SynergyFF: A single shooting method for simulating crouch gait using muscle synergy feedforward control as a CPG

The central pattern generator (CPG) is a functional network responsible for the creation of motor patterns. It has been suggested that a locomotor CPG consists of a basic rhythm generator and an interneuronal network, essentially implementing muscle synergy. Predictive simulations without tracking experimental data have been used to study gait pathologies, and to explain the causal-effect relationship between muscle deficiencies and abnormal gaits. However, frequent stiff-knee and crouch gaits have proven difficult to simulate when performing single shooting predictive simulations of bilateral plantarflexor weakness.

We trained a musculoskeletal model with 9 degrees of freedom and 14 Hill-type musculotendon actuators to walk by optimizing the parameters of the controller with respect to an objective function. We developed a single shooting predictive simulation method using muscle-synergy feedforward control (called the SynergyFF method) as the CPG and explored its advantages.

When studying gaits associated with plantarflexor weakness, our SynergyFF model was able to simulate crouch gait that could not be simulated by the classical reflex-based model. Our model was also able to simulate calcaneal and stiff-knee gaits. The success of our simulations was primarily driven by more accurate modeling of the knee flexion moment.

The CPG based on synergy hypothesis was able to produce realistic simulations of abnormal gaits. We attribute this result to the critical role played by muscle synergy in our framework for the generation of crouch gait. More generally, our simulations provide support for the existence of CPG and the hypothesis of muscle synergy.