Adversarial Evaluation of Autonomous Vehicles in Lane-Change Scenarios

Chen, Baiming; Li, Liang

doi:10.1109/tits.2021.3091477

Cited by 40 publications

(21 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[58] uses A2C [115] to control one surrounding vehicle in the car following scenarios. [55] uses DDPG [116] to generate adversarial policy to control surrounding agents to generate lane changes scenarios. [56] uses Multi-agent DDPG [117] to control two surrounding vehicles (which are called Nonplayer Characters) to attack the ego vehicle.…”

Section: Adversarial Policymentioning

confidence: 99%

A Survey on Safety-Critical Driving Scenario Generation -- A Methodological Perspective

Ding¹,

Xu²,

Arief³

et al. 2022

Preprint

View full text Add to dashboard Cite

Autonomous driving systems have witnessed a significant development during the past years thanks to the advance in machine learning-enabled sensing and decision-making algorithms. One critical challenge for their massive deployment in the real world is their safety evaluation. Most existing driving systems are still trained and evaluated on naturalistic scenarios collected from daily life or heuristically-generated adversarial ones. However, the large population of cars, in general, leads to an extremely low collision rate, indicating that the safety-critical scenarios are rare in the collected real-world data. Thus, methods to artificially generate scenarios become crucial to measure the risk and reduce the cost. In this survey, we focus on the algorithms of safety-critical scenario generation in autonomous driving. We first provide a comprehensive taxonomy of existing algorithms by dividing them into three categories: data-driven generation, adversarial generation, and knowledge-based generation. Then, we discuss useful tools for scenario generation, including simulation platforms and packages. Finally, we extend our discussion to five main challenges of current works -fidelity, efficiency, diversity, transferability, controllability -and research opportunities lighted up by these challenges.

show abstract

Section: Adversarial Policymentioning

confidence: 99%

A Survey on Safety-Critical Driving Scenario Generation -- A Methodological Perspective

Ding¹,

Xu²,

Arief³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…(i) NPC vehicles: parameters for a control policy (which, when given the current state information, outputs next location [53], [54], [58] and speed [54], [58], or acceleration [59], [69]) (ii) NPC pedestrians: parameters for a control policy [54] (which, when given the current state information, outputs next location and speed).…”

Section: Control At Every Stepmentioning

confidence: 99%

“…A scenario is considered critical if a safety measure is below a user-specified threshold. Two of the most commonly used measures are the minimum distance between the ego car and any other objects [43], [46], [47], [48], [53], [54], [55], [57], [59], [60], [62], [65], [66], [67], [68], [69], [70], [72], [73], [74] during the simulation (denoted as min distance) and minimum Time To Collision [100] (or its variants) between the ego car and any of other objects [43], [45], [51], [52], [61], [62], [63] during the simulation (denoted as min TTC). Many works (e.g.…”

Section: Misbehavior Propertymentioning

confidence: 99%

“…When this value is equal to zero, the ego car is considered to drive to an invalid region. Task Failure scenarios are those where the ego car fails to reach its original destination which is usually due to a deadlock with an NPC vehicle [47], [69], [74].…”

Section: Misbehavior Propertymentioning

confidence: 99%

“…These properties describe the behaviors of the NPC vehicles and usually also contribute to the responsibility of the ego car when an accident happens. Rule-Following scenarios are those in which the NPC vehicles must follow the traffic rules [69]. Natural scenarios are those in which all NPC vehicles do not collide with each other or walls and arrive at the destination on time [53].…”

Section: Npc Propertymentioning

confidence: 99%

See 2 more Smart Citations

A Survey on Scenario-Based Testing for Automated Driving Systems in High-Fidelity Simulation

Zhong¹,

Tang²,

Zhou³

et al. 2021

Preprint

View full text Add to dashboard Cite

Automated Driving Systems (ADSS) have seen rapid progress in recent years. To ensure the safety and reliability of these systems, extensive testings are being conducted before their future mass deployment. Testing the system on the road is the closest to real-world and desirable approach, but it is incredibly costly. Also, it is infeasible to cover rare corner cases using such real-world testing. Thus, a popular alternative is to evaluate an ADS's performance in some well-designed challenging scenarios, a.k.a. scenario-based testing. High-fidelity simulators have been widely used in this setting to maximize flexibility and convenience in testing what-if scenarios. Although many works have been proposed offering diverse frameworks/methods for testing specific systems, the comparisons and connections among these works are still missing. To bridge this gap, in this work, we provide a generic formulation of scenario-based testing in high-fidelity simulation and conduct a literature review on the existing works. We further compare them and present the open challenges as well as potential future research directions.

show abstract