Normative Conflict Resolution in Gesture-Based Human-Robot Collaboration: A Personalized Recognition Method with Meta-Learning

Human-robot collaboration (HRC), particularly in complex assembly, is often hindered by a “human-machine normative conflict,” whereby conventional “one-size-fits-all” gesture recognition systems prioritize system accuracy but limit operator autonomy. To mitigate this conflict, this study develops a personalized gesture recognition framework that shifts the interaction paradigm from machine-prescribed to human-defined. Improving an Augmented Reality platform for habitual gesture acquisition and a novel GCN-ML-SVM model for intention inference, the framework facilitates rapid adaptation to both unseen users or gestures. The model integrates Graph Convolutional Networks for spatial feature extraction and a Model-Agnostic Meta-Learning strategy for few-shot fine-tuning to ensure robust performance. Systematic experiments demonstrate that the framework achieves a high recognition accuracy while granting high self-defined flexibility within the supported static-gesture vocabulary. Furthermore, a quantitative evaluation reveals a significant reduction in cognitive load. This research provides a replicable methodology for balancing operational precision with interactional naturalness in human-centric industrial environments.