Abstract:Variable magnetic force adsorption wall climbing robot is a kind of crawling robot with fast and flexible movement mode. However, its adsorption force is difficult to control, and its obstacle climbing stability is poor, which makes it difficult to ensure the smooth crawling of the robot. In order to realize the autonomous obstacle avoidance of the wall climbing robot on the outer surface of large building structures and improve the adsorption tightness between the robot and the moving plane, a variable magnetic force adsorption wall climbing robot control system based on Netvlad neural network is designed. According to the PCB control requirements, the external SRAM device and the sensor module are connected, and the closing of the pneumatic valve is controlled by the electric drive provided by the drive I / O port circuit. The hardware structure design of the variable magnetic force adsorption wall climbing robot control system is completed. Set up the Netvlad neural network system, determine the range of porting parameters by dividing the tasks of the control instruction program, realize porting of the control protocol, and complete the control system design of the variable magnetic force adsorption wall climbing robot based on the Netvlad neural network in combination with the relevant hardware application structure. The experimental results show that under the action of the designed system, the measured distance between the position of the obstacle and the position of the wall climbing robot is not more than 30cm, which can effectively achieve autonomous obstacle avoidance and ensure the close adsorption between the robot and the moving plane.