Research On The Control System Of Wall-Climbing Robot Based On ARM And Linux

Based on ARM and Linux platform,a robot control system for the ship-cleaning wall-climbing robot is developed in this paper.According to the requirements of the control system,the distributed control scheme of the host and slave computer of the wall-climbing robot is determined.The host computer and the slave computer use ARM9 chip as the control core.The slave computer peripheral circuit also includes: Zigbee communication module,analog camera,remote map transmission system transmitter,high pressure water and vacuum detection circuit,high pressure water vacuum valve and air motor speed control circuit,MMA7455 three axis acceleration sensor circuit.The host computer also includes LCD display circuit,rocker circuit,remote image transmission receiver circuit,Zigbee communication module circuit.The wall-climbing robot realizes the automatic operation by image recognition.The image of working condition collected firstly by the control system of wall-climbing robot and grayed and binarized to realize the extraction of the tracking guide line.The automatic operation of the climbing wall robot can be completed by tracking the tracking line.In order to achieve the stability,accuracy and fastness of the climbing robot tracking track,the Q-learning algorithm is introduced to the control system of wall-climbing robot.The algorithm take the position of tracking line in the image as the state value,and the diffential of the air pressure between the left and right air motors as the action.The introduce of the Q-learning algorithm realized the reinforcement learning tracking and improved the problem that classical PID algorithm cannot choose the optimal policy for different environment.Thus it proved the accuracy of the wall-climbing robot tracking the tracking line.The control system is based on the Linux operating system.The Linux operating system has the advantages of high reliability,high stability,and can let the processes run concurrently.The using of the Linux system can solve the problem of system scheduling and real-time display of system working conditions.The hardware can only manipulated in kernel level,so kenel driver program should be developed as well as application layer program.The testings of wireless data-transmission reliability,the image data accuracy and stability and the control effect of control system show that the wireless data-transmission of the control system can make sure its bit error rate is-610 with a confidence level of 99%,the digital image signals are stable after image acquisition and decoding and the image displays without delay,the control system also ensure the control of the robot is in stable state.When the wall-climbing robot runs automatically,the reinforcement learning tracking algorithm can converge in testing area and traverse all the area.