Motion Control of High-Dimensional Musculoskeletal Systems with Hierarchical Model-Based Planning

TL;DR

This work tackles the challenge of controlling high-dimensional musculoskeletal systems without extensive training or reward engineering. It introduces $MPC^2$, a hierarchical framework that combines a sampling-based model predictive controller for target posture planning with a morphology-aware proportional low-level controller, enabling zero-shot, near-real-time control of a 700-actuator full-body model. The approach demonstrates stable control across standing, walking on varied terrains, and sports imitation, and supports rapid, automatic cost-function design via Bayesian optimization, dramatically reducing human tuning time. By enabling training-free planning and cross-morphology generalization (e.g., ostrich and arm models), MPC$^2$ offers a practical pathway to robust, real-time biomechanical control in both biological-inspired and robotic contexts.

Abstract

Controlling high-dimensional nonlinear systems, such as those found in biological and robotic applications, is challenging due to large state and action spaces. While deep reinforcement learning has achieved a number of successes in these domains, it is computationally intensive and time consuming, and therefore not suitable for solving large collections of tasks that require significant manual tuning. In this work, we introduce Model Predictive Control with Morphology-aware Proportional Control (MPC^2), a hierarchical model-based learning algorithm for zero-shot and near-real-time control of high-dimensional complex dynamical systems. MPC^2 uses a sampling-based model predictive controller for target posture planning, and enables robust control for high-dimensional tasks by incorporating a morphology-aware proportional controller for actuator coordination. The algorithm enables motion control of a high-dimensional human musculoskeletal model in a variety of motion tasks, such as standing, walking on different terrains, and imitating sports activities. The reward function of MPC^2 can be tuned via black-box optimization, drastically reducing the need for human-intensive reward engineering.

Figures (12)

• Figure 1: Movement control of whole-body human musculoskeletal system over a diverse set of motion control tasks. The videos of the control performances are on the https://lnsgroup.cc/research/MPC2.html.
• Figure 2: Workflow of Model Predictive Control with Morphology-aware Proportional Control (MPC$^2$). Solid arrows indicate control pipeline, and dashed arrows indicate planning procedure.
• Figure 3: Model Predictive Control with Morphology-aware Proportional Control (MPC$^2$)
• Figure 4: Control sequences of MPC$^2$ in (a) Stand, (b) Rough, (c) Walk, (d) Slope and (e) Stair (f) Ostrich walk tasks. The simulation speed of Stair task is set to 10% due to slower contact computation.
• Figure 5: Walking control under sudden muscle failure. (a) Illustration of muscle injury. (b) Control sequences of MPC$^2$. The dotted line shows the time when the muscles fail.