Development of Control Algorithms for an Adaptive Running Platform for a Musculoskeletal Rehabilitation System.

An essential component of modern musculoskeletal rehabilitation systems is treadmills of various sizes, the control of which may rely either on manual adjustment of treadmill speed, fixed for the entire training session, or on automatic regulation based on analysis of the user's movements and velocity. The aim of this study was to experimentally compare the control functions of an adaptive treadmill designed for musculoskeletal rehabilitation and to assess the influence of the hardware configuration and tracking systems on user stability and the smoothness of transient processes. Two running platforms (of different lengths, one equipped with handrails and one without), two tracking systems (virtual reality trackers and a computer vision system using the MediaPipe Pose model), and three control functions-linear, nonlinear, and proportional-integral-derivative (PID)-were investigated. A set of metrics with both metrical and physiological interpretability was proposed (including positional stability, duration and amplitude of transient processes in position and velocity, subjective assessment, and others), all integrated into a single quality control criterion. This study presents extensive experimental research comparing various designs of adaptive running platforms and tracking systems, exploring the relationships between the available working area length and user comfort, and determining the optimal parameters for the selected control functions. The optimal control function was identified as the linear law for the tracking system based on virtual reality trackers and the PID function for the computer-vision-based tracking system. The conducted experiments made it possible to formulate recommendations regarding the minimum permissible working area length of treadmill platforms and the selection of tracking systems and control functions for musculoskeletal rehabilitation systems. The obtained results are of practical relevance for developing adaptive rehabilitation simulators and creating control algorithms that ensure smooth and stable treadmill motion, thereby enhancing user comfort, efficiency, and safety during musculoskeletal rehabilitation exercises.