Wu Dongming ; Chang Jiahao ; Jia Fan ; Liu Yingfei ; Wang Tiancai ; Shen Jianbing

Abstract

Topology reasoning aims to comprehensively understand road scenes and presentdrivable routes in autonomous driving. It requires detecting road centerlines(lane) and traffic elements, further reasoning their topology relationship,i.e., lane-lane topology, and lane-traffic topology. In this work, we firstpresent that the topology score relies heavily on detection performance on laneand traffic elements. Therefore, we introduce a powerful 3D lane detector andan improved 2D traffic element detector to extend the upper limit of topologyperformance. Further, we propose TopoMLP, a simple yet high-performancepipeline for driving topology reasoning. Based on the impressive detectionperformance, we develop two simple MLP-based heads for topology generation.TopoMLP achieves state-of-the-art performance on OpenLane-V2 benchmark, i.e.,41.2% OLS with ResNet-50 backbone. It is also the 1st solution for 1st OpenLaneTopology in Autonomous Driving Challenge. We hope such simple and strongpipeline can provide some new insights to the community. Code is athttps://github.com/wudongming97/TopoMLP.