Automated Curriculum Design for High-dimensional Human Motor Learning: Experimental Data

Designing effective practice schedules for high-dimensional motor learning tasks remains a challenge, especially when skill states are unobservable and task performance may not reflect true learning. We propose an automated curriculum design framework that combines a human motor learning model and personalized real-time skill estimation with Stochastic Nonlinear Model Predictive Control (SNMPC) for task sequencing in de-novo (novel) motor learning paradigms. Compared to a random curriculum and a heuristic performance-based curriculum, our SNMPC approach accelerates skill acquisition and improves task performance. Simulation and human-subject studies using a hand exoskeleton demonstrate significant gains in learning efficiency, highlighting the potential of model-based, individualized curricula for motor rehabilitation and complex skill training.