車線変更シナリオにおける自動運転車の敵対的評価
Adversarial Evaluation of Autonomous Vehicles in Lane-Change Scenarios
深層強化学習で危険な敵対的環境を生成し、自動運転車を効率的に評価する適応的フレームワークを提案。車線変更シナリオで検証し、生成された敵対的環境が被験車両の性能を大きく低下させることを示した。
著者: Baiming Chen, Xiang Chen, Wu Qiong, Liang Li
分類: cs.RO, cs.LG, stat.ML
原文アブストラクト
Autonomous vehicles must be comprehensively evaluated before deployed in cities and highways. However, most existing evaluation approaches for autonomous vehicles are static and lack adaptability, so they are usually inefficient in generating challenging scenarios for tested vehicles. In this paper, we propose an adaptive evaluation framework to efficiently evaluate autonomous vehicles in adversarial environments generated by deep reinforcement learning. Considering the multimodal nature of dangerous scenarios, we use ensemble models to represent different local optimums for diversity. We then utilize a nonparametric Bayesian method to cluster the adversarial policies. The proposed method is validated in a typical lane-change scenario that involves frequent interactions between the ego vehicle and the surrounding vehicles. Results show that the adversarial scenarios generated by our method significantly degrade the performance of the tested vehicles. We also illustrate different patterns of generated adversarial environments, which can be used to infer the weaknesses of the tested vehicles.
