Context-aware cognitive assistive assembly system based on online human action recognition
Huachang Liang, Hao Ran Zheng, Yuqian Lu · 2024
In response to the demand for mass personalised manufacturing in Industry 5.0, we propose a novel human-centric cognitive assistive assembly system. This system is engineered to enhance assembly efficiency by offering workers contextual instructions. These instructions are dynamically generated based on real-time analysis of worker’s actions and task progress, effectively reducing cognitive load and improving workflow efficiency. Leveraging and advancing cutting-edge computer vision and machine learning technologies, our system accurately identifies human activities during assembly. The recognised action integrates with a reference task graph to provide feedback, including step-by-step assembly guidance, future action suggestions, and progress updates via intuitive augmented reality overlays directly on the workbench. Our results demonstrate the capability of our system in assembly activity recognition, thereby ensuring precise guidance tailored to workers’ performance. The proposed assistive assembly system enables a human-centric assembly scenario which reduces worker stress through straightforward, intuitive support and improves assembly efficiency through adaptive instructions. This research contributes to the evolving field of personalised manufacturing, highlighting the synergy between human skills and digital augmentation in advanced manufacturing environments.