Fig 1 - uploaded by Li Li
Content may be subject to copyright.
Source publication
This paper presents a collision free car following model for connected automated vehicles. It considers different kinds of delays during driving, and build a model to make the best usage of each delay. The model is efficient, safe and stable.
Contexts in source publication
Context 1
... the perspective of Vehicle-to-Vehicle (V2V) communication technology's development, the evolution of vehicle driving can be divided into four stages, as shown in Fig. 1. Stage 1 is the traditional human-driving pattern, which is now still the main mode in real life. Stage 2 is the info-assistant driving pattern, in which the vehicle can obtain information of the surrounding environment (including road conditions, signal status at intersections, and acceleration, speed, and location of other vehicles) ...
Context 2
... 2.2 s whenever the vehicle is travelling faster than 15 m/s ( . A time headway of 1.1 s is chosen for comparison because of the high efficiency the CFFM demonstrated in the last section. A platoon of 5 vehicles are simulated and the simulation step is set as 0.1 s. Simulation results of the FVs with ACC and CFFM control are shown in Fig. 9 and Fig. 10, respectively, where both use the exactly same trajectory for the first vehicle. With the help of V2V communication technology, small communication delays and high communication frequency in CFFM make it possible to keep a very small following gap (about 0.6 s), which results in a much higher efficiency compared with ACC. As for string ...
Context 3
... occurs, for example, in emergency situations, such as a sudden appearance of pedestrian ahead, and gentle braking examples include situations when the head vehicle sees/anticipates a slowdown (e.g., downstream queue) or a complete stop (e.g., a red light). Simulation results for the CACC equipped platoon and CFFM equipped platoon are shown in Fig. 11 and Fig. 12, ...
Context 4
... for example, in emergency situations, such as a sudden appearance of pedestrian ahead, and gentle braking examples include situations when the head vehicle sees/anticipates a slowdown (e.g., downstream queue) or a complete stop (e.g., a red light). Simulation results for the CACC equipped platoon and CFFM equipped platoon are shown in Fig. 11 and Fig. 12, ...
Context 5
... time setting for CACC. In conclusion, though the CFFM is not as good as CACC in terms of string stability, it performs better when it comes to safety and efficiency. (Newell, 2002). In this theory, an nth vehicle follows the same trajectory as the (n-1)th vehicle except for a translation in space dn and time í µí¼ í µí± , as shown in Eq. (12). Fig. 13(a) shows how the nth vehicle changes its speed according to the (n-1)th vehicle in Newell's theory (a piecewise linear approximation). The initial speed of both nth and (n-1)th vehicle is vn, with a stable space headway Sn, and the final speed for both the nth and the (n-1)th vehicle and their space headway are vn' and Sn', respectively. ...
Context 6
... initial speed of both nth and (n-1)th vehicle is vn, with a stable space headway Sn, and the final speed for both the nth and the (n-1)th vehicle and their space headway are vn' and Sn', respectively. Under this rule, compared with the (n-1)th vehicle, the nth vehicle changes its speed with a í µí¼ í µí± delay in time and a dn advance in space. Fig. 13(b) shows part of the five vehicles' trajectories in Fig. 12, which is based on CFFM. Obviously, the two sub-figures look very similar. A close look at the raw data of the CFFM reveals that during the stable following phase, the trajectory of the FV is simply a translation of the trajectory of its PV by a distance Dn and a time Tn, where ...
Context 7
... stable space headway Sn, and the final speed for both the nth and the (n-1)th vehicle and their space headway are vn' and Sn', respectively. Under this rule, compared with the (n-1)th vehicle, the nth vehicle changes its speed with a í µí¼ í µí± delay in time and a dn advance in space. Fig. 13(b) shows part of the five vehicles' trajectories in Fig. 12, which is based on CFFM. Obviously, the two sub-figures look very similar. A close look at the raw data of the CFFM reveals that during the stable following phase, the trajectory of the FV is simply a translation of the trajectory of its PV by a distance Dn and a time Tn, where Dn equals to the speed of the FV multiplied by the total ...
Context 8
... Part of trajectories of five vehicles of CFFM in Fig.12. Fig. 13 Comparison between Newell's trajectory replication rule and the CFFM. ...
Similar publications
Quantifying drivers' perceived risk is important in the design and evaluation of the behaviour of automated vehicles (AVs) and in predicting takeovers by the driver. A 'Driver's Risk Field' (DRF) function has been previously shown to be able to predict manual driving behaviour in several simulated scenarios. In this paper, we tested if the DRF-base...
Model-based and learning-based methods are two major types of methodologies to model car following behaviors. Model-based methods describe the car-following behaviors with explicit mathematical equations, while learning-based methods focus on getting a mapping between inputs and outputs. Both types of methods have advantages and weaknesses. Meanwhi...
Various measures are being implemented from 2017 to 2019 to make the city center of Oslo car free. In this paper we discuss effects and consequences of reduced car-access to the 1,5 km 2 area. Some of the already introduced measures are removal of on-street parking spaces and implementing of a new driving pattern. In this paper we present findings...
Citations
... However, no study has made the Gipps model collision-free. Furthermore, there is an increasing need for collision-free CF models for ACC because ACC controls the CF maneuvers of CAVs without collision risk (Feng et al., 2021;Li & Wanjing, 2017). ...
Car following (CF) models are used in microscopic traffic simulation tools to help assess the effects of a new road design or to assess the effect of change in traffic flow. In 1981, Gipps developed a collision avoidance CF model using the Newtonian laws of motion to describe the motion of each vehicle in a stream of traffic. It is one of the most widely used CF models in both research and practice. Although it is claimed that the Gipps model produces collision-free results, the model produces a collision when the intention of the following vehicle is to brake harder than the perceived deceleration of lead vehicle. For the ease of simulations, a traffic simulation tool is expected to not show unrealistic crashes. This study was carried out to make the Gipps model collision-free in all conditions. It first highlights the conditions where the original Gipps model produces a collision. Then the study derives an equation for a collision-free Gipps CF model. This modified Gipps CF model produces collision-free results that always maintain a safe spacing with the lead vehicle.
... Li and Ma [21] propose a Collision-Free car-Following Model (CFFM) for connected automated vehicles that can prevent rear-end collisions when the preceding vehicle suddenly brakes. The first step is to obtain an acceleration set that can ensure safety under predictable dangerous conditions, and then to decide on the exact acceleration by considering efficiency and comfort. ...
Validating the safety of Autonomous Vehicles (AVs) operating in open-ended, dynamic environments is challenging as vehicles will eventually encounter safety-critical situations for which there is not representative training data. By increasing the coverage of different road and traffic conditions and by including corner cases in simulation-based scenario testing, the safety of AVs can be improved. However, the creation of corner case scenarios including multiple agents is non-trivial. Our approach allows engineers to generate novel, realistic corner cases based on historic traffic data and to explain why situations were safety-critical. In this paper, we introduce Probabilistic Lane Graphs (PLGs) to describe a finite set of lane positions and directions in which vehicles might travel. The structure of PLGs is learnt directly from spatio-temporal traffic data. The graph model represents the actions of the drivers in response to a given state in the form of a probabilistic policy. We use reinforcement learning techniques to modify this policy and to generate realistic and explainable corner case scenarios which can be used for assessing the safety of AVs.
... Studies that focus on the influence of AVs and CAVs on traffic networks show that connectivity and cooperation are crucial factors for improving traffic conditions, regardless of the availability of automation [14,15,16]. Also, CAVs have the potential to act as actuators of traffic control by receiving information from a traffic management center or the infrastructure, such as detectors, through V2I communication [17]. However, the efficiency of traffic management with CAVs highly depends on the penetration rate of CAVs. ...
Traffic management methods aim to increase the infrastructure’s capacity to lower congestion levels. Using vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-everything (V2X) connectivity technologies, connected autonomous vehicles (CAVs) have the potential to operate as actuators for traffic control. In this study, a CAV-based alternative approach for traffic management is proposed, and its performance is compared to that of lane control signals (LCS) and variable speed limits (VSL), which are also traffic management systems. When a shockwave is detected due to an incident, the CAVs on the road slow until they reach the speed of the observed shockwave (SWS), according to this proposed procedure. Thus, the incoming traffic flow towards the incident is slowed, preventing the queue behind from extending. In a simulation of the urban mobility (SUMO) environment, the suggested method is evaluated for 4800 scenarios on a three-lane highway by varying the market penetration rate of CAVs in traffic flow, the control distances, the incident lane, and the duration. The proposed method reduces the incidence of density values of over 38 veh/km/lane and 28 veh/km/lane in the vicinity of the incident region by 12.68 and 8.15 percent, respectively. Even at low CAV market penetration rates, the suggested technique reduces traffic density throughout the network and in the location of the incident site by twice as much as the LCS system application.
... Vehicle following, the most common drivers' behavior in traffic, exerts more important influence on many factors including traffic flow characteristics, traffic safety, and traffic simulation results. The car-following model serves as a basic algorithm of traffic simulation tools (such as SUMO and VIS-SIM) and an indispensable control algorithm for automated vehicles [1,2]. The model is aimed at replicating drivers' car-following behavior. ...
... (1) A car-following decision algorithm based on Generative Adversarial Networks is proposed, and a method for establishing a driving decision database based on Euclidean distance as a decision index is proposed (2) We design the framework based on the evidence that driving for a longer time will lead to a longer reaction time of the drivers (3) Analyze the model using China's online car-hailing trajectory data ...
The car-following model describes the microscopic behavior of the vehicle. However, the existing car-following models set the drivers’ reaction time to a fixed value without considering its dynamics. In order to improve the accuracy of car-following model, this paper proposes Deep Feature Learning-based Car-Following Model (DeepCF), a car-following model based on fatigue driving and Generative Adversarial Networks (GAN). The model is composed of the drivers’ reaction time model and the car-following decision algorithm. First, we regard driving fatigue as the starting point to study the influence of driving time and the acceleration of the preceding vehicle on the drivers’ reaction time, and develop a coarse-grained drivers’ reaction time model. Secondly, considering the impact of fatigue driving on car-following decisions, we utilize GAN to generate a driving decision database based on reaction time and use Euclidean distance as a decision search indicator. Finally, we conduct experiments on a real data set, and the results indicate that our DeepCF model is superior to baseline models.
Deep reinforcement learning (DRL) based car-following control (CFC) models are widely applied in the longitudinal motion control tasks of automated vehicles by self-learning for the optimal control policy. However, DRL algorithms easily produce unsafe commands and have low robustness, especially in complex car-following scenarios. To improve the DRL-based CFC model, this paper combines the deep deterministic policy gradient (DDPG) based CFC model with the deep optical flow estimation (DOFE) based CFC model that can overcome the shortcomings of DDPG-based one which is denoted as cooperative car-following model (DDPGoF). The DDPG-based CFC model utilizes prioritized experience replay which can intrinsically accelerate the learning speed. Meanwhile, the proposed DOFE-based CFC model employs the recurrent all-pairs field transforms algorithm (RAFT) and EfficientNet to perceive the motion variation of the surrounding vehicles, motorcycles, etc. The real vehicle driving data sets are applied to calibrate and validate the proposed DDPGoF-based CFC model while several assessment criteria are established to evaluate its overall performance. As a result, the DDPGoF-based CFC model is superior to DDPG-based one in avoiding crashes, improving car-following stability , riding comfort, and fuel economy of HEV.
Microscopic traffic models (MTMs) are widely used for assessing the impacts of (connected) autonomous vehicles ((C)AVs). These models utilize car‐following (CF) and lane‐changing models to replicate the (C)AVs driving behaviors. Numerous studies are being lately published regarding the approximation of the driving behaviors of (C)AVs (especially CF behavior) with many state‐of‐the‐art modelling methods. Still, there is no established CF model to mimic the accurate behavior of (C)AVs. Researchers often utilize existing mathematical CF models as well as limited data‐driven models for (C)AVs modelling. Meanwhile, several studies conduct simulation‐based impact assessments with various key performance indicators (KPIs). Identification of these KPIs is a crucial step for future studies. Hence, this paper presents a comprehensive outlook on different CF models with their adopted parameters for (C)AVs modelling and investigates how and in which aspects might the CF behaviors of (C)AVs are different from human‐driven vehicles. In addition, the recent publications in data‐driven CF models including their methodologies are explicitly discussed. This work also reviews simulation‐based studies with the reported impacts and used KPIs. Finally, in light of the findings of this paper, several future research needs are highlighted.
