Adaptive Stress Testing without Domain Heuristics using Go-Explore

Koren, Mark; Kochenderfer, Mykel J.

doi:10.1109/itsc45102.2020.9294729

Cited by 11 publications

(9 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…GE has two phases, a tree search exploration phase, and a DRL robustification phase. While the original version of GE uses the state of the simulator when building the tree and training the robust policy, Koren and Kochenderfer (2020) modified the algorithm to use the history of disturbances instead, reducing the access requirements of the simulator.…”

Section: Deep Reinforcement Learningmentioning

confidence: 99%

“…In an intersection scenario, the system under test approaches a stoplight , stop sign (Abeysirigoonawardena et al, 2019), crosswalk (Koren et al, 2018), or other form of intersection and must proceed through without a failure. Failures can include collisions with pedestrians (Koren & Kochenderfer, 2020) or other vehicles . Failures may also include violations of traffic laws (Kress-Gazit & Pappas, 2008) or other rule-sets, such as those designed to prevent at-fault collisions (Hekmatnejad et al, 2020).…”

Section: Autonomous Drivingmentioning

confidence: 99%

“…Within these applications, researchers have developed domain specific techniques to improve performance. The most common approach is to use miss distance, the physical distance between the test vehicle and other vehicles in the simulation, as a reward heuristic to allow easier optimization (Koren & Kochenderfer, 2020). Miss distance is also the underlying principle for robust semantics of temporal formulas when applied to autonomous vehicles (Zhang et al, 2018).…”

Section: Autonomous Drivingmentioning

confidence: 99%

See 2 more Smart Citations

A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systems

CorsoAnthony¹,

Moss

KorenMark

et al. 2021

jair

Self Cite

View full text Add to dashboard Cite

Autonomous cyber-physical systems (CPS) can improve safety and efficiency for safety-critical applications, but require rigorous testing before deployment. The complexity of these systems often precludes the use of formal verification and real-world testing can be too dangerous during development. Therefore, simulation-based techniques have been developed that treat the system under test as a black box operating in a simulated environment. Safety validation tasks include finding disturbances in the environment that cause the system to fail (falsification), finding the most-likely failure, and estimating the probability that the system fails. Motivated by the prevalence of safety-critical artificial intelligence, this work provides a survey of state-of-the-art safety validation techniques for CPS with a focus on applied algorithms and their modifications for the safety validation problem. We present and discuss algorithms in the domains of optimization, path planning, reinforcement learning, and importance sampling. Problem decomposition techniques are presented to help scale algorithms to large state spaces, which are common for CPS. A brief overview of safety-critical applications is given, including autonomous vehicles and aircraft collision avoidance systems. Finally, we present a survey of existing academic and commercially available safety validation tools.

show abstract

Section: Deep Reinforcement Learningmentioning

confidence: 99%

Section: Autonomous Drivingmentioning

confidence: 99%

Section: Autonomous Drivingmentioning

confidence: 99%

See 1 more Smart Citation

A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systems

CorsoAnthony¹,

Moss

KorenMark

et al. 2021

jair

Self Cite

View full text Add to dashboard Cite

show abstract

“…DRL has shown state-of-the art results in playing Atari games [82], playing chess [74], and robot manipulation from camera input [83]. In recent years, different DRL techniques have been applied to falsification and most-likely failure analysis [22], [69], [72], [84]- [88].…”

Section: Deep Reinforcement Learningmentioning

confidence: 99%

“…While the original version of GE uses the state of the simulator when building the tree and training the robust policy, Koren and Kochenderfer [88] modified the algorithm to use the history of disturbances instead, reducing the access requirements of the simulator. GE was used for the falsification of an autonomous vehicle and was able to find counterexamples more reliably than MCTS on problems with long horizons and BA was able to find more probable failures than MCTS, or DRL.…”

Section: Deep Reinforcement Learningmentioning

confidence: 99%

A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systems

Corso,

Moss,

Koren

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Autonomous and semi-autonomous systems for safety-critical applications require rigorous testing before deployment. Due to the complexity of these systems, formal verification may be impossible and real-world testing may be dangerous during development. Therefore, simulation-based techniques have been developed that treat the system under test as a black box during testing. Safety validation tasks include finding disturbances to the system that cause it to fail (falsification), finding the most-likely failure, and estimating the probability that the system fails. Motivated by the prevalence of safety-critical artificial intelligence, this work provides a survey of state-of-the-art safety validation techniques with a focus on applied algorithms and their modifications for the safety validation problem. We present and discuss algorithms in the domains of optimization, path planning, reinforcement learning, and importance sampling. Problem decomposition techniques are presented to help scale algorithms to large state spaces, and a brief overview of safety-critical applications is given, including autonomous vehicles and aircraft collision avoidance systems. Finally, we present a survey of existing academic and commercially available safety validation tools.

show abstract

Discovery and Analysis of Rare High-Impact Failure Modes Using Adversarial RL-Informed Sampling

Lipkis,

Agogino

2024

Autonomous Agents and Multiagent Systems. Best and Visionary Papers

View full text Add to dashboard Cite

Adaptive Stress Testing without Domain Heuristics using Go-Explore

Cited by 11 publications

References 13 publications

A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systems

A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systems

A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systems

Discovery and Analysis of Rare High-Impact Failure Modes Using Adversarial RL-Informed Sampling

Contact Info

Product

Resources

About