Design And Implementation Of Binocular Imaging Obstacle Detection System Based On Embedded Platform

As a key technology integrating software and hardware in the indoor robot system,obstacle detection technology has a decisive influence on whether the robot can complete the established functions.The complex and varied indoor environment determines that such indoor robots can accurately detect and avoid obstacles.This is also an important reason why such indoor mobile robots are still costly and difficult to popularize.The detection of obstacles through non-contact technology is the current mainstream direction,mainly including ultrasonic,infrared detection and laser ranging.Ultrasonic and infrared technologies have poor performance for objects with small reflected signal area,and the laser scanning area is large.High precision,but due to the high cost,it is difficult to popularize at one time.Therefore,there is a need for a detection system that is low cost and meets the requirements of the actual work scene.To this end,this thesis designs and implements an obstacle detection system based on visual imaging technology.The main work and innovations of this thesis are:Based on the analysis of the current mainstream indoor robot obstacle detection technology advantages and disadvantages,the system realized in this thesis is based on binocular vision technology,combined with the low cost and high accuracy of the overall requirements,using the "Zhang Zhengyou calibration method" to complete the embedded based The preprocessing operation of the image distortion of the dual camera acquires high-precision parameters of the camera required for subsequent image processing.This thesis selects the Raspberry Pi platform based on low cost,high integration,complete functions and easy operation to complete the design.The article completes the preprocessing of the original image captured by the camera by using the distortion elimination algorithm and the bilateral filtering method.In this thesis,based on the characteristics of the application scenario,the two matching methods are compared,and the actual test results are used to select a more suitable method for the scenario proposed in this thesis.Through continuous testing and parameter adjustment,the final results show that the proposed system can extract the distance information of obstacles in the range of 250mm?500mm with over 95%accuracy.Based on the test results,this thesis proposes a method for detecting obstacles in subregions based on the actual scene and the particularity of the indoor mobile robot platform.The method further develops the three-dimensional coordinate information of the feature points obtained by matching.The processing is used to assist the robot to detect and avoid obstacles.The test results of the actual scene verify the effectiveness of the method.In order to solve the challenges and problems of obstacle detection in indoor robots,under the condition of ensuring low cost and high accuracy,this paper completed the design and implementation of binocular imaging obstacle detection system based on embedded platform,which realized within a certain range.The accurate detection function of the obstacle distance information acquires the effective data required for the auxiliary robot to complete the obstacle avoidance function.