BucketKD: A Safety-Aware Bucket-Based Knowledge Distillation Framework for End-to-End Motion Planning

2026-07-12

AI Summary

This work addresses the challenge of deploying computationally intensive end-to-end motion planning models in resource-constrained autonomous driving systems, where conventional knowledge distillation approaches often discard safety-critical information during scene representation simplification. To overcome this limitation, the authors propose an adaptive binning–based knowledge distillation framework that discretizes key environmental variables into semantically rich bins and incorporates a time-to-collision (TTC)–aware waypoint attention mechanism to preserve essential planning behaviors. Experimental results on the Bench2Drive dataset in the CARLA simulation platform demonstrate that the proposed method achieves superior planning accuracy and safety compared to existing techniques, even under high model compression ratios.

Abstract

End-to-end motion planning has emerged as a promising paradigm in autonomous driving, directly mapping raw sensor data to control commands via deep neural networks. Despite its advantages, its large model size hinders deployment in resource-constrained platforms. In this paper, we present BucketKD, a bucket-based knowledge distillation framework that yields compact and safety-aware end-to-end planners. Compared to the state-of-the-art approach, which relies on simplified planning state representations, BucketKD discretizes critical environmental variables into adaptive buckets that capture richer scene semantics while preserving efficiency. In addition, we design a safety-aware waypoint attention mechanism that evaluates each waypoint's risk level by accounting for both obstacle proximity and relative motion through a time-to-collision (TTC) formulation widely used in transportation research. This enables the student model to better retain safety-critical behaviors during distillation. Extensive experiments in CARLA using the Bench2Drive dataset show that BucketKD significantly outperforms the state-of-the-art in both planning accuracy and safety while maintaining strong compression ratios.

Problem

• end-to-end motion planning
• knowledge distillation
• safety-aware
• model compression
• autonomous driving

Innovation

• bucket-based knowledge distillation
• safety-aware planning
• end-to-end motion planning
• time-to-collision (TTC)
• adaptive discretization