We think
that this is a common problem and we have put forward some general steps that
could help to address such problems.
Our future work will unfold among two major lines. Firstly, we will explore how
the additional technical requirements can be integrated into the auction protocol for
smart intersections, and re-evaluate the potential impact on their performance. In
Author's personal copy

412
J.-A.Santos et al.
1 3
parallel, we will further look into the impact and consequences of the legal and ethi-
cal issues that have been identified. Secondly, we aim at performing studies similar
to the one outlined in this article to other mechanisms, enabled by intelligent inter-
actions with smart infrastructures, not only in the field of smart transportation, but
also in domains such as smart grids and smart governance.
Acknowledgements This work has been partially supported by the Spanish Ministry of Economy
and Competitiveness, and the Spanish Ministry of Science, Innovation and Universities, co-funded by
EU FEDER Funds, through Grants TIN2015-65515-C4-X-R (SURF) and RTI2018-095390-B-C33
(InEDGEMobility).
References
Bartolini C, Lenzini G, Santos C (2018) A legal validation of a formal representation of GDPR articles.
In: TERECOM 2018, regulatory and legal compliance
Beiker SA (2012) Legal aspects of autonomous driving. Santa Clara Law Rev 52:1145–1156
Bench-Capon T, Modgil S (2017) Norms and value based reasoning: justifying compliance and violation.
Artif Intell Law. https ://doi.org/10.1007/s1050 6-017-9194-9
Billhardt H, Julián V, Corchado JM, Fernandez A (2015) Human-agent societies: challenges and issues.
Int J Artif Intell 13(1):28–44
Bobes Sánchez MJ (2007) La teoría del domino público y el Derecho de carreteras. Iustel, Madrid
Boella G, Humphreys L, Muthuri R, Rossi P, van der Torre L (2014) A critical analysis of legal require-
ments engineering from the perspective of legal practice. In: 2014 IEEE 7th international workshop
on requirements engineering and law (RELAW), Karlskrona, 2014, pp 14–21
Brynjolfsson E, McAfee A (2016) The second machine age. Work, progress, and prosperity in a time of
brilliant technologies. W. W. Norton & Company, New York
Caliskan A, Bryson JJ, Narayanan A (2017) Semantics derived automatically from language corpora con-
tain human-like biases. Science 356:183–186. https ://doi.org/10.1126/scien ce.aal42 30
Casanovas P (2013) Agreement and relational justice: a perspective from philosophy and sociology of
law. In: Ossowski S etal (eds) Agreement technologies. Springer, Dordrecht, pp 19–41
Casanovas P, Morris N, González-Conejero J, Teodoro E, Adderley R (2018) Minimisation of incidental
findings, and residual risks for security compliance: the SPIRIT project. In: TERECOM 2018, regu-
latory and legal compliance
Centeno R, Billhardt H, Hermoso R, Ossowski S (2009) Organising MAS—a formal model based on
organisational mechanisms. In: Proceedings of the 2009 ACM symposium on applied computing
(SAC), pp 740–746
Chen T, Chen F, Chen C (2017) Review on driverless traffic from management perspective. https ://doi.
org/10.1051/matec conf/20171 24030 02
Council of Europe (2018) Discrimination, artificial intelligence, and algorithmic decision-making, 2018.
https ://rm.coe.int/discr imina tion-artifi cial -intel ligen ce-and-algor it hmi c-decis ion-makin g/16809
25d73 . Accessed 26 Sept 2019
De Filippi P, Wright A (2018) Blockchain and the law. The rule of code. Harvard University Press,
Cambridge
Dresner K, Stone PA (2008) Multiagent approach to autonomous intersection management. J Artif Intell
Res 31:591–656
Etzioni A, Etzioni O (2016) Designing AI systems that obey our laws and values. Commun ACM
59:29–31
European Commission (2019) Communication from the Commission to the European Parliament, the
Council, the European economic and social Committee and the Committee of the Regions (2019),
building trust in human-centric artificial intelligence, COM/2019/168 final
Floridi L, Sanders JW (2004) On the morality of artificial agents. Minds Mach 14:349–379
Fornara N, Lopes Cardoso H, Noriega P, Oliveira E, Tampitsikas C (2013) Modelling agent institutions.
In: Ossowski S etal (eds) Agreement technologies. Springer, Dordrecht, pp 277–308
Author's personal copy

413
1 3
Legal andethical implications ofapplications based on…
García-Herrero J (2017) ¿Cómo se audita un Algoritmo?. http://www.lawan dtren ds.com/notic ias/tic/
como-se-audit a-un-algor itmo.html. Accessed 26 Sept 2019
Goodman B, Flaxman S (2017) European Union regulations on algorithmic decision-making and a “right
to explanation”. Artif Intell Mag 38(3):2017. https ://doi.org/10.1609/aimag .v38i3 .2741
Górriz López C (2017) Tecnología blockchain y contratos inteligentes. In: Navas Navarro S (ed) Inteli-
gencia artificial. Tecnología, Derecho, Tirant lo Blanch, Valencia, pp 151–195
Hernáez Esteban E (2018) Inteligencia artificial y vehículos autónomos: el régimen de la responsabilidad
civil ante los nuevos retos tecnológicos. Rev Aranzadi Derecho Nuevas Tecnol 48:7
Kaminski ME (2019) The right to explanation, explained (June 15, 2018). University of Colorado Law
Legal Studies Research Paper No. 18-24; Berkeley Technology Law Journal, vol. 34, No. 1. http://
doi.org/10.2139/ssrn.31969 85
Kockelman K etal (2018) Bringing smart transport to Texans: ensuring the benefits of a connected and
autonomous transport system in Texas—final report. Center for Transportation Research Bureau of
Engineering Research. The University of Texas, pp 59–89. https ://libra ry.ctr.utexa s.edu/ctr-publi
catio ns/0-6838-3.pdf. Accessed 26 Sept 2019
Kroll JA, Huey J, Barocas S, Felten EW, Reidenberg JR, Robinson DG, Yu H (2017) Accountable algo-
rithms. Univ Pa Law Rev 165:633–705
Levin S (2016) A beauty contest was judged by AI and the Robots didn’t like dark skin. The Guardian
(London, 8 September). https ://www.thegu ardia n.com/techn ology /2016/sep/08/artifi cial -intel ligen
ce-beaut y-conte st-doesn t-like-black -peopl e. Accessed 26 Sept 2019
Lohmann M (2016) Liability issues concerning self-driving vehicles. Eur J Risk Regul 7(2):335–340.
https ://doi.org/10.1017/S1867 299X0 00057 54
Marchant GE, Lindor RA (2012) The coming collision between autonomous vehicles and the liability
system. Santa Clara Law Rev 52:1321
Millard C (2018) Blockchain and law: Incompatible codes? Comput Law Secur Rev 34(4):843–846. https
://doi.org/10.1016/j.clsr.2018.06.006
Namazi E, Li J, Lu C (2019) Intelligent intersection management systems considering autonomous vehi-
cles: a systematic literature review. IEEE Access 7:91946–91965
Navas Navarro S (2017) Derecho e inteligencia artificial desde el diseño. Aproximaciones. In: Navas
Navarro S (ed) Inteligencia artificial. Tecnología, Derecho, Tirant lo Blanch, Valencia, pp 23–72
Omicini A, Ossowski S, Ricci A (2004) Coordination infrastructures in the engineering of multiagent
systems. In: Bergenti F, Gleizes MP, Zambonelli F (eds) Methodologies and software engineer-
ing for agent systems—the agent-oriented software engineering handbook. Springer, Dordrecht, pp
273–296
Ossowski S etal (2013) Agreement technologies. Law, governance and technology series (LGTS) no. 8.
Springer, Dordrecht.
Packin NG, Lev-Aretz Y (2018) Learning algorithms and discrimination. In: Barfield W, Pagallo U (eds)
Research handbook on the law of artificial intelligence. Elgar Publishing, Cheltenham, pp 88–113.
https ://doi.org/10.4337/97817 86439 055
Pagallo U, Durante M (2016) The pros and cons of legal automation and its governance. Eur J Risk
Regul. https ://doi.org/10.1017/S1867 299X0 00057 42
Parker A, Nitschke G (2017) How to best automate intersection management. In: IEEE congress on evo-
lutionary computation (CEC). https ://ieeex plore .ieee.org/docum ent/79694 48. Accessed 26 Sept
2019
Pearson J (2016) It’s too late—we’ve already taught AI to be racist and sexist, in motherboard. https ://
mothe rboar d.vice.com/en_us/artic le/weve-alrea dy-taugh t-artifi cial -intel ligen ce-to-be-racis t-sexis t.
Accessed 26 Sept 2019
Perez S (2016) Microsoft silences its new A.I. bot tay, after Twitter users teach it racism. Tech Crunch,
24. https ://techc runch .com/2016/03/24/micro soft-silen ces-its-new-a-i-bot-tay-after -twitt er-users
-teach -it-racis m/. Accessed 26 Sept 2019
Prakken H (2017) On the problem of making autonomous vehicles conform to traffic law. Artif Intell Law
25(3):341–363
Rabinia A, Ghanavati A (2018) The FOL-based legal-GRL (FLG) framework: towards an automated
goal modeling approach for regulations. In: Proceedings of the IEEE 8th international model-driven
requirements engineering workshop (MoDRE), Banff, AB, pp 58–67
Santamaría-Pastor JA (1985) Comentarios al artículo 19. In: Garrido-Falla F (ed) Comentarios a la Con-
stitución. Civitas, Madrid
Author's personal copy

414
J.-A.Santos et al.
1 3
Shubbak M (2013) Self-driving cars: legal, social, and ethical aspects. http://doi.org/10.2139/ssrn.29318
47
Surden H (2012) Computable contracts. UC Davis Law Rev 46(629):629–700
Surden H (2017) Values embedded in legal artificial intelligence. University of Colorado Law Legal
Studies Research Paper No. 17, p 15
Surden H, Williams M-A (2016) Technological opacity, predictability, and self-driving cars. Cardozo
Law Rev 38:121–181
Taeihagh A, Lim HSM (2019) Governing autonomous vehicles: emerging responses for safety, liability,
privacy, cybersecurity, and industry risks. Transp Rev 39(1):103–128. https ://doi.org/10.1080/01441
647.2018.14946 40
Tutt A (2017) An FDA for algorithms. Adm Law Rev 69:83–123
Vasirani M, Ossowski S (2009) Evaluating policies for reservation-based intersection control. In: Pro-
ceedings of the 14th Portuguese conference on artificial intelligence, pp 39–50
Vasirani M, Ossowski S (2011) Learning and coordination for autonomous intersection control. Appl
Artif Intell 25(3):193–216
Vasirani M, Ossowski S (2012) A market-inspired approach for intersection management in urban road
traffic networks. J Artif Intell Res 43:621–659
von Unger-Sternberg A (2018) Autonomous driving: regulatory challenges raised by artificial decision
making and tragic choices. In: Barfield W, Pagallo U (eds) Research handbook on the law of artifi-
cial intelligence. Elgar Publishing, Cheltenham, pp 88–113. https ://doi.org/10.4337/97817 86439 055
Walker-Osborn C, Chan C (2017) Artificial intelligence and the law. ITNOW 59(1):36–37. https ://doi.
org/10.1093/itnow /bwx01 7
Zuiderveen Borgesius F, Poort J (2017) Online price discrimination and EU data privacy law. J Consum
Policy 40:347. https ://doi.org/10.1007/s1060 3-017-9354-z
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations.
Author's personal copy