Fast User Adaptation for Human Motion Prediction in Physical Human–Robot Interaction

Accurate prediction of human movements is required to enhance the efficiency physical human–robot interaction. Behavioral differences across various users are crucial factors that limit motion. Although recent neural network-based modeling methods have improved their accuracy, most did not consider an effective adaptations different users, thereby employing same model parameters for all users. To deal with this insufficiently addressed challenge, we introduce a meta-learning framework facilitate rapid adaptation unseen In study, propose structure and algorithm specialized enable fast user in predicting cooperative situations robots. The proposed comprises shared adaptive parameters, each addressing user's general individual movements. Using only small amount data from user, adjusted user-specific through two-step process: initialization via separate network few gradient steps. Regarding motion dataset has 20 collaborating robotic device, method outperforms existing non-meta-learning baselines