Interaction-Aware Trajectory Repair in Compliance with Formalized Traffic Rules

Traffic-rule compliance is crucial for motion planning of automated vehicles. If an initially-planned trajectory violates traffic rules, we suggest to repair it instead of completely replanning it to save computational time. However, there exists no trajectory repair framework that considers the interactions among traffic participants, potentially leading to conservative driving behaviors. To address this issue, we propose for the first time an interaction-aware trajectory repair algorithm based on game theory. Our novel algorithm predicts the influence of the repaired trajectory on other traffic participants and then executes the trajectory candidate with the best outcome. To demonstrate our repair mechanism, we integrate it into a receding-horizon motion planning framework. Our approach is evaluated using the CommonRoad benchmark suite, revealing that-compared to the interaction-unaware repair strategy-our approach avoids unnecessarily conservative driving behaviors and achieves a higher repair rate.