自动驾驶模型训练方法、轨迹确定方法及自动驾驶车辆
By evaluating the rationality of candidate trajectories and modulating the loss value to train the autonomous driving model, the problem of conservative driving schemes in complex environments is solved, and more reasonable and flexible driving trajectory selection is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING BAIDU NETCOM SCI & TECH CO LTD
- Filing Date
- 2024-03-27
- Publication Date
- 2026-07-17
AI Technical Summary
When faced with complex real-world situations, autonomous driving models often output conservative driving plans, failing to effectively handle emergencies or special circumstances, resulting in low rationality of driving trajectories.
By evaluating the rationality of candidate trajectories, the loss value is modulated to train the autonomous driving model, enabling it to output diverse driving schemes. Human feedback is then incorporated to optimize the evaluation model and improve the rationality of trajectory selection.
It improves the rationality of autonomous driving models' driving trajectories in complex environments, ensures flexible responses in special situations, and outputs multi-mode candidate trajectories to enhance the diversity and rationality of decision-making.
Smart Images
Figure CN118171105B_ABST