![[2602.16073] ScenicRules: An Autonomous Driving Benchmark with Multi-Objective Specifications and Abstract Scenarios](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.16073] ScenicRules: An Autonomous Driving Benchmark with Multi-Objective Specifications and Abstract Scenarios
Summary
The paper presents ScenicRules, a benchmark for evaluating autonomous driving systems that balances multiple objectives like safety and efficiency in complex environments.
Why It Matters
As autonomous driving technology evolves, it is crucial to have robust benchmarks that accurately reflect real-world complexities. ScenicRules addresses the gap in existing benchmarks by incorporating prioritized multi-objective specifications, enhancing the evaluation of driving systems in stochastic environments. This has significant implications for improving safety and performance in autonomous vehicles.
Key Takeaways
- ScenicRules introduces a benchmark for autonomous driving systems focused on multi-objective evaluation.
- The framework allows for prioritization of driving objectives, improving alignment with human driving judgments.
- The benchmark includes diverse scenarios to test agent performance in near-accident situations.
Computer Science > Robotics arXiv:2602.16073 (cs) [Submitted on 17 Feb 2026] Title:ScenicRules: An Autonomous Driving Benchmark with Multi-Objective Specifications and Abstract Scenarios Authors:Kevin Kai-Chun Chang, Ekin Beyazit, Alberto Sangiovanni-Vincentelli, Tichakorn Wongpiromsarn, Sanjit A. Seshia View a PDF of the paper titled ScenicRules: An Autonomous Driving Benchmark with Multi-Objective Specifications and Abstract Scenarios, by Kevin Kai-Chun Chang and 4 other authors View PDF HTML (experimental) Abstract: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...
