LoRD: 分布シフトに適応する微分可能な運転ポリシー
LoRD: Adapting Differentiable Driving Policies to Distribution Shifts
自動運転の微分可能なスタック（予測・計画・制御）を分布シフトに適応させるため、低ランク残差デコーダ（LoRD）とマルチタスク微調整を提案し、破滅的忘却を最大23.33%改善、閉ループOOD運転スコアを9.93%向上させた。
著者: Christopher Diehl, Peter Karkus, Sushant Veer, Marco Pavone, Torsten Bertram
分類: cs.RO, cs.AI, cs.LG
原文アブストラクト
Distribution shifts between operational domains can severely affect the performance of learned models in self-driving vehicles (SDVs). While this is a well-established problem, prior work has mostly explored naive solutions such as fine-tuning, focusing on the motion prediction task. In this work, we explore novel adaptation strategies for differentiable autonomy stacks consisting of prediction, planning, and control, perform evaluation in closed-loop, and investigate the often-overlooked issue of catastrophic forgetting. Specifically, we introduce two simple yet effective techniques: a low-rank residual decoder (LoRD) and multi-task fine-tuning. Through experiments across three models conducted on two real-world autonomous driving datasets (nuPlan, exiD), we demonstrate the effectiveness of our methods and highlight a significant performance gap between open-loop and closed-loop evaluation in prior approaches. Our approach improves forgetting by up to 23.33% and the closed-loop OOD driving score by 9.93% in comparison to standard fine-tuning.
