障碍物的检测方法、装置、设备及存储介质
By utilizing historical detection records and gradient change category clustering in autonomous driving, the problem of ignoring data coherence between frames is solved, improving the efficiency and real-time performance of obstacle detection, and enhancing the accuracy and flexibility of detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GUANGZHOU WERIDE TECH LTD CO
- Filing Date
- 2022-09-21
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies, when fusing information from different types of detection sources in autonomous driving, neglect the data coherence between frames, resulting in low efficiency and poor real-time performance in obstacle detection.
By acquiring obstacle information of the current frame and projecting it onto a preset network model, reusable and non-reusable grids are determined based on the historical detection records of the network model. Non-reusable grids are processed by category clustering based on gradient changes, and target obstacle information is output.
It improves the efficiency and real-time performance of obstacle detection, reduces computational resource consumption, and enhances the accuracy and flexibility of obstacle detection by reusing reusable grid information.
Smart Images
Figure CN115661783B_ABST