知覚・注目・運転:安全な自動運転のための空間的注意の学習
Perceive, Attend, and Drive: Learning Spatial Attention for Safe Self-Driving
運動計画に直接向けたスパース注意モジュールを備えたEnd-to-End自動運転ネットワークを提案し、重要な領域に集中計算することで安全性と解釈性を向上させた。
著者: Bob Wei, Mengye Ren, Wenyuan Zeng, Ming Liang, Bin Yang, Raquel Urtasun
分類: cs.RO, cs.CV, cs.LG
原文アブストラクト
In this paper, we propose an end-to-end self-driving network featuring a sparse attention module that learns to automatically attend to important regions of the input. The attention module specifically targets motion planning, whereas prior literature only applied attention in perception tasks. Learning an attention mask directly targeted for motion planning significantly improves the planner safety by performing more focused computation. Furthermore, visualizing the attention improves interpretability of end-to-end self-driving.
