Author(s)

Hu Jiahao

Corresponding Author

Hu Jiahao

Abstract

With intelligent manufacturing technology entering a new era, the bionic mechanical arm has also developed quickly. Centered on innovation-led development ideas, researchers need to propose dynamic obstacle avoidance methods that meet the needs of industrial automation. Based on the dynamic evolution of deep learning, we construct a theoretical analysis framework for real-time visual processing algorithms according to the inherent logic of visual processing and explain the dynamic obstacle avoidance mechanism related to YOLOv 5's manipulator and the obstacle avoidance strategy generated by the real-time visual feedback loop mechanism. In addition, the potential to achieve more efficient and safer goals in intelligent manufacturing is explored in terms of the practical derivation of deep learning algorithms and mechanical arm control. In this research, by constructing a dynamic obstacle avoidance model of a bionic mechanical arm based on YOLOv5, the influence of various visual processing parameters, neural network structure, and error level on the avoidance results is analyzed, which provides theoretical support and technical reference for the real-time visual processing algorithm of a bionic mechanical arm.

Keywords

YOLOv5; Bionic mechanical arm; Dynamic obstacle avoidance; Real-time visual processing