Extracting motor synergies from random movements for low-dimensional task-space control of musculoskeletal robots

Utilising redundancy in musculoskeletal systems for adaptive stiffness and muscle failure compensation: a model-free inverse statics approach

Vertebrates possess a biomechanical structure with redundant muscles, enabling adaptability in uncertain and complex environments. Harnessing this inspiration, musculoskeletal systems offer advantages like variable stiffness and resilience to actuator failure and fatigue. Despite their potential, the complex structure presents modelling challenges that are difficult to explicitly formulate and control. This difficulty arises from the need for comprehensive knowledge of the musculoskeletal system, including details such as muscle arrangement, and fully accessible muscle and joint states. Whilst existing model-free methods do not need explicit formulations, they also underutilise the benefits of muscle redundancy. Consequently, they necessitate retraining in the event of muscle failure and require manual tuning of parameters to control joint stiffness limiting their applications under unknown payloads. Presented here is a model-free local inverse statics controller for musculoskeletal systems, employing a feedforward neural network trained on motor babbling data. Experiments with a musculoskeletal leg model showcase the controller's adaptability to complex structures, including mono and bi-articulate muscles. The controller can compensate for changes such as weight variations, muscle failures, and environmental interactions, retaining reasonable accuracy without the need for any additional retraining.

An in vitro hand simulator for simultaneous control of hand and wrist movements

A human hand is a complex biomechanical system, in which bones, ligaments, and musculotendon units dynamically interact to produce seemingly simple motions. A new physiological hand simulator has been developed, in which electromechanical actuators apply load to the tendons of extrinsic hand and wrist muscles to recreate movements in cadaveric specimens in a biofidelic way. This novel simulator simultaneously and independently controls the movements of the wrist (flexion/extension and radio-ulnar deviation) and flexion/extension of the fingers and thumb. Control of these four degrees of freedom (DOF) is made possible by actuating eleven extrinsic muscles of the hand. The coupled dynamics of the wrist, fingers, and thumb, and the over-actuated nature of the human musculoskeletal system make feedback control of hand movements challenging. Two control algorithms were developed and tested. The optimal controller relies on an optimization algorithm to calculate the required tendon tensions using the collective error in all DOFs, and the action-based controller loads the tendons solely based on their actions on the controlled DOFs (e.g., activating all flexors if a flexing moment is required). Both controllers resulted in hand movements with small errors from the reference trajectories (<3.4); however, the optimal controller achieved this with 16% lower total force. Owing to its simpler structure, the action-based controller was extended to enable feedback control of grip force. This simulator has been shown to be a highly repeatable tool (<0.25 N and <0.2 variations in force and kinematics, respectively) for in vitro analyses of human hand biomechanics.

On the Relationship Between Muscle Synergies and Redundant Degrees of Freedom in Musculoskeletal Systems

It has been suggested that the human nervous system controls motions in the task (or operational) space. However, little attention has been given to the separation of the control of the task-related and task-irrelevant degrees of freedom. Aim: We investigate how muscle synergies may be used to separately control the task-related and redundant degrees of freedom in a computational model. Approach: We generalize an existing motor control model, and assume that the task and redundant spaces have orthogonal basis vectors. This assumption originates from observations that the human nervous system tightly controls the task-related variables, and leaves the rest uncontrolled. In other words, controlling the variables in one space does not affect the other space; thus, the actuations must be orthogonal in the two spaces. We implemented this assumption in the model by selecting muscle synergies that produce force vectors with orthogonal directions in the task and redundant spaces. Findings: Our experimental results show that the orthogonality assumption performs well in reconstructing the muscle activities from the measured kinematics/dynamics in the task and redundant spaces. Specifically, we found that approximately 70% of the variation in the measured muscle activity can be captured with the orthogonality assumption, while allowing efficient separation of the control in the two spaces. Implications: The developed motor control model is a viable tool in real-time simulations of musculoskeletal systems, as well as model-based control of bio-mechatronic systems, where a computationally efficient representation of the human motion controller is needed.

A Synergy-Based Motor Control Framework for the Fast Feedback Control of Musculoskeletal Systems

This paper presents a computational framework for the fast feedback control of musculoskeletal systems using muscle synergies. The proposed motor control framework has a hierarchical structure. A feedback controller at the higher level of hierarchy handles the trajectory planning and error compensation in the task space. This high-level task space controller only deals with the task-related kinematic variables, and thus is computationally efficient. The output of the task space controller is a force vector in the task space, which is fed to the low-level controller to be translated into muscle activity commands. Muscle synergies are employed to make this force-to-activation (F2A) mapping computationally efficient. The explicit relationship between the muscle synergies and task space forces allows for the fast estimation of muscle activations that result in the reference force. The synergy-enabled F2A mapping replaces a computationally heavy nonlinear optimization process by a vector decomposition problem that is solvable in real time. The estimation performance of the F2A mapping is evaluated by comparing the F2A-estimated muscle activities against the measured electromyography (EMG) data. The results show that the F2A algorithm can estimate the muscle activations using only the task-related kinematics/dynamics information with ∼70% accuracy. An example predictive simulation is also presented, and the results show that this feedback motor control framework can control arbitrary movements of a three-dimensional (3D) musculoskeletal arm model quickly and near optimally. It is two orders-of-magnitude faster than the optimal controller, with only 12% increase in muscle activities compared to the optimal. The developed motor control model can be used for real-time near-optimal predictive control of musculoskeletal system dynamics.