Mechanical arm six-degree-of-freedom real-time grabbing method based on deep reinforcement learning

Abstract

The invention relates to a mechanical arm six-degree-of-freedom real-time grabbing method based on deep reinforcement learning. The mechanical arm six-degree-of-freedom real-time grabbing method based on deep reinforcement learning comprises the following steps of 1, acquiring image information of an object on a grabbing operation table through a binocular camera; 2, performing target detection training on the image by using a YOLOv5 pruning network model; 3, establishing a reinforcement learning network model; 4, completing mechanical arm grabbing movement through forward and inverse kinematics of a robot; and 5, carrying out reinforcement learning model training, so that a mechanical arm completes the grabbing action. The invention overcomes the defects of the prior art, provides a real-time object detection system which is easy to implement and high in applicability and is based on the YOLOv5 pruning network and the Policy Gradient reinforcement learning method, and the system can realize rapid and real-time target detection and complete the grabbing action while ensuring high precision.

Description

technical field [0001] The invention belongs to a real-time grasping object technology of a manipulator based on deep reinforcement learning, and specifically relates to YOLOv5 pruning network, Kinevt forward and reverse kinematics, CoppeliaSim Edu simulation software and Policy Gradient reinforcement learning strategy. Background technique [0002] Grasping is a fundamental and important problem in robotics. Despite its importance, the solutions to this problem have been unsatisfactory. However, with the rapid development of deep learning and reinforcement learning in recent years, it has provided many feasible ideas for the intelligent grasping method of the manipulator. Real-time object detection technology is a research hotspot in the field of computer vision in recent years. This technology includes the design of lightweight object detection networks, the production of target data sets, and the research on model deployment carriers. Among them, one of the most obvious ...

AI Technical Summary

Problems solved by technology

However, due to potential failures such as sensor storage and transmission, the 3D point cloud depth data input to the network is not stable compared with traditional 2D RGB image data.

Images