Performance criteria for generating predictive optimal control simulations of bicycle pedaling

Humans trade off whole-body energy cost to avoid overburdening muscles while walking

Metabolic cost minimization is thought to underscore the neural control of locomotion. Yet, avoiding high muscle activation, a cause of fatigue, often outperforms energy minimization in computational predictions of human gait. Discerning the relative importance of these criteria in human walking has proved elusive, in part, because they have not been empirically decoupled. Here, we explicitly decouple whole-body metabolic cost and 'fatigue-like' muscle activation costs (estimated from electromyography) by pitting them against one another using two distinct gait tasks. When experiencing these competing costs, participants (n = 10) chose the task that avoided overburdening muscles (fatigue avoidance) at the expense of higher metabolic power (p < 0.05). Muscle volume-normalized activation more closely models energy use and was also minimized by the participants' decision (p < 0.05), demonstrating that muscle activation was, at best, an inaccurate signal for metabolic energy. Energy minimization was only observed when there was no adverse effect on muscle activation costs. By decoupling whole-body metabolic and muscle activation costs, we provide among the first empirical evidence of humans embracing non-energetic optimality in favour of a clearly defined neuromuscular objective. This finding indicates that local muscle fatigue and effort may well be key factors dictating human walking behaviour and its evolution.

Using Bayesian inference to estimate plausible muscle forces in musculoskeletal models

Background

Musculoskeletal modeling is currently a preferred method for estimating the muscle forces that underlie observed movements. However, these estimates are sensitive to a variety of assumptions and uncertainties, which creates difficulty when trying to interpret the muscle forces from musculoskeletal simulations. Here, we describe an approach that uses Bayesian inference to identify plausible ranges of muscle forces for a simple motion while representing uncertainty in the measurement of the motion and the objective function used to solve the muscle redundancy problem.

Methods

We generated a reference elbow flexion–extension motion and computed a set of reference forces that would produce the motion while minimizing muscle excitations cubed via OpenSim Moco. We then used a Markov Chain Monte Carlo (MCMC) algorithm to sample from a posterior probability distribution of muscle excitations that would result in the reference elbow motion. We constructed a prior over the excitation parameters which down-weighted regions of the parameter space with greater muscle excitations. We used muscle excitations to find the corresponding kinematics using OpenSim, where the error in position and velocity trajectories (likelihood function) was combined with the sum of the cubed muscle excitations integrated over time (prior function) to compute the posterior probability density.

Results

We evaluated the muscle forces that resulted from the set of excitations that were visited in the MCMC chain (seven parallel chains, 500,000 iterations per chain). The estimated muscle forces compared favorably with the reference forces generated with OpenSim Moco, while the elbow angle and velocity from MCMC matched closely with the reference (average RMSE for elbow angle = 2°; and angular velocity = 32°/s). However, our rank plot analyses and potential scale reduction statistics, which we used to evaluate convergence of the algorithm, indicated that the chains did not fully mix.

Conclusions

While the results from this process are a promising step towards characterizing uncertainty in muscle force estimation, the computational time required to search the solution space with, and the lack of MCMC convergence indicates that further developments in MCMC algorithms are necessary for this process to become feasible for larger-scale models.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12984-022-01008-4.

A direct collocation framework for optimal control simulation of pedaling using OpenSim

The direct collocation (DC) method has shown low computational costs in solving optimization problems in human movements, but it has rarely been used for solving optimal control pedaling problems. Thus, the aim of this study was to develop a DC framework for optimal control simulation of human pedaling within the OpenSim modeling environment. A planar bicycle-rider model was developed in OpenSim. The DC method was formulated in MATLAB to solve an optimal control pedaling problem using a data tracking approach. Using the developed DC framework, the optimal control pedaling problem was successfully solved in 24 minutes to ten hours with different objective function weightings and number of nodes from two different initial conditions. The optimal solutions for equal objective function weightings were successful in terms of tracking, with the model simulated pedal angles and pedal forces within ±1 standard deviation of the experimental data. With these weightings, muscle tendon unit (MTU) excitation patterns generally matched with burst timings and shapes observed in the experimental EMG data. Tracking quality and MTU excitation patterns were changed little by selection of node density above 31, and the optimal solution quality was not affected by initial guess used. The proposed DC framework could easily be turned into a predictive simulation with other objective functions such as fastest pedaling rate. This flexible and computationally efficient framework should facilitate the use of optimal control methods to study the biomechanics, energetics, and control of human pedaling.

A Conceptual Blueprint for Making Neuromusculoskeletal Models Clinically Useful

The ultimate goal of most neuromusculoskeletal modeling research is to improve the treatment of movement impairments. However, even though neuromusculoskeletal models have become more realistic anatomically, physiologically, and neurologically over the past 25 years, they have yet to make a positive impact on the design of clinical treatments for movement impairments. Such impairments are caused by common conditions such as stroke, osteoarthritis, Parkinson’s disease, spinal cord injury, cerebral palsy, limb amputation, and even cancer. The lack of clinical impact is somewhat surprising given that comparable computational technology has transformed the design of airplanes, automobiles, and other commercial products over the same time period. This paper provides the author’s personal perspective for how neuromusculoskeletal models can become clinically useful. First, the paper motivates the potential value of neuromusculoskeletal models for clinical treatment design. Next, it highlights five challenges to achieving clinical utility and provides suggestions for how to overcome them. After that, it describes clinical, technical, collaboration, and practical needs that must be addressed for neuromusculoskeletal models to fulfill their clinical potential, along with recommendations for meeting them. Finally, it discusses how more complex modeling and experimental methods could enhance neuromusculoskeletal model fidelity, personalization, and utilization. The author hopes that these ideas will provide a conceptual blueprint that will help the neuromusculoskeletal modeling research community work toward clinical utility.

Inclusion of actuator dynamics in simulations of assisted human movement

Simulation of musculoskeletal systems using dynamic optimization is a powerful approach for studying the biomechanics of human movements and can be applied to human-robot interactions. The simulation results of human movements augmented by robotic devices may be used to evaluate and optimize the device design and controller. However, simulations are limited by the accuracy of the models which are usually simplified for computation efficiency. Typically, the powered robotic devices are often modeled as massless, ideal torque actuators that is without mass and internal dynamics, which may have significant impacts on the simulation results. This article investigates the effects of including the mass and internal dynamics of the device in simulations of assisted human movement. The device actuator was modeled in various ways with different detail levels. Dynamic optimization was used to find the muscle activations and actuator commands in motion tracking and predictive simulations. The results showed that while the effects of device mass and inertia can be small, the electrical dynamics of the motor can significantly impact the results. This outcome suggests the importance of using an accurate actuator model in simulations of human movement augmented by assistive devices. NOVELTY: Demonstrating the effects of including mass and internal dynamics of the actuator in simulations of assisted human movement A new OpenSim electric motor actuator class to capture the electromechanical dynamics for use in simulation of human movement assisted by powered robotic devices.

Bilevel Optimization for Cost Function Determination in Dynamic Simulation of Human Gait

Predictive simulation based on dynamic optimization using musculoskeletal models is a powerful approach for studying human gait. Predictive musculoskeletal simulation may be used for a variety of applications from designing assistive devices to testing theories of motor control. However, the underlying cost function for the predictive optimization is unknown and is generally assumed a priori. Alternatively, the underlying cost function can be determined from among a family of possible cost functions, representing an inverse optimal control problem that may be solved using a bilevel optimization approach. In this study, a nested evolutionary approach is proposed to solve the bilevel optimization problem. The lower level optimization is solved by a direct collocation method, and the upper level is solved by a genetic algorithm. We demonstrate our approach to solve different bilevel optimization problems, including finding the weights among three common performance criteria in the cost function for normal human walking. The proposed approach was found to be effective at solving the bilevel optimization problems. This approach should provide practical utility in designing assistive devices to aid mobility, and could yield insights about the control of human walking.