The Reasonable Crowd: Towards evidence-based and interpretable models of driving behavior
This work addresses the challenge of behavior specification for autonomous vehicles, providing an evidence-based and interpretable approach, though it is incremental as it builds on existing rule-based and machine learning methods in a limited domain.
The study tackled the problem of specifying desired driving behavior for autonomous vehicles by creating a dataset of 92 traffic scenarios with human preferences and comparing interpretable models (e.g., Bayesian networks) against a hand-crafted rulebook. They found that interpretable models achieved high classification accuracy using 14 expert-designed rules, and the rulebook offered high interpretability without major performance loss.
Autonomous vehicles must balance a complex set of objectives. There is no consensus on how they should do so, nor on a model for specifying a desired driving behavior. We created a dataset to help address some of these questions in a limited operating domain. The data consists of 92 traffic scenarios, with multiple ways of traversing each scenario. Multiple annotators expressed their preference between pairs of scenario traversals. We used the data to compare an instance of a rulebook, carefully hand-crafted independently of the dataset, with several interpretable machine learning models such as Bayesian networks, decision trees, and logistic regression trained on the dataset. To compare driving behavior, these models use scores indicating by how much different scenario traversals violate each of 14 driving rules. The rules are interpretable and designed by subject-matter experts. First, we found that these rules were enough for these models to achieve a high classification accuracy on the dataset. Second, we found that the rulebook provides high interpretability without excessively sacrificing performance. Third, the data pointed to possible improvements in the rulebook and the rules, and to potential new rules. Fourth, we explored the interpretability vs performance trade-off by also training non-interpretable models such as a random forest. Finally, we make the dataset publicly available to encourage a discussion from the wider community on behavior specification for AVs. Please find it at github.com/bassam-motional/Reasonable-Crowd.