Real-Time Haptic Testbed for Path-Following Control in Rehabilitation: System and Performance Characterization

Abstract

Developing path-following haptic guidance for body weight support (BWS) rehabilitation systems requires validated, risk-free real-time infrastructure before implementing advanced control strategies. This paper presents the architecture and preliminary validation of a scaled haptic testbed integrating a Beckhoff CX2043 real-time computer with a Novint Falcon haptic device. The system employs a distributed architecture: MATLAB/Simulink controllers compiled to TwinCAT PLC code executed on the embedded real-time platform, communicating via UDP with a multi-threaded C++ application interfacing the haptic device. We characterize the system's timing performance, achieving 1?kHz control loop execution on the real-time controller with current end-to-end latencies of approximately 20?ms. Preliminary tests with sit-to-stand trajectories derived from motion capture data confirm functional path-following capability. This work establishes the foundational infrastructure for subsequent development and comparison of assist-as-needed control strategies on a safe test bed that can be translated to “the FLOAT” BWS system (Reha-Stim Medtec AG, Schlieren, Switzerland).