Exploiting Connected Autonomous Vehicles to Improve Mixed Traffic Safety and Efficiency in Realistic Scenarios
Citation:
Garg, Mohit, Exploiting Connected Autonomous Vehicles to Improve Mixed Traffic Safety and Efficiency in Realistic Scenarios, Trinity College Dublin, School of Computer Science & Statistics, Computer Science, 2024Download Item:
Abstract:
Human-driven vehicles (HDVs) produce stop-and-go waves i.e., high speed variations, due to humans large-reaction times and perception errors. This leads to decrease in both traffic safety and efficiency. In contrast, connected autonomous vehicles (CAVs) exploit information from surrounding vehicles via V2V communication, which enables them to adapt their speed based on the precise information (position, speed and acceleration) from the vehicles ahead of them within their communication range. This is expected to result in improved traffic safety and efficiency in pure CAV traffic. In the near future, however, CAVs and HDVs will coexist on the road (mixed traffic), and it may take decades to transform existing transportation systems into fully connected and autonomous environments. In mixed traffic scenarios, information exchange among vehicles may not always be possible due to the presence of human-driven vehicles that do not have communication capabilities. Furthermore, as wireless vehicular networks are unreliable, information from other vehicles can be delayed or lost, which brings more challenges for CAVs in utilizing information from other vehicles. While the impact of CAVs on traffic safety and efficiency at different market penetration rates (MPRs) has been studied extensively, CAVs have typically been assumed to operate in large-scale traffic scenarios with perfect communication. The few studies that consider imperfect communication include only a very small number of vehicles. Furthermore, most existing work on investigating the impact of CAVs in mixed traffic and unreliable communication environments is based on the simple predecessor-following information flow topology, where a CAV takes information only from the vehicle it follows. In such scenarios, a CAV degrades its longitudinal driving mode to sensor-based control if the preceding vehicle is a HDV, or when it cannot obtain information due to communication failures. This results in reductions in both traffic safety and efficiency. Although traffic safety can be improved by increasing the CAV time headway, this is at the expense of a reduction in traffic efficiency. To avoid CAV mode degradation, a number of studies propose equipping HDVs with communication devices. While this approach can avoid CAV mode degradation if the preceding vehicle is a HDV, it cannot avoid degradation in case of communication failures. In addition, the cost of retrofitting HDVs with communication-capability is likely to be prohibitive for at least a proportion of their owners. Exploiting information from multiple leading vehicles within their communication range might increase the resilience of CAV control algorithms, contributing to improved traffic safety and efficiency (compared to single-vehicle information-based control). While a few researchers have attempted to develop CAV car-following control algorithms considering information from more than one leading vehicle, for CAVs operation in unreliable communication and/or mixed traffic environment, most of these studies assume that the number and type of vehicles they can get information from is fixed, which does not hold in the presence of communication failures or HDVs. In recent years, a few robust car-following control algorithms been developed for CAVs operation in mixed traffic environment that can handle varying the number and type of vehicles they can get information from, but they have only been validated in very limited scenarios for a small number of vehicles (with a fixed sequence of CAVs and HDVs) in a platoon, assuming perfect communication. These three assumptions (fully connected traffic, perfect communication or fixed sequence of vehicles) are unrealistic to investigate the impact of CAVs on mixed traffic safety and efficiency in realistic scenarios in terms of traffic composition (mixed traffic), communication (not assumed to be reliable) and traffic scenario with real traffic demand (i.e., typically a large number of vehicles). To fill these research gaps, the goal of this work is to investigate the impact of CAV penetration rates when they can exploit information received from multiple, rather than a single, leading vehicles in realistic scenarios. To accomplish this, firstly, a car-following control algorithm is designed for CAVs to be able to cope with varying numbers and types of vehicles they can get information from for CAV operation in mixed traffic and unreliable communication environments. Then, CAV controller parameters are tuned such that it can provide significant improvement in both mixed traffic safety and efficiency. Finally, in order to better adapt to varying the number and type of vehicles they can get information from, in mixed traffic and unreliable communication environments, an adaptive information weights assignment approach is developed for the proposed controller. These approaches are evaluated using simulation studies of CAVs in realistic communication and traffic scenarios at different market penetration rates, using real motorway traffic data (the M50 motorway, in Ireland). Considering realistic scenarios in terms of imperfect communication, humans large reaction time and perception errors, and traffic scenarios, these simulation studies assess the effect of different penetration rates of CAVs on traffic safety and efficiency using the timeto- collision (TTC) performance metric (for the safety evaluation), as well as the travel time metric (to evaluate traffic efficiency). Results have reported two primary findings. Firstly, they show that imperfections in V2V communication links and drivers large reaction time have adverse effects on both safety and efficiency. Secondly, they illustrate that by properly tuning of CAV controller parameters (control gains and time headways), in realistic scenarios both traffic safety and efficiency still improve significantly as the CAV penetration rate increases. The results of this work thus represent a significant step towards the deployment of CAVs on motorways in the near future.
Sponsor
Grant Number
Trinity College Dublin (TCD)
Science Foundation Ireland (SFI)
Description:
APPROVED
Author: Garg, Mohit
Sponsor:
Trinity College Dublin (TCD)Science Foundation Ireland (SFI)
Advisor:
Bouroche, MelaniePublisher:
Trinity College Dublin. School of Computer Science & Statistics. Discipline of Computer ScienceType of material:
ThesisCollections
Availability:
Full text availableMetadata
Show full item recordThe following license files are associated with this item: