Research On AGV Obstacle Detection And Path Planning Based On Visio

With the wide application of Automated Guided Vehicles(AGVs)in industry,logistics and e-commerce,the technology of obstacle detection and path planning of AGVS has been put forward higher requirements.In this paper,for the obstacle detection problem that AGV faces in the process of work,the machine vision technology is used to carry out in-depth research on the detection algorithm;Aiming at the path planning problem of AGV,the traditional DWA algorithm was optimized to improve its efficiency.Firstly,the obstacle detection method based on corner domain features is introduced.Based on the corner detection algorithm of FAST,the concept of image pyramid is introduced,and an improved FAST feature point detection algorithm based on image pyramid is proposed.Simulation experiments show that the recognition rate of different obstacles such as AGV and tray reaches more than 80%.Second,a variety of neural network models are simulated and analyzed.DeepLabV3+network model is selected to segment the obstacle body in the storage environment,so as to extract ROI.An obstacle detection algorithm based on the combination of DeepLabV3+ and the improved FAST corner detection method is proposed.The simulation results show that the combined algorithm has a high recognition accuracy of 95.48% for AGV and tray,which is suitable for AGV application in practical engineering.Thirdly,aiming at the problem of path planning when AGV encounters obstacles in the process of work,a DWA algorithm based on improved evaluation function is proposed,which overcomes the shortcoming of traditional DWA algorithm that locks up in the area with "L and C" shaped obstacles,and enables AGV to re-plan the path after encountering obstacles.The algorithms mentioned above are used for simulation verification,and the experimental research is carried out in the AGV warehouse as the experimental background.The final results show that the combined algorithm based on DeepLabV3+ and improved FAST corner detection can accurately detect the obstacles within the specified range under the condition of meeting the AGV working speed.The DWA algorithm based on the improved evaluation function can re-regulate the optimal path after encountering obstacles,effectively avoid the "L" shape and "C" shape obstacle area,and successfully complete the task of picking up goods and transporting goods.