Research On Obstacle Detection Technology Of Amphibious Robot Based On Binocular Vision

Amphibious robot can adapt to complex environments and has excellent motion capabilities,and can replace humans in performing various special tasks such as communication and rescue.In recent years,intelligence has become a research hotspot in robotics and robots are expected to show more autonomy and be able to perform more complex and unconventional tasks.Obstacle detection is a kind of environment perception and the prerequisite for achieving autonomous obstacle avoidance of robots is to correctly detect obstacles.At present,the research scenarios on obstacle avoidance are mainly concentrated on monoxenous platforms and few optimization schemes for amphibious robot application scenarios.This paper uses vision as the main detection method to study amphibious obstacle detection technology.On the one hand,it provides new solutions for vision-based obstacle detection technology;on the other hand,it can promote the development of practical amphibious robots.This paper is aimed at amphibious obstacle detection technology,The main research work is as follows:Investigate the system scheme of obstacle detection system.According to the requirements of the amphibious robot in this paper,combined with the analysis of various obstacle avoidance sensors.,the obstacle avoidance sensor scheme with binocular vision as the main and sonar as the auxiliary is determined.In the condition of ensuring the reliability and speed of the system,the software and hardware platform and the structural design of the binocular vision system have been completed.The binocular camera was calibrated and stereo corrected,and the sonar was calibrated to determine the global coordinate system of the obstacle detection system.In addition,an two-times calibration method is designed to improve the camera calibration accuracy of Open CV.The comparison experiment with the MATLAB calibration method verifies its effectiveness.Investigate the technical route of binocular vision obstacle detection.Aiming at the shortcomings of traditional visual obstacle detection algorithms such as limited adaptability,poor real-time performance,and incomplete obstacle information,an obstacle detection method based on depth map segmentation is designed.To improve the real-time performance,the BM algorithm of Open CV was selected for disparity calculation.A three-dimensional connected domain detection algorithm based on edge separation was developed,which can realize the rapid segmentation of different depth regions in the depth map.The information of the three-dimensional area that is helpful to locate the spatial position of the three-dimensional area and improve the stability of the information is retained,and the calculation method of the relevant information is given.In order to obtain complete obstacle information,the small area filtering algorithm and the area merging algorithm are designed respectively,and the static filtering method of obstacle information is also summarized.Investigate the obstacle detection method based on the fusion of vision and motion sensors.In order to track obstacles in the global coordinate system,an information fusion method based on motion synthesis is designed.According to the principle of spatial continuity,the matching feature information of multiple obstacles that can be used for inter-frame matching is designed.In order to make full use of the existing information,a hierarchical matching method is designed,based on which the inter-frame filtering of obstacles and the calculation of speed information are realized.Investigate the obstacle detection method based on the fusion of vision and sonar.According to the principle of the visual detection algorithm in this paper,a fast median filter plus Gaussian filter combination of underwater image preprocessing method is designed.A general strategy for determining the information fusion method based on the working state of the obstacle avoidance sensor is proposed.Finally,amphibious obstacle detection experiments were carried out in order to comprehensively verify and analyze the performance of the amphibious obstacle detection method.First,the software and hardware platform required for the experiment is built,and the software flow of the obstacle detection algorithm is designed based on the idea of parallel programming.Then the running period and delay time of the obstacle detection algorithm are measured.Finally,a static obstacle detection experiment,a dynamic obstacle detection experiment and an underwater obstacle detection experiment are carried out.The experimental results verify the effectiveness of the method in this paper.