A Hybrid Controller Integrating Finite-State Impedance and Electromyography-Driven Musculoskeletal Model for Robotic Active Ankle Prostheses: A Case Study

Abstract

Finding a balance between adaptability for unexpected or non-standard movements and repeatability for cyclic standard movements is a known control challenge for active lower-limb prostheses. This study integrates a volitional electromyography (EMG)-driven musculoskeletal model controller with a finite-state machine impedance controller. During the stance phase, the hybrid controller is activated, and during the swing phase, it functions as a position controller. To provide a seamless transition to the swing phase of the gait, the controller first runs in hybrid mode until it validates that the foot is entirely off the ground. The Gastrocnemius and Tibialis Anterior muscles were modeled using a Hill-type muscle model to function around an ankle joint. The system uses input from ankle sensors and EMG data from antagonist muscle pairs to activate the muscle models. In addition, muscle parameters within the model are optimized using a surrogate-based optimization technique to improve the neuromuscular model controller’s performance and responsiveness. A transtibial amputation participant assessed the hybrid controller online for various activities, including ambulation on level ground, stairs, and ramps, using EMG signals from the residual limb as input. Combined with the impedance controller, the EMG control showed exceptional efficiency and repeatability, particularly for stair ascent, stair descent, and ramp descent. Moreover, according to user feedback, the user felt more involved and capable of using their muscles to control the motion, which improved their overall sensation of stability and control during these activities. The hybrid controller functioned more like an impedance controller during ramp ascents and walking on level terrain.

Keywords

• Direct myoelectric control
• electromyography (EMG)
• model-based design (MBD)
• neuromuscular model (NM)
• prosthesis control
• surrogate-based optimization (SBO)