Research On Obstacle Avoidance System Of Hexapod Walking Firefighting Robot Based On Machine Vision

With the continuous improvement of the concept of national industrial modernization,multi-legged robots have been greatly developed in different fields,especially in the fire rescue is widely used,this kind of robot can be a good substitute for human to complete dangerous tasks.Hexapod robots have been studied by domestic and foreign laboratory scholars because of their excellent mobility and wide application environment,among which parallel hexapod walking robots have become the focus of research because of their overall rigidity,strong bearing capacity,motion stability and other characteristics.Aiming at the complex and changeable fire scene,the autonomous movement of fire-fighting robot is highly required.The perception of the current environment and the path planning autonomous obstacle avoidance implemented after the detection of obstacles are urgent problems to be solved.This topic is based on the science and Technology planning project of Changchun City(application research of non-bionic walking stair climbing fire robot),aiming to solve the problem of fire detection and path planning for avoiding obstacles in the working process of hexapod fire walking robot,and realize the functions of fire warning,obstacle autonomous avoidance and path planning of hexapod fire walking robot.First of all,this paper designs the overall obstacle avoidance scheme of the hexapod walking fire robot,and the way which uses based on machine vision to perceive the current environment information,in which Zhang’s checkerboard and MATLAB toolbox are selected to obtain the internal and external parameters of the selected binocular camera and complete the calibration.Then,Open CV image processing software is used for stereo matching of left and right views to obtain the depth information of the current environment.Secondly,after obtaining the current environment information,this paper selects the YOLOv5 end-to-end target detection algorithm based on three aspects to improve the detection speed and accuracy greatly.Finally,the kinematics analysis of the hexapod walking fire robot is carried out.Three different gaits are designed by simulation and the corresponding obstacle avoidance strategies are implemented according to the different obstacle sizes.At the same time,the path planning algorithm is improved on the basis of the traditional A-star path search algorithm to improve the search efficiency and reduce the search error.After completing the above research content,build the software test environment.With the help of the simulation test tool of MATLAB software,the obstacle avoidance system of the hexapod fire robot designed based on machine vision was simulated and tested in the real environment.The test results show that the scheme design is logical and the system function is normal.The results of ranging experiments at different distances show that the results meet the design needs.In the flame detection test,it is shown that the identification accuracy of the improved algorithm is improved to some extent in terms of speed.In the path planning simulation test,it is shown that the search accuracy of the improved algorithm is improved and the search error is reduced.Finally,in the real environment,the hexapod walking fire robot can safely pass through obstacles and reach the designated place with a high success rate.Therefore,the proposed project has certain feasibility,and provides reference value for the further research of obstacle avoidance technology of hexapod fire-fighting robot based on machine vision.It also makes the application of hexapod walking robot in domestic fire rescue field more intelligent.