Design and Implementation of Vehicle Safety Robot Based on Computer Vision

Abstract

This paper designs and implements a vehicle safety robot based on machine vision. When a car breaks down on the highway, the robot can automatically drive to the designated location, place a warning triangle, monitor the coming vehicles, and give the driver an alarm when necessary. The robot can automatically drive back when ordered. The robot uses Lane Detection Algorithm based on OpenCV to detect road line and adjust its driving gesture dynamically, an Optical Flow Alarming Algorithm is implemented to monitor the coming vehicle, and the communication is implemented through a 2.4 GHz module. Through experiments, we measure the TPR (true positive rate) of the Lane Detection Algorithm, we also test the end-to-end computing time of the system and test the stability of the communication of the system. The analysis shows that the design and implementation of the system can meet the requirement of the highway scenario.