On the Road to Portability: Compressing End-to-End Motion Planner for Autonomous Driving
Authors:
- 16 June 2024
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 15099-15108
Abstract
End-to-end motion planning models equipped with deep neural networks have shown great potential for enabling full autonomous driving. However, the oversized neu-ral networks render them impractical for deployment on resource-constrained systems, which unavoidably requires more computational time and resources during reference. To handle this, knowledge distillation offers a promising approach that compresses models by enabling a smaller stu-dent model to learn from a larger teacher model. Neverthe-less, how to apply knowledge distillation to compress motion planners has not been explored so far. In this paper, we propose PlanKD, the first knowledge distillation frame-work tailored for compressing end-to-end motion planners. First, considering that driving scenes are inherently complex, often containing planning-irrelevant or even noisy in-formation, transferring such information is not beneficial for the student planner. Thus, we design an information bottleneck based strategy to only distill planning-relevant information, rather than transfer all information indiscrim-inately. Second, different waypoints in an output planned trajectory may hold varying degrees of importance for motion planning, where a slight deviation in certain crucial waypoints might lead to a collision. Therefore, we devise a safety-aware waypoint-attentive distillation module that as-signs adaptive weights to different waypoints based on the importance, to encourage the student to accurately mimic more crucial waypoints, thereby improving overall safety. Experiments demonstrate that our PlanKD can boost the performance of smaller planners by a large margin, and significantly reduce their reference time.Keywords
All Related Versions
Funding Information
- NSFC (62122013,U2001211)
This publication has 36 references indexed in Scilit:
- GLAMD: Global and Local Attention Mask Distillation for Object Detectors2022
- Cross-Modality Knowledge Distillation Network for Monocular 3D Object Detection2022
- ST-P3: End-to-End Vision-Based Autonomous Driving via Spatial-Temporal Feature Learning2022
- Complex Trajectory Tracking Using PID Control for Autonomous DrivingInternational Journal of Intelligent Transportation Systems Research, 2019
- Knowledge Adaptation for Efficient Semantic Segmentation2019
- Structured Knowledge Distillation for Semantic Segmentation2019
- CIRL: Controllable Imitative Reinforcement Learning for Vision-Based Self-driving2018
- End-to-End Learning of Driving Models from Large-Scale Video Datasets2017
- Deep Residual Learning for Image Recognition2016
- Optimizing the Gaussian kernel function with the formulated kernel target alignment criterion for two-class pattern classificationPattern Recognition, 2013