Vehicle tracking from Bird-Eye view

Engineering notes

Key topics: autonomous driving, autonomous vehicle, bev, path planning, nuscenes, kitti, perception, planning. See the paper for implementation details and experimental results.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Perception algorithms in 3D space are key research contents in the fields of autonomous vehicles and robots. 3D perception in a bird's-eye view (BEV) has attracted a lot of attention in recent years, because the BEV's representation of traffic scenes contains abundant target depth information and dimensional information, and the detection results can be directly applied to downstream tasks such as trajectory tracking, path planning, etc. Motivated by this observation, We propose a tracking module based on bird's-eye view and a target correlation method optimized for tracking tasks. The method we propose has advantages over other algorithms in performance on the Nuscenes dataset and the KITTI dataset. We further demonstrate that by projecting 3D objects under the bird's-eye view, we can achieve significant improvements in monocular 3D perception tasks, including detection and tracking.

Links and sources

• Official / arXiv page

Comments