A novel modular architecture for a neural controller for predictive simulations of stand-to-walk motions

Neural control simulations are a powerful tool for evaluating principles of motor control that cannot be tested directly in conventional experimental settings. However, current controllers struggle with physiological grounding and are restricted to a small set of discrete behaviours, despite the inherent heterogeneity of human movement. Here, we propose a new neural controller, the Internal Model-based Modular Controller (IMMC), which can transition between multiple lower-limb motor tasks within a single simulation. The architecture comprises a simplified model of the Mesencephalic Locomotor Region (MLR), which sends control signals to activate internal models (IMs). These IMs organise functional muscle synergies into task-specific networks that generate coordinated and stereotyped activity across multiple muscles. The IMMC performs Stand-To-Walk, Stand-To-Backward-Walking, Walking-To-Stand, and Backward-Walk-To-Stand transitions, and modulates gait speed by adjusting a single control signal. Across all behaviours, the simulated kinematics, muscle activation patterns and ground reaction forces were generally consistent with experimental observations. These findings suggest that interactions between internal models and functional muscle synergies offer a plausible mechanism for producing heterogeneous motor behaviours, providing a conceptual basis for more flexible neuromuscular control architectures.