Efficient Black-box Assessment of Autonomous Vehicle Safety

Abstract

While autonomous vehicle (AV) technology has shown substantial progress, we still lack tools for rigorous and scalable testing. Real-world testing, the $\text{de-facto}$ evaluation method, is dangerous to the public. Moreover, due to the rare nature of failures, billions of miles of driving are needed to statistically validate performance claims. Thus, the industry has largely turned to simulation to evaluate AV systems. However, having a simulation stack alone is not a solution. A simulation testing framework needs to prioritize which scenarios to run, learn how the chosen scenarios provide coverage of failure modes, and rank failure scenarios in order of importance. We implement a simulation testing framework that evaluates an entire modern AV system as a black box. This framework estimates the probability of accidents under a base distribution governing standard traffic behavior. In order to accelerate rare-event probability evaluation, we efficiently learn to identify and rank failure scenarios via adaptive importance-sampling methods. Using this framework, we conduct the first independent evaluation of a full-stack commercial AV system, Comma AI's OpenPilot.

References in corpus (8)

• Variational Inference with Normalizing Flows
• On a Formal Model of Safe and Scalable Self-driving Cars
• Certifying Some Distributional Robustness with Principled Adversarial Training
• CARLA: An Open Urban Driving Simulator
• Provable Bounds for Learning Some Deep Representations
• Deep Imitative Models for Flexible Inference, Planning, and Control
• FuzzerGym: A Competitive Framework for Fuzzing and Learning
• VERIFAI: A Toolkit for the Design and Analysis of Artificial Intelligence-Based Systems

Cited by in corpus (12)

• A Survey of End-to-End Driving: Architectures and Training Methods
• A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systems
• Neural Bridge Sampling for Evaluating Safety-Critical Autonomous Systems
• Non-Gaussian Chance-Constrained Trajectory Planning for Autonomous Vehicles under Agent Uncertainty
• Neural Network Guided Evolutionary Fuzzing for Finding Traffic Violations of Autonomous Vehicles
• AdvSim: Generating Safety-Critical Scenarios for Self-Driving Vehicles
• Multimodal Safety-Critical Scenarios Generation for Decision-Making Algorithms Evaluation
• Self-Improving Safety Performance of Reinforcement Learning Based Driving with Black-Box Verification Algorithms
• Transfer Importance Sampling -- How Testing Automated Vehicles in Multiple Test Setups Helps With the Bias-Variance Tradeoff
• FormulaZero: Distributionally Robust Online Adaptation via Offline Population Synthesis
• Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures
• Evaluation Framework for Performance Limitation of Autonomous Systems under Sensor Attack