Emotional State Correlation in Human-Robot Collaborative Manufacturing via Deep CCA
Аннотация
Human-robot collaboration in manufacturing enhances productivity and safety by combining human skills with robotic precision. Monitoring and responding to human emotional states during collaborative tasks is essential for effective interaction and operator well-being. Existing methods often rely on single-modal sensors or heuristic-based adaptations, which fail to capture complex correlations between human emotions and task dynamics, resulting in suboptimal robot responses and reduced efficiency. Deep CCA-Based Adaptive Control (DCAC) addresses these challenges by applying Deep Canonical Correlation Analysis to model nonlinear correlations between multimodal human emotional signals, including facial expressions, speech, and physiological data, and robot actions. This framework enables real-time adaptive robot behavior, allowing adjustments in speed, assistance, and task allocation based on the operator's emotional state. Experimental results in collaborative manufacturing scenarios demonstrate improvements in task efficiency, a reduction in operator stress, and an enhanced overall quality of human-robot interaction.