[Retracted] Motion Control System of IoT Intelligent Robot Based on Improved ResNet Model

Abstract

In order to further optimize the precision and efficiency of intelligent robot navigation system control, an IoT intelligent robot motion control system based on the improved ResNet model is proposed. Based on the deep learning method, using the Faster R-CNN target detection architecture and the ResNet50 convolutional neural network, the network is trained according to the characteristics of the operation target of the distribution line maintenance robot system. On this basis, combined with the binocular vision ranging principle, the coordinates of the job target in the camera coordinate system are measured, and the coordinates are converted into the robot base coordinate system through hand-eye calibration, so as to complete the spatial positioning of the job target. The results showed that the errors of the binocular measurement methods adopted by the system are all within 1%. Conclusion. The method can well adapt to the complex background of the operation scene, the change of illumination, and the partial occlusion of the target and can meet the requirements of the distribution line maintenance robot for the measurement and positioning of the target space.