一种面向自动驾驶场景的多目标跟踪方法及汽车
By using the UMRTrack target tracking model, online noise adaptive filtering and multi-source evidence robust association are adopted to solve the problems of insufficient robustness and accuracy of existing 3D multi-target tracking methods in complex traffic scenarios, and adaptive trajectory management and continuity are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUAQIAO UNIVERSITY
- Filing Date
- 2026-06-18
- Publication Date
- 2026-07-17
AI Technical Summary
Existing 3D multi-target tracking methods suffer from insufficient robustness and accuracy in complex dynamic traffic scenarios due to issues such as fixed observation noise covariance, data association cost function dependence on geometry, lack of adaptability in trajectory lifecycle management, and lack of online reconnection mechanism after trajectory breakage.
The UMRTrack target tracking model is adopted, and the observation noise covariance matrix is constructed through an online noise adaptive filtering method. Combined with a multi-source evidence robust correlation strategy and a trajectory segment graph optimization mechanism, adaptive Kalman gain adjustment, multi-source information fusion and trajectory consistency management are realized.
It improves the robustness and accuracy of 3D multi-target tracking, reduces the association error rate, ensures the continuity of the trajectory and the stability of the identification, and adapts to changes in complex traffic scenarios.
Smart Images
Figure CN122415682A_ABST