Human musculoskeletal system

Modeling

Human musculoskeletal system integrate the skeleton and muscles (as an actuator) to generate the body movement. Modeling this system consists in considering as an input the activation of muscles with a natural activation from brain (voluntary contraction), or with an artificial activation, based on Functional Electrical Stimulation (FES) applied on muscle or on motor nerves. Then, excited muscles are contracted and produce forces that generates the human body motion.

Parameters estimation

This research activity consists in estimating the physiological and dynamical parameters of human musculoskeletal model to make the model fit with the real system. This is a challenging aspect since the access to the state and measurements are complex with non-invasive context. Also, parameters estimation techniques that require to excite parameters through an appropriate trajectories, are not suitable since it may be unsafe for human. The challenge is to estimate enough parameters to express well the human behavior with the model and simulation. Another important issue concerns the high variabilities in behavior and parameters between subjects (inter-subject variability) and over time (intra-subject variability).

FES-based control

The Functional Electrical Stimulation (FES) mau be use to activate artificially muscles and thus generate the motion. Therefore, FES-based control of musculoskeletal system consists in calculating FES patterns and applying them in open or closed-loop mode. Our control strategies are model-based and exploit a physiological, complex and highly non-linear model of muscles. Several muscle modeling were used in my works, where the main one combine the well-known Hill's model with the microscopic Huxley's model.

Several control strategies of musculoskeletal system are investigated in my works. One strategy applied in open-loop and tested on a panel of real subjects with Spinal Cord Injury (SCI) is based on optimization of muscle activities criteria (Benoussaad et al. 2015). Another nonlinear control method based on the flatness of the system is invetigate as well (Benoussaad et al. 2020).

Human motion capture and analysis

IMU-based motion capture

The IMU are often sensors that measures acceleration (3-axis), rotation velocities (3-axis) and absolute orientation (3-axis). In our works, we use IMU as an alternative to classical optical motion capture systems which ra expensive, complex in terms of calibration and setting-up, limited in terms of space and suffer from the problem of occultation w.r.t cameras. However, the use of IMU implies a certain number of challenges, particulary when the acceleration data are used to estimate the position and drift issues. Hence, we previously used the accelerations data to estimate the human foot clearance in walking gait, where the drift were corrected using the walking context information (Benoussaad et al. 2016).

Meanwhile, the IMU, and particularly rotation velocities (gyroscopes) combined with accelerometers, were used with human model to capture and track its movement in real time. This were introduced in a virtual reality system to anaimate a human avatar for interactive and immersive simulations and for ergonomics analysis purposes.