The invention discloses a Yo-v9 multi-mode road adhesion coefficient estimation algorithm fused with dynamics, and belongs to the technical field of automotive electronics. Comprising the following steps: capturing a front road surface in real time through a vehicle-mounted infrared and visible light camera, fusing through a dynamic fusion algorithm, and sending to an image processing algorithm; carrying out feature extraction on the processed algorithm by adopting an improved Yo-v9 algorithm to obtain useful information of the road surface; a global adaptive attention mechanism is adopted to improve the feature extraction network, and an adhesion interval of a front road surface is obtained through table look-up after target detection and classification; tire dynamics is adopted to estimate the adhesive force of the current road surface, and the stability of the vehicle is controlled. According to the invention, the problem that the vehicle stabilization system cannot well estimate the current road adhesion coefficient and cannot stably control the vehicle is solved. The stability of vehicles under road surfaces made of rain, snow, mud and various materials can be improved, and the road safety is improved, so that the personal safety and the social stability are guaranteed.
本发明公开了一种融合动力学的Yolo‑v9多模态路面附着系数估计算法,属于汽车电子技术领域。包括以下步骤:通过车载红外与可见光相机将前方路面实时捕获,通过动态融合算法进行融合后送入图像处理算法;采用改进的Yolo‑v9算法对处理的算法进行特征提取,获取到路面有用信息;采用全局自适应注意力机制对特征提取网络进行改进,通过目标检测分类后查表得到前方路面的附着力区间;采用轮胎动力学对当下路面进行附着力估计,对车辆稳定性进行控制。本发明解决了车辆稳定系统不能很好地对当前路面附着系数进行估计,不能够稳定地控制车辆的问题。有助于改进雨雪、泥泞以及各种材质路面下的车辆稳定性,提升了道路安全,从而保障了人身安全以及社会稳定性。
Yo-v9 multi-mode road adhesion coefficient estimation algorithm fused with dynamics
融合动力学的Yolo-v9多模态路面附着系数估计算法
2024-08-06
Patent
Electronic Resource
Chinese
Vision and dynamics fused road adhesion coefficient estimation method
European Patent Office | 2020
|Image and vehicle dynamics fused road adhesion coefficient estimation method
European Patent Office | 2024
|Road adhesion coefficient interactive estimation method based on vision and dynamics
European Patent Office | 2020
|British Library Conference Proceedings | 2022
|