Learning to falsify automated driving vehicles with prior knowledge
This work addresses the problem of scalable and rigorous testing for automated driving vehicles, which is incremental as it builds on existing falsification methods by integrating prior knowledge.
The paper tackles the challenge of testing automated driving vehicles by proposing a learning-based falsification framework that incorporates prior knowledge to limit scenario variance and guide the learning process, resulting in non-trivial falsifying scenarios with higher reward for an adaptive cruise controller compared to purely learning-based or model-based approaches.
While automated driving technology has achieved a tremendous progress, the scalable and rigorous testing and verification of safe automated and autonomous driving vehicles remain challenging. This paper proposes a learning-based falsification framework for testing the implementation of an automated or self-driving function in simulation. We assume that the function specification is associated with a violation metric on possible scenarios. Prior knowledge is incorporated to limit the scenario parameter variance and in a model-based falsifier to guide and improve the learning process. For an exemplary adaptive cruise controller, the presented framework yields non-trivial falsifying scenarios with higher reward, compared to scenarios obtained by purely learning-based or purely model-based falsification approaches.