ScenicRules: An Autonomous Driving Benchmark with Multi-Objective Specifications and Abstract Scenarios
This work addresses the need for better benchmarks in autonomous driving research by providing a tool for more rigorous evaluation, though it is incremental as it builds on existing formal modeling approaches.
The authors tackled the problem of evaluating autonomous driving systems in complex environments with conflicting objectives by introducing ScenicRules, a benchmark that combines multi-objective prioritized rules and formal scenario models, and experimental results show it aligns with human judgments and effectively exposes agent failures.
Developing autonomous driving systems for complex traffic environments requires balancing multiple objectives, such as avoiding collisions, obeying traffic rules, and making efficient progress. In many situations, these objectives cannot be satisfied simultaneously, and explicit priority relations naturally arise. Also, driving rules require context, so it is important to formally model the environment scenarios within which such rules apply. Existing benchmarks for evaluating autonomous vehicles lack such combinations of multi-objective prioritized rules and formal environment models. In this work, we introduce ScenicRules, a benchmark for evaluating autonomous driving systems in stochastic environments under prioritized multi-objective specifications. We first formalize a diverse set of objectives to serve as quantitative evaluation metrics. Next, we design a Hierarchical Rulebook framework that encodes multiple objectives and their priority relations in an interpretable and adaptable manner. We then construct a compact yet representative collection of scenarios spanning diverse driving contexts and near-accident situations, formally modeled in the Scenic language. Experimental results show that our formalized objectives and Hierarchical Rulebooks align well with human driving judgments and that our benchmark effectively exposes agent failures with respect to the prioritized objectives. Our benchmark can be accessed at https://github.com/BerkeleyLearnVerify/ScenicRules/.