A Framework for Learning Scoring Rules in Autonomous Driving Planning Systems

2025-02-17

AI Summary

In autonomous driving trajectory scoring, handcrafted rule-based approaches suffer from difficulties in modeling complex spatiotemporal dependencies, labor-intensive parameter tuning, and poor generalization across diverse driving scenarios. To address these challenges, we propose FLoRA, a learnable temporal logic scoring framework that enables end-to-end learning of interpretable, differentiable temporal logic specifications directly from positive-only expert demonstrations—without requiring counterfactual or negative examples. FLoRA jointly optimizes both the logical structure and quantitative parameters via a differentiable temporal logic representation, structure-aware gradient optimization, and positive-example-driven learning, trained on real-world NuPlan data. Experiments in closed-loop simulation demonstrate that FLoRA significantly outperforms both handcrafted expert rules and neural scoring models, yielding substantial improvements in safety and comfort metrics. Moreover, FLoRA operates as a plug-and-play module, seamlessly integrating with diverse trajectory generators.

Technology Category

• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Search and Optimization: Learning to Search
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

In autonomous driving systems, motion planning is commonly implemented as a two-stage process: first, a trajectory proposer generates multiple candidate trajectories, then a scoring mechanism selects the most suitable trajectory for execution. For this critical selection stage, rule-based scoring mechanisms are particularly appealing as they can explicitly encode driving preferences, safety constraints, and traffic regulations in a formalized, human-understandable format. However, manually crafting these scoring rules presents significant challenges: the rules often contain complex interdependencies, require careful parameter tuning, and may not fully capture the nuances present in real-world driving data. This work introduces FLoRA, a novel framework that bridges this gap by learning interpretable scoring rules represented in temporal logic. Our method features a learnable logic structure that captures nuanced relationships across diverse driving scenarios, optimizing both rules and parameters directly from real-world driving demonstrations collected in NuPlan. Our approach effectively learns to evaluate driving behavior even though the training data only contains positive examples (successful driving demonstrations). Evaluations in closed-loop planning simulations demonstrate that our learned scoring rules outperform existing techniques, including expert-designed rules and neural network scoring models, while maintaining interpretability. This work introduces a data-driven approach to enhance the scoring mechanism in autonomous driving systems, designed as a plug-in module to seamlessly integrate with various trajectory proposers. Our video and code are available on xiong.zikang.me/FLoRA.

Problem

• Learning interpretable scoring rules for autonomous driving.
• Optimizing rules from real-world driving demonstrations.
• Enhancing trajectory selection in motion planning systems.

Innovation

• Learning interpretable scoring rules
• Optimizing rules from real-world data
• Seamless integration with trajectory proposers