MapVision: CVPR 2024 Autonomous Grand Challenge Mapless Driving Tech Report

TL;DR

MapVision introduces a novel approach for mapless autonomous driving using multi-perspective images and SD maps, employing advanced encoding and detection techniques to improve scene understanding and driving accuracy.

Contribution

The paper presents a new method combining multi-view images, SD maps, and innovative detection strategies to enhance autonomous driving without HD maps.

Findings

Achieved an OLUS score of 0.58.

Improved detection of traffic elements and road areas.

Enhanced geometric encoding with map encoder pre-training.

Abstract

Autonomous driving without high-definition (HD) maps demands a higher level of active scene understanding. In this competition, the organizers provided the multi-perspective camera images and standard-definition (SD) maps to explore the boundaries of scene reasoning capabilities. We found that most existing algorithms construct Bird's Eye View (BEV) features from these multi-perspective images and use multi-task heads to delineate road centerlines, boundary lines, pedestrian crossings, and other areas. However, these algorithms perform poorly at the far end of roads and struggle when the primary subject in the image is occluded. Therefore, in this competition, we not only used multi-perspective images as input but also incorporated SD maps to address this issue. We employed map encoder pre-training to enhance the network's geometric encoding capabilities and utilized YOLOX to improve traffic element detection precision. Additionally, for area detection, we innovatively introduced LDTR and auxiliary tasks to achieve higher precision. As a result, our final OLUS score is 0.58.