Object Recognition and Grasping Detection Method of Construction Machinery Robot Relying on Deep Learning

Abstract

At present, robots are widely used in the field of machinery manufacturing, which improves the production speed on the basis of ensuring the safety of workers. As a result, the traditional mode of enterprises based on manual operation has begun to be transformed into industrial robots as the main body. Construction machinery robots need to capture information about the surrounding environment in production operations. Traditional object recognition methods cannot adapt to complex working environments. Therefore, how to effectively identify target objects and successfully grasp objects has become a challenge for robots. In addition, the grasping detection(GD) method required by the robot to complete the task also relies on the known information of the target object and cannot effectively deal with the complex and changeable unknown environment. To this end, this paper designs a robot GD system based on deep learning, constructs a GD model and system framework through a convolutional neural network(CNN), and introduces a multi-target object grasping recognition algorithm to improve the grasping accuracy of the robot. The simulation experiment of the GD system proves that the accuracy rate of the system successfully grasping objects is over 95%.