基于单目视觉测距的车辆自动刹车辅助系统设计

Based on Monocular Vision Ranging Automatic Vehicle Brake Assist System Design

摘要:

针对红外测距与超声波测距探测距离与真实距离差异大,刹车控制智能性不强的问题,设计并实现了基于单目视觉测距的车辆自动刹车辅助系统,系统硬件主要由图像采集模块、图像处理模块、以及电子制动模块组成;通过基于单目视觉测距算法实现软件编程,并在Matlab平台上完成测试,实验中车辆以35 km/h驶向障碍物,使在20～70 m的实际距离进行仿真测距,单目摄像头俯仰角测定在88°～90°之间,对前方实时车距进行测量,并通过与汽车电子控制单元之间的数据交换对车辆制动进行辅助控制；实验结果表明:在实测距离和系统介入距离均在70 m以内时,算法相对误差平均在2%左右,车辆停止地点距障碍物在4.5 m左右,说明系统满足车辆智能化辅助制动的实时性要求。

Abstract:

For infrared distance and ultrasonic detection range and the actual distance difference is big, the problem of intelligent brake control is not strong, was designed and implemented based on monocular vision range of vehicle brake assist system, the system hardware is mainly composed of image acquisition module, the embedded microcontroller processor, and the composition of electronic brake module. By ranging algorithm based on monocular vision software programming, and complete the test on the Matlab platform, the experiments of vehicle to 35 km/h to obstacles, make in 20 ~ 70 m for the calculation of the actual distance range, monocular camera pitching Angle measurement between 88°~90°, and for real-time measuring distance ahead, and through the data exchange between auto electronic control unit for auxiliary control of vehicle brake. The experimental results show that the measured distance and system in distance were within 70 m, the arithmetic average relative error is around 2%, the vehicle stop location away from obstacles in 4.5 m or so, this system meet the real-time requirement the vehicle intelligent auxiliary brake.