Research On Obstacle Recognition And Path Planning Of Mine Rescue Robot

After a safety accident occurs in the environment of a coal mine,a mine rescue robot enters the disaster scene for rescue activities,ensuring the safety of people's lives and property.Obstacle recognition and path planning are the key technologies for mine rescue robots,and in-depth research on them is an important prerequisite for the robots to successfully complete rescue tasks.Therefore,this thesis mainly carries out the following research around these two key technologies:(1)Calibration about the stereo vision system of the robot.In this thesis,in order to solve the internal parameters and relative position relationship of the left and right cameras,the calibration experiments of monocular and binocular vision systems are performed by Zhang Zhengyou calibration method and Matlab toolbox,and the projection error analysis is performed on the calibration results.The Fusiello algorithm is used to perform binocular epipolar correction on the images to lay the foundation for stereo matching of the images.(2)Research on obstacle recognition of mine rescue robot.Firstly,this thesis analyzes the types and characteristics of obstacles in the mine environment.Secondly,the denoising algorithm based on the improved wavelet transform is used to filter and reduce the noise of the infrared image of the mine,and the image is deblurred by the Wiener filtering algorithm.Then,the Canny operator is used to detect the image edge features in a short period of time,preparing for the image recognition research of obstacles.Then,a Graph Cut algorithm incorporating an improved watershed segmentation based on markers is used to extract two-dimensional edge and area features of obstacles.Finally,stereo binocular matching processing is performed on the collected binocular images to obtain the spatial information of obstacles,which provides data support for subsequent path planning tasks of the mine rescue robot.(3)Research on robot path planning based on improved ant colony optimization.This algorithm is mainly suitable for point-to-point path planning in a global static environment.Firstly,through a large number of simulation experiments,the optimal parameter combination of ant colony optimization for path planning problems is obtained.Secondly,the initial pheromone is differentiated to accelerate the convergence speed of ant colony optimization.The path heuristic function is substituted for the estimated cost function to improve the search efficiency of ant colony optimization.The pheromone is updated,and the iterative threshold parameter is introduced to adjust the pheromone volatility coefficient,so as to prevent the ant colony optimization from falling into the local optimal solution.Finally,the optimal path obtained by the improved ant colony optimization is smoothed.The comparison of MATLAB simulation experiments proves the effectiveness and superiority of the improved ant colony optimization.(4)Research on path planning of mine rescue robot based on ant colony optimization with potential field.In this thesis,the ant colony optimization with potential field is applied to the complex post-disaster mine environmental path planning to complete the rescue task in the designated area.The thesis combines artificial potential field method and ant colony optimization and uses ant colony search mechanism for global path planning.It uses the positive feedback and global optimization characteristics of ant colony optimization and the principle of avoiding obstacles and feedback control by artificial potential field method.The ant colony optimization with potential field solves the local minimum problem of artificial potential field method and the blind search problem of ant colony optimization.Finally,MATLAB is used to compare the simulation experiments on structured and unstructured mine environment raster maps,which proves the feasibility of this algorithm.The thesis has 87 pictures,10 tables,and 97 references.