Research On Coverage Path Planning For Robots Based On Machine Vision

Mobile robots are the mechanical devices that are able to move independently,and playan important role in the modern production and life.In the field of mobile robot research,coverage path planning is a very hot issue.Coverage path planning is such an issue that the robot must visit all reachable regions in the designated work area.The typical coverage path planning algorithms generally have problems of long distances of paths and frequent turnings,which leads to the low coverage rate of the robot and the high loss of time and energy.To solve the above problems,the object distance detection technique is adopted in this thesis,with the coverage method of back and forth motion,and new ways of edge detection and the exhausted walk for the robot are proposed,with the advantages of high coverage rate,low repetition rate and short time.The main works are as follows:(1)Aiming at the problem of discontinuity in image edge detection,the Canny edge detection operator is improved.Firstly,the color image is transformed into the HSI color space,and then the Canny edge detection is performed on the H,S and I color spaces respectively,with three detection results.The three results are superimposed to obtain the valuation of each pixel,and then the technique which uses a high value to raise the value of its surrounding underestimated pixels is used to re-evaluate pixels in the image.Finally,each pixel of the image is judged to obtain the highest valuation pixels for the edge,and the other pixels are non-edge.The results with the above method are better than the typical operator.(2)In the current coverage path planning algorithm,the robot needs to learn the regional environment,and then plan the path scheme.This results in low coverage rate,high repetition rate and high energy consumption.This thesis defines two states of the robot: the back and forth motion state and the guidance to a next target state.After encountering an obstacle to end the back and forth motion,the robot is guided to an unvisited region with the shortest path.After arriving at the new target,the robot switches to the back and forth motion state.While covering the region,the robot is able to congnize the environment with this method,improving its efficiency.(3)Aiming at the shortcoming of the fixed target in the inverse distance transform,an improvement is made in this thesis.The distance value is accumulated from the starting point,so as to find the target point dynamically and find the most efficient point.The experimental results show that the proposed algorithm has higher coverage rate and efficiency than the existing algorithms.