A Human Digital Twin Based Framework for Human–Robot Hand-Over Task Intention Recognition

Industry 5.0, the upcoming industrial revolution, places a strong emphasis on human-centric smart manufacturing, redefining the state and role of humans in the process of human–robot collaboration (HRC). The human–robot hand-over task plays a crucial role of HRC, and finding ways to enable robots to understand human handover intentions is an urgent problem to be solved. In this study, a human digital twin-based framework for human–robot hand-over task intention recognition is proposed to enhance the execution efficiency of human–robot hand-over tasks. This framework aims to foster a seamless integration and cooperation between humans and robots, facilitating deep human–robot fusion. Then, considering the multi-scale characteristics and temporal correlation of human intention information, a feature extractor based on multi-scale convolutional neural network and bidirectional long-short-term memory (MSCNN-BiLSTM) is devised to improve the cognitive and collaboration abilities of robots. A series of optimization experiments are conducted to enhance the performance of the MSCNN-BiLSTM model. The superiority of the proposed framework is demonstrated by comparing with traditional human–robot hand-over intention recognition algorithms, which provides an approach to achieve better human–robot fusion and more efficient execution of hand-over tasks.