Figure 1 - uploaded by Milan Jelisavcic
Content may be subject to copyright.
Source publication
Evolutionary robotics using real hardware has been almost exclusively restricted to evolving robot controllers, but the technology for evolvable morphologies is advancing quickly. We discuss a proof-of-concept study to demonstrate real robots that can reproduce. Following a general system plan, we implement a robotic habitat that contains all syste...
Contexts in source publication
Context 1
... versus virtual evaluations. The majority of evolutionary robotics follows scheme A shown in Figure 1, where the complete evolutionary process takes place in simulation. A handful of systems are based on scheme B, where some (usually not all) fitness evaluations are performed on real hardware. ...
Context 2
... alternative is online evolution, a.k.a. embodied evolution [44], where robots undergo evolution during their operational period; see scheme C and scheme D in Figure 1. ...
Context 3
... provide feedback for the learning process, an overhead camera and attached localization server track the robotʼs position using ReacTIVision. 2 This software implements tracking of fidu- cial markers ( Figure 10) that are printed on the cover of each robot. The software captures output from the camera, analyzes it to find the markersʼ positions on the screen, and sends the infor- mation about the ID and the position to a client application. ...
Context 4
... speed of the 3D printer is a crucial factor here; in our case 3 h was needed to print one block. The initial population consisted of two robots, the spider and the gecko, as shown in Figure 12 (left). The extended population shown in Figure 12 (right) is explained further in Appendix 2. ...
Context 5
... initial population consisted of two robots, the spider and the gecko, as shown in Figure 12 (left). The extended population shown in Figure 12 (right) is explained further in Appendix 2. ...
Context 6
... new morphogenesis process started with this genome; component parts were printed and as- sembled. The image on the left-hand side of Figure 13 shows the result, the first robot baby par- ented by two parent robots in a real-world (not simulated) environment. Figure 12. ...
Context 7
... image on the left-hand side of Figure 13 shows the result, the first robot baby par- ented by two parent robots in a real-world (not simulated) environment. Figure 12. Overview of the entire habitat and the initial population consisting of the spider and the gecko (left ) and the extended population with the parents and the offspring in the habitat (right ). ...
Context 8
... in the definition of RoboGen), the mutation is valid as well, meaning that it does not need to follow any paradigm (such as color ) from the parent genomes. To illustrate this notion, Figure 16 shows a mutated child where the leaf of the right subtree was exchanged for two red nodes. ...
Context 9
... Attachment position on parent part: 0-3 2. Part type 3. Unique identifier Figure 15. Offspring after recombination operator is applied. ...
Similar publications
The Robotic Remote Laboratory (RRL) controls the Robot labs via the Internet and applies the Robot experiment in easy and advanced way. If we want to enhance the RRL system, we must study requirements of the Robot experiment deeply. One of key requirements of the Robot experiment is the Control algorithm, which includes all important activities to...
Citations
... Dog robots can increase bonding but technological limits result in users ending the interaction (Löffler, et al. 2020). An evolutionary robot morphology aims at evolving the design and software of the robot autonomously (Jelisavcic, et al. 2017). By using evolutionary principles to solve problems by changing the morphology, software, etc. of the robot (Doncieux, et al. 2015). ...
... These four are mentioned by Löffler, et al. (2020) although we termed them as morphologies. Amorphism is where the robot is designed with a lack of order or changes over time akin to evolutionary robots mentioned by Jelisavcic, et al. (2017). One morphology not previously identified is Neomorphism which applies where the morphology of the robot feels new or is a combination of multiple. ...
Robots are becoming increasingly prevalent in the workplace. As Industry 5.0 pursues human-centric technologies, a greater understanding of what effects aesthetics has on those interacting with robots is needed. This paper sets out robot morphology as a way to characterise key form types, and proposes seven classifications: anthropomorphism, zoomorphism, phytomorphism, artemorphism, functiomorphism, amorphism, and neomorphism. Through an assessment of the current robot aesthetic landscape, design dimensions are identified with examples that can inform future robot design.
... Here again, the majority of related work consists of applying learning algorithms to the evolvable brains of robots with fixed bodies 17-26 . However, it has been argued that morphological robot evolution must include a learning stage immediately after reproduction 27 , and some recent studies have investigated the combination of body evolution, brain evolution, and learning [28][29][30][31][32][33][34][35][36] . ...
Evolutionary robot systems offer two principal advantages: an advanced way of developing robots through evolutionary optimization and a special research platform to conduct what-if experiments regarding questions about evolution. Our study sits at the intersection of these. We investigate the question “What if the 18th-century biologist Lamarck was not completely wrong and individual traits learned during a lifetime could be passed on to offspring through inheritance?” We research this issue through simulations with an evolutionary robot framework where morphologies (bodies) and controllers (brains) of robots are evolvable and robots also can improve their controllers through learning during their lifetime. Within this framework, we compare a Lamarckian system, where learned bits of the brain are inheritable, with a Darwinian system, where they are not. Analyzing simulations based on these systems, we obtain new insights about Lamarckian evolution dynamics and the interaction between evolution and learning. Specifically, we show that Lamarckism amplifies the emergence of ‘morphological intelligence’, the ability of a given robot body to acquire a good brain by learning, and identify the source of this success: newborn robots have a higher fitness because their inherited brains match their bodies better than those in a Darwinian system.
... To gain the full benefit of the evolutionary approach, it is desirable to do both, but morphological evolution is very challenging to implement with real robot hardware, since it requires a system capable of realising a wide range of body plans with highly variable requirements. For this reason, evolution of real robots is often only applied at the controller level, with few examples of morphological evolution progressing beyond simulated models (Pollack and Lipson, 2000;Hiller and Lipson, 2011;Brodbeck et al., 2015;Jelisavcic et al., 2017;Auerbach et al., 2018;Kriegman et al., 2020a;Kriegman et al., 2020b). However, to achieve the objective of evolving robot bodies that are of practical use in real-world applications, they must be implemented in hardware (Eiben, 2014). ...
... The star topology described in Section 2.4.1 defines a maximum number of 8 organs that can be connected directly to the head. Although this limit is greater than other platforms in literature (Jelisavcic et al., 2017;Auerbach et al., 2018;Miras et al., 2020), the genome decoding has to accommodate this limit. ...
The evolutionary robotics field offers the possibility of autonomously generating robots that are adapted to desired tasks by iteratively optimising across successive generations of robots with varying configurations until a high-performing candidate is found. The prohibitive time and cost of actually building this many robots means that most evolutionary robotics work is conducted in simulation, but to apply evolved robots to real-world problems, they must be implemented in hardware, which brings new challenges. This paper explores in detail the design of an example system for realising diverse evolved robot bodies, and specifically how this interacts with the evolutionary process. We discover that every aspect of the hardware implementation introduces constraints that change the evolutionary space, and exploring this interplay between hardware constraints and evolution is the key contribution of this paper. In simulation, any robot that can be defined by a suitable genetic representation can be implemented and evaluated, but in hardware, real-world limitations like manufacturing/assembly constraints and electrical power delivery mean that many of these robots cannot be built, or will malfunction in operation. This presents the novel challenge of how to constrain an evolutionary process within the space of evolvable phenotypes to only those regions that are practically feasible: the viable phenotype space. Methods of phenotype filtering and repair were introduced to address this, and found to degrade the diversity of the robot population and impede traversal of the exploration space. Furthermore, the degrees of freedom permitted by the hardware constraints were found to be poorly matched to the types of morphological variation that would be the most useful in the target environment. Consequently, the ability of the evolutionary process to generate robots with effective adaptations was greatly reduced. The conclusions from this are twofold. 1) Designing a hardware platform for evolving robots requires different thinking, in which all design decisions should be made with reference to their impact on the viable phenotype space. 2) It is insufficient to just evolve robots in simulation without detailed consideration of how they will be implemented in hardware, because the hardware constraints have a profound impact on the evolutionary space.
... Learning allows to fine-tune the coupling between body and brain and also allows for more degrees of freedom that account for environmental changes. The majority of such research consists of applying learning algorithms to the evolvable brains of robots with fixed bodies [ [16][17][18][19][20][21][22][23][24][25], but some have also evolved the morphologies [ [26][27][28][29][30][31][32][33][34][35]. ...
Evolutionary robot systems offer two principal advantages: an advanced way of developing robots through evolutionary optimization and a special research platform to conduct what-if experiments regarding questions about evolution. Our study sits at the intersection of these. We investigate the question ''What if the 18th-century biologist Lamarck was not completely wrong and individual traits learned during a lifetime could be passed on to offspring through inheritance?'' We research this issue through simulations with an evolutionary robot framework where morphologies (bodies) and controllers (brains) of robots are evolvable and robots also can improve their controllers through learning during their lifetime. Within this framework, we compare a Lamarckian system, where learned bits of the brain are inheritable, with a Darwinian system, where they are not. Analyzing simulations based on these systems, we obtain new insights about Lamarckian evolution dynamics and the interaction between evolution and learning. Specifically, we show that Lamarckism amplifies the emergence of 'morphological intelligence', the ability of a given robot body to acquire a good brain by learning, and identify the source of this success: 'newborn' robots have a higher fitness because their inherited brains match their bodies better than those in a Darwinian system.
... of Joint modules (Fig. 1c). The Core module is unique for each robot and represents the robot "head" that, in the original physical incarnation [27], contains the main logic board and the battery. The Core module has four connection points where other modules can be attached. ...
... Connected neighbor CPG nodes can synchronize the oscillations of their joints and induce global locomotion patterns. The number of configurable parameters for a robot brain is thus j + c , where j is the number of joints, and c is the number of connections between joints; in Fig. 2 we show an example of how the nodes would be connected in a robot made of 8 joints configured as in the "spider" robot from [27]. ...
In the field of evolutionary robotics, choosing the correct genetic representation is a complicated and delicate matter, especially when robots evolve behaviour and morphology at the same time. One principal problem is the lack of methods or tools to investigate and compare representations. In this paper we introduce and evaluate such a tool based on the biological notion of heritability. Heritability captures the proportion of phenotypic variation caused by genotypic variation and is often used to better understand the transmissibility of traits in real biological systems. As a proof of concept, we compare the heritability of various robot traits in two systems, one using a direct (tree based) representation and one using an indirect (grammar based) representation. We measure changes in heritability during the course of evolution and investigate how direct and indirect representation can be biased towards more exploration or exploitation throughout the course of evolution. The empirical study shows that heritability can be a useful tool to analyze different representations without running complete evolutionary processes using them.
... There are examples where the evolution of morphology is done in simulation before a select few individuals are built and tested in the real world [8], but this often comes with a large decrease in performance due to the reality gap. There are a few examples of work that combines predefined modules in new ways to build robot bodies in the real world [9,10]. This allows adapting the bodies of robots to the environment they operate in, but this is still a bit limited due to the simple nature of the robots, relative slow process of building, and need for human intervention. ...
For many years, the main target of robotics research has been how to improve control and behavior to allow robots to solve more challenging tasks. Recently, the concept of embodied intelligence has shifted this focus from pure control to a more holistic approach. The mind, body, environment, and the interactions between all these are all highly important for the performance of a robot system. This has previously been investigated in virtual robots in simulation or very simple physical robots in the lab, but the benefits for more advanced and capable real-world robots are still to be determined. We present a case study with several examples of real-world adaptation of morphology, and discuss the benefits and challenges as this direction becomes more important in the field of robotics. Taking an embodied approach promises more adaptive and robust robots, but there are many challenges that need to be addressed before wide-spread adoption in the robotics community.
... The field encompasses the use of artificial EAs to design morphology [4], control [5] or a combination of the two for both simulated and physical robots [6][7][8][9]. Furthermore, it has progressed to cover evolution with a wide range of materials, ranging from soft, flexible materials that can be easily deformed under pressure to provide adaptation [10], through modular approaches that evolve configurations of pre-existing components [11] to approaches that exploit the ability to rapidly print hard-plastic components in a diverse array of forms [12]. ...
We survey and reflect on how learning (in the form of individual learning and/or culture) can augment evolutionary approaches to the joint optimization of the body and control of a robot. We focus on a class of applications where the goal is to evolve the body and brain of a single robot to optimize performance on a specified task. The review is grounded in a general framework for evolution which permits the interaction of artificial evolution acting on a population with individual and cultural learning mechanisms. We discuss examples of variations of the general scheme of ‘evolution plus learning’ from a broad range of robotic systems, and reflect on how the interaction of the two paradigms influences diversity, performance and rate of improvement. Finally, we suggest a number of avenues for future work as a result of the insights that arise from the review.
This article is part of a discussion meeting issue ‘The emergence of collective knowledge and cumulative culture in animals, humans and machines’.
... However, this situation is changing rapidly and after the first major transition from "wetware" to software in the 20th century, evolution is at the verge of a second one, this time from software to hardware (Eiben and Smith 2015). Recent advances in and integration of evolutionary computation, robotics, 3D-printing, and automated assembly are enabling systems of physical robots that can autonomously reproduce and evolve (Brodbeck et al., 2015;Jelisavcic et al., 2017;Vujovic et al., 2017;Hale et al., 2019;Howard et al., 2019;Ellery 2020). ...
Rapid developments in evolutionary computation, robotics, 3D-printing, and material science are enabling advanced systems of robots that can autonomously reproduce and evolve. The emerging technology of robot evolution challenges existing AI ethics because the inherent adaptivity, stochasticity, and complexity of evolutionary systems severely weaken human control and induce new types of hazards. In this paper we address the question how robot evolution can be responsibly controlled to avoid safety risks. We discuss risks related to robot multiplication, maladaptation, and domination and suggest solutions for meaningful human control. Such concerns may seem far-fetched now, however, we posit that awareness must be created before the technology becomes mature.
... A major difficulty for this endeavor is that the processes of behaviour-and morphology-adaptation happen on different time-scales: While the adaptation of behaviour and movement strategies can be done in seconds, the change of design or morphological parameters of robots requires hours or even days for the production of a new robot body. This problem has lead to three main strategies for the adaptation of robot morphology: Algorithms optimize morphologies using (1) data generated from known analytical models or (differentiable) simulators [7,30,33,48,19], (2) limited small-scale robots which can be produced within hours or minutes [30,16,15,26], or (3) re-configurable/ shape-shifting robots [23,36]. Relying exclusively on simulations or analytical models is by far the most popular strategy, because it allows the generation of vast numbers of robot populations and generations due to today's easy access to computational resources and the possibility to mass-parallelize simulations [10]. ...
The co-adaptation of robot morphology and behaviour becomes increasingly important with the advent of fast 3D-manufacturing methods and efficient deep reinforcement learning algorithms. A major challenge for the application of co-adaptation methods to the real world is the simulation-to-reality-gap due to model and simulation inaccuracies. However, prior work focuses primarily on the study of evolutionary adaptation of morphologies exploiting analytical models and (differentiable) simulators with large population sizes, neglecting the existence of the simulation-to-reality-gap and the cost of manufacturing cycles in the real world. This paper presents a new approach combining classic high-frequency deep neural networks with computational expensive Graph Neural Networks for the data-efficient co-adaptation of agents with varying numbers of degrees-of-freedom. Evaluations in simulation show that the new method can co-adapt agents within such a limited number of production cycles by efficiently combining design optimization with offline reinforcement learning, that it allows for the direct application to real-world co-adaptation tasks in future work
... Each robot is composed of three different types of modules: one Core module (Figure 1a), an arbitrary number of Brick modules (Figure 1b), and an arbitrary number of Joint modules (Figure 1c). The Core module is unique for each robot and represents the robot "head" that, in the original physical incarnation [11], contains the main logic board and the battery. The Core module has four connection points where other modules can be attached. ...
... Connected neighbor CPG nodes can synchronize the oscillations of their joints and induce global locomotion patterns. The number of configurable parameters for a robot brain is thus j+c, where j is the number of joints, and c is the number of connections between joints; in Figure 2 we show an example of how the nodes would be connected in a robot made of 8 joints configured as in the "spider" robot from [11]. ...
In the field of evolutionary robotics, choosing the correct encoding is very complicated, especially when robots evolve both behaviours and morphologies at the same time. With the objective of improving our understanding of the mapping process from encodings to functional robots, we introduce the biological notion of heritability, which captures the amount of phenotypic variation caused by genotypic variation. In our analysis we measure the heritability on the first generation of robots evolved from two different encodings, a direct encoding and an indirect encoding. In addition we investigate the interplay between heritability and phenotypic diversity through the course of an entire evolutionary process. In particular, we investigate how direct and indirect genotypes can exhibit preferences for exploration or exploitation throughout the course of evolution. We observe how an exploration or exploitation tradeoff can be more easily understood by examining patterns in heritability and phenotypic diversity. In conclusion, we show how heritability can be a useful tool to better understand the relationship between genotypes and phenotypes, especially helpful when designing more complicated systems where complex individuals and environments can adapt and influence each other.
