An Efficient Human Activity Recognition Framework Based on Graph Convolutional Network for Human-Robot Collaboration

In this paper, a collaborative human-robot object processing framework based on skeleton data and graph convolutional human action recognition is proposed, and network architecture search and adaptive graph mechanisms are introduced. By recognizing the pose that the human carries the object, it is passed to the collaborative robot to make a matching object pickup action. The method is driven by the human skeleton data captured from RGB images, recognizes human actions by adaptive learning graph structure, and builds a graph convolution model and training in the collected dataset. Human-robot collaboration experiments are conducted in a laboratory environment, and the experimental results show that the method achieves high accuracy and real-time performance, and effectively accomplishes the human-robot collaboration task.

I. Introduction

Currently, robots are extensively employed in diverse production settings to mitigate workforce shortages. However, conventional industrial robots are predominantly programmed in advance to carry out certain duties and lack the ability to adapt their actions based on real-time environmental assessments. Furthermore, these robots pose challenges in terms of operation, cost, and the need for isolation from human presence throughout their functioning. The introduction of collaborative robots has mitigated the limitations of conventional robots, which are primarily engineered to aid people in doing specified activities collectively [1]. These robots may easily be adjusted to various settings and utilized for diverse purposes through straightforward programming [2]. Consequently, academics have shown significant interest in enhancing the agility and precision of collaborative robots for human-robot collaboration (HRC) [3]–[6].