| In consideration of many problems,such as long manual inspection time in the exhibition hall,shallow inspection level,and insufficient handling of abnormal exhibits in time,the exhibition hall inspection robot system is designed.During the inspection process of the inspection robot,the reliability and stability of the remote control communication system determine the inspection quality.At the same time,the obstacle avoidance and efficiency problems of path planning in the autonomous inspection process also need to be solved urgently.This paper takes the inspection robot system as the research object,and conducts research and design on its remote control communication system and autonomous driving system.The inspection robot system of the exhibition hall is composed of three units: the inspection robot body,the communication control unit and the monitoring room control station.The inspection robot communication control unit is designed,and the key issues such as the data transmission module and the image transmission module are studied to complete the image transmission,data transmission and other functional indicators.At the same time,the technical indicators of the control station in the monitoring room and the parameters of remote control equipment are designed to meet the requirements of various indicators of the inspection robot system.Based on the inspection robot body and the depth camera sensor,the inspection robot motion model,sensor model and map model are established.The global path planning A~* algorithm of the autonomous driving system is studied.The time factor is introduced to optimize the A~* algorithm evaluation function.The estimated cost weight is increased,In order to speed up the search for the optimal node and significantly improve the efficiency of path planning.In view of the local dynamic obstacle avoidance problem in the inspection process of the inspection robot,the local path planning Double Q-learning algorithm is researched.The Double BP Q-learning algorithm is proposed,which uses two BP neural network fitting value functions of the same structure to speed up the model learning efficiency.Add preferential experience playback and parameter transfer mechanisms to improve generalization capabilities.Set up sparse discrete,dense discrete,and special obstacle environment simulation comparison experiments to verify the feasibility of the improved algorithm model.On the basis of the research and design of the inspection robot remote control system and autonomous driving system,the actual robot inspection application of the exhibition hall is carried out,and the human-computer interaction interface of the remote control system software is determined to be simple and friendly.The exhibition hall is used as the actual environment to verify the path of the autonomous driving system Planning is scientific. |
