RECTOR: Priority-Aware Rule-Based Reranking for Compliance-Aware Autonomous Driving Trajectory Selection

2026-05-24

AI Summary

This work addresses the limitations of existing autonomous driving systems that rely solely on model confidence for trajectory selection, often neglecting multi-level constraints such as safety, traffic regulations, and comfort, thereby leading to frequent violations. To remedy this, the authors propose a hierarchical rulebook-based trajectory reranking method grounded in a priority ordering: safety ≻ legality ≻ road rules ≻ comfort. Their approach integrates differentiable rule proxies and a scene-conditioned applicability mechanism, coupled with a deterministic ε-lexicographic decision policy, enabling provable rule compliance without retraining the underlying prediction model. Evaluated on the Waymo Open Motion Dataset, the method reduces safety and legal violation rates from 28.58% to 20.42% and overall violations from 40.32% to 32.41% compared to confidence-based selection, while maintaining 96% robust rejection capability under adversarial confidence perturbations.

Abstract

Autonomous driving stacks must pick one trajectory from a multi-modal candidate set; choosing by model confidence ignores safety, traffic-law, and comfort constraints. We present \textsc{RECTOR} (Rule-Enforced Constrained Trajectory Orchestrator), a post-generation reranking layer that scores candidates against a tiered rulebook (Safety~$\succ$~Legal~$\succ$~Road~$\succ$~Comfort) via differentiable proxies and a scene-conditioned applicability mechanism, then selects with a deterministic $\varepsilon$-lexicographic rule that preserves cross-tier priority by construction -- without retraining the predictor. On the Waymo Open Motion Dataset \texttt{validation\_interactive} split (43{,}219 augmented instances, $K{=}6$), under Protocol~B (28-rule proxy catalog, oracle applicability) rule-aware selection cuts Safety+Legal violations from 28.58\% to 20.42\% and Total from 40.32\% to 32.41\% versus confidence-only on the same candidates. A uniform-weight weighted-sum baseline matches binary compliance on this benchmark -- the empirical lift comes from rule-aware ranking, while the lexicographic guarantee is the structural differentiator no weight calibration can replicate. Under adversarial confidence corruption, confidence-only selection fails in 100\% of scenarios while both rule-aware selectors reject the injected mode in $\sim$96\%. All figures are proxy-evaluator results (not a safety certificate), open-loop, 5\,s horizon, U.S.\ rules, validation split.

Problem

• autonomous driving
• trajectory selection
• compliance
• safety constraints
• traffic rules

Innovation

• rule-based reranking
• lexicographic optimization
• compliance-aware trajectory selection
• autonomous driving
• differentiable rule proxies