This paper proposes an energy-based training assistance controller for a unicycle riding simulator inspired by Human Adaptive Mechatronics (HAM). We focus on sagittal plane (pitch) balance for beginners and derive a simplified longitudinal plane unicycle model, where pedaling is represented as an action–reaction torque between the wheel and the rider–saddle body. After time normalization, the saddle dynamics is expressed in a form suitable for energy analysis. Using the natural Hamiltonian of the uncontrolled system, we design a minimum-intervention pumping–damping controller that modifies the energy flow only when necessary. The assistance is smoothly activated outside a training core region defined by a saddle-angle bound: a damping term suppresses excessive motion, and a pumping term prevents trapping in a tilted posture when the energy becomes too small. The proposed framework offers physically interpretable, localized assistance while preserving the natural unicycle dynamics required for skill learning.
Kubota et al. (Mon,) studied this question.