Research On Carpet Recognition And Path Planning Of Sweeping Robot

At present,in the actual cleaning process of the sweeping robot,after the sweeping machine enters the carpet,it will be trapped in it due to the chassis rollers and the dust suction structure,and it cannot clean up for the user.In addition,the sweeping robot is not allowed to enter the carpet in the wet mopping cleaning mode.Therefore,in view of the problem of the sweeper touching the carpet during cleaning,this paper uses digital image processing to extract texture features from the carpet and floor pictures in the surrounding environment of the sweeping robot,and then uses the image classification algorithm to recognize the carpet,and use the carpet recognition result as obstacles that are introduced into the sweeping robot environment map,and finally point-to-point path planning is carried out.The main contents of the thesis are as follows:(1)Apply the texture feature extraction algorithm and the image classification algorithm to the sweeping robot to achieve the purpose of carpet recognition.The gray level co-occurrence matrix is used for texture feature extraction,and the support vector machine is used to classify the carpet and the floor,and the hyperplane equation that distinguishes the carpet and the floor is obtained.The window direction,phase element spacing and characteristic parameters of the gray-level co-occurrence matrix are analyzed in detail,and the local binary mode extraction and comparison experiment and the comparison experiment between the support vector machine kernel function and the texture feature dimension are carried out.Finally,choose the best carpet recognition solution.(2)Designed the carpet recognition system software of the sweeping robot.Software developers can obtain the final source code for carpet recognition through simple configuration,and automatically check whether the accuracy of the source code carpet recognition meets the requirements.The carpet recognition software of the sweeping robot improves the work efficiency of software programmers and saves a lot of precious time.(3)An improved trajectory planning algorithm for PRM is proposed.By taking the boundary points of obstacles as the definite sampling points,and establishing the best feasible area for the free space of the raster map,the dispersion of random sampling points of the traditional PRM algorithm is reduced,and the path search is clear,thereby improving the algorithm’s time And space utilization.Simulation and physical experiments are carried out to fully verify the feasibility of the algorithm.Finally,the carpet recognition and path planning algorithm verification of the sweeping robot is carried out.The environment map is created by combining the gray level co-occurrence matrix and the support vector machine algorithm,and the improved PRM algorithm is used for point-to-point path planning.This improved algorithm completes the goal of not touching the carpet in the cleaning mode and autonomous recharging mode.