The Road Detection Technology of Vision Navigation for Picking Robot

摘要

Vision navigation technology of the autonomous walking robot has been conducted in the research. Based on the characteristics of picking robot working environment, the visual system for the independent walking car is designed, which applies monocular camera to obtain images of marking lines on the surface of road for pre processing. Using the image gray contrast constraints determines the information of the effective area on the surface of road. Then the Hough linear detection combined with constraint will be used for the recognition and detection of the indication line on the surface, and the information of visual examination path will be fed back to the car control system as following. Through the ultrasonic sensors on the head of the walking car, the walking car can detect the obstacle at the same time. According to the real-time feedback data of ultrasonic sensor, fuzzy propulsion and calculation determine the information of the obstacles on the road surface ahead of time in order to achieve the function of real-time obstacle avoidance. The result of visual navigation test shows that maximum deviation of the car walking straight line based on visual navigation system is 1.5cm and the maximum deviation of curve walking is 6cm. The success rate of obstacle avoidance is 96.67%, which can meet the requirement of the vision navigation of picking robot and provide technical support for the autonomous navigation and intelligent operation of picking robot.