Rethinking Imitation-based Planner for Autonomous Driving

TL;DR

This paper evaluates imitation-based autonomous driving planners using the nuPlan benchmark, identifying key features and data augmentation techniques, revealing an imitation gap, and proposing a strong baseline model with competitive performance and better generalization.

Contribution

It provides a comprehensive study on imitation-based planners, introduces the PlanTF baseline, and highlights the importance of features, data augmentation, and the imitation gap.

Findings

Pure imitation-based planner can match state-of-the-art performance.

Effective data augmentation reduces compounding errors.

Identifies an overlooked imitation gap affecting learning systems.

Abstract

In recent years, imitation-based driving planners have reported considerable success. However, due to the absence of a standardized benchmark, the effectiveness of various designs remains unclear. The newly released nuPlan addresses this issue by offering a large-scale real-world dataset and a standardized closed-loop benchmark for equitable comparisons. Utilizing this platform, we conduct a comprehensive study on two fundamental yet underexplored aspects of imitation-based planners: the essential features for ego planning and the effective data augmentation techniques to reduce compounding errors. Furthermore, we highlight an imitation gap that has been overlooked by current learning systems. Finally, integrating our findings, we propose a strong baseline model-PlanTF. Our results demonstrate that a well-designed, purely imitation-based planner can achieve highly competitive performance compared to state-of-the-art methods involving hand-crafted rules and exhibit superior generalization capabilities in long-tail cases. Our models and benchmarks are publicly available. Project website https://jchengai.github.io/planTF.