车辆控制方法、装置、设备及存储介质
By segmenting roads and building a perception model library, and dynamically calling appropriate perception models, the problem of insufficient perception of a single perception model in complex environments is solved, thus achieving accurate perception and safe driving.
CN116985840BActive Publication Date: 2026-07-17TENCENT CLOUD COMPUTING (BEIJING) CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TENCENT CLOUD COMPUTING (BEIJING) CO LTD
- Filing Date
- 2022-09-27
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In existing technologies, single perception models struggle to cope with the diverse driving situations and environmental conditions in the real world, resulting in insufficient perception accuracy and impacting driving safety.
Method used
By segmenting the road, a perception model library is constructed for different environmental conditions, and the most suitable perception model is dynamically called to perform perception tasks during driving.
Benefits of technology
It enables accurate perception of all possible situations that may occur during vehicle operation, improves driving safety and perception effectiveness, reduces perception difficulty, and promotes the implementation and promotion of autonomous driving.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116985840B_ABST
Abstract
本申请公开了一种车辆控制方法、装置、设备及存储介质。本申请实施例可应用于自动驾驶领域或辅助驾驶领域。本申请实施例提供的车辆控制方法实现了切片式感知。详细来讲，对于任意一个道路段，会构建该道路段在不同环境条件下的道路图像集，进而根据该道路段在不同环境条件下的道路图像集，训练该道路段在不同环境条件下的感知模型，进而形成感知模型库。这样，在行驶过程中，车辆便可根据行驶途中的道路段信息和处于相应道路段时的环境条件，从感知模型库动态调用最适配的感知模型进行感知。由于针对不同道路段、不同环境条件训练了不同的感知模型，因此能够对车辆行驶过程中所有可能出现的状况进行精准感知，进而确保了驾驶安全性。
Need to check novelty before this filing date? Find Prior Art