C-TRAIL: A Commonsense World Framework for Trajectory Planning in Autonomous Driving

2026-03-31

AI Summary

This work addresses the unreliability of large language models (LLMs) in autonomous driving trajectory planning, which can lead to safety hazards. To mitigate this issue, the authors propose C-TRAIL, a commonsense trajectory reasoning and adaptive ensemble framework that integrates LLM-generated commonsense knowledge into motion planning through a closed-loop “recall–plan–update” mechanism. The framework incorporates dual trust assessments and introduces a Dirichlet-distribution-based trust policy to weight and inject commonsense priors into Monte Carlo Tree Search (MCTS). Trust estimates and planning policies are jointly updated in response to environmental feedback. Experimental results demonstrate that C-TRAIL achieves an average reduction of 40.2% in Average Displacement Error (ADE) and 51.7% in Final Displacement Error (FDE), along with a 16.9 percentage-point improvement in success rate across the Highway-env, highD, and rounD datasets.

Abstract

Trajectory planning for autonomous driving increasingly leverages large language models (LLMs) for commonsense reasoning, yet LLM outputs are inherently unreliable, posing risks in safety-critical applications. We propose C-TRAIL, a framework built on a Commonsense World that couples LLM-derived commonsense with a trust mechanism to guide trajectory planning. C-TRAIL operates through a closed-loop Recall, Plan, and Update cycle: the Recall module queries an LLM for semantic relations and quantifies their reliability via a dual-trust mechanism; the Plan module injects trust-weighted commonsense into Monte Carlo Tree Search (MCTS) through a Dirichlet trust policy; and the Update module adaptively refines trust scores and policy parameters from environmental feedback. Experiments on four simulated scenarios in Highway-env and two real-world levelXData datasets (highD, rounD) show that C-TRAIL consistently outperforms state-of-the-art baselines, reducing ADE by 40.2%, FDE by 51.7%, and improving SR by 16.9 percentage points on average. The source code is available at https://github.com/ZhihongCui/CTRAIL.

Problem

• autonomous driving
• trajectory planning
• large language models
• commonsense reasoning
• safety-critical systems

Innovation

• commonsense reasoning
• trust mechanism
• trajectory planning
• Monte Carlo Tree Search
• large language models