A new paper proposes a hierarchical Bayesian credibility framework to address the challenges of pricing liability for autonomous vehicles. The framework accounts for sparse data, shifting operational domains, and non-stationary risk across software updates by pooling information across cities, software versions, and territories using a learned operational design domain (ODD) similarity kernel. This approach was demonstrated on Waymo crash data and showed that partial pooling significantly outperforms no pooling, with the learned kernel offering advantages that become detectable with more deployed cities. AI
RANK_REASON The cluster contains a research paper published on arXiv detailing a new methodology for pricing autonomous vehicle liability. [lever_c_demoted from research: ic=1 ai=0.7]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →