In stroke rehabilitation, the assessments of the severity of stroke that are based on objective and robust measurements are the key to improve the efficacy of the rehabilitation efforts. It is essential, therefore, to complement the existing tools, where the assessments are partly relied on therapists’ subjective judgements, with a tool that can quantify important indicators of stroke recovery. One such indicator is the level of spasticity. The reliability of the current methods of measuring the severity of spasticity can be significantly improved by incorporating a feasible way to measure muscle forces and activations during stroke assessment. However, most of the present methods of estimating muscle forces require input parameters that are difficult to obtain in a clinical setting. A musculoskeletal arm model has been developed to bridge the gap between the domains of muscle forces estimation and stroke rehabilitation assessment. The project is divided into three stages. In the first stage, a biomechanical arm model that computes the joint torques with kinematic data from sensors is developed. The model has three features that eliminate the need for parameters that are difficult to obtain thus making it a feasible tool in clinical settings. The first is the use of a hybrid method that combines the data from sensors and a shoulder rhythm model to compute the orientation of the shoulder complex. The second is a method to compute the elbow joint angles without the need to compute the ambiguous carrying angle. The third is a method of estimating the inertial properties using published data, scaled by parameters that can be easily measured. The musculoskeletal properties of the human arm are added to the model in the second stage. The muscle model consists of 22 muscles that span from the thorax via the shoulder and the upper arm to the forearm. The muscle path is defined using Obstacle Set method where the anatomical structures are modelled using regular-shaped rigid bodies. Dynamics of the muscle is computed based on the Hill’s type muscle model that consists of an active contractile element, a passive parallel element and a series element. Due the difficulties in defining the moment arms, an optimization routine is designed to compute the optimal moment arms for each muscle for a subject. The muscle-sharing problem is solved using optimization which minimises the square of sum of muscle stresses. The muscle activation predicted by the model is compared to EMG signal for validation. In the final stage of this project, the model is used in the application of spasticity assessment. The tonic stretch reflex threshold (TSRT) which is an indicator for the severity of spasticity is computed using the model. Fifteen patient subjects participated in the experiments where they were assessed by two qualified therapists using Modified Ashworth Scale (MAS), and their motions and EMG signals were captured at the same time. Using the arm model, the TSRT of each patient was measured and ranked. The estimated muscle activation profiles have a high correlation (0.707) to the EMG signal profiles. The null hypothesis that the rankings of the severity using the model and the MAS assessment have no correlation has been tested, and was rejected convincingly (p ≈ 0.0003). These findings suggest that the model has the potential to complement the existing practices by providing an alternative evaluation method.

We consider a new approach to digital human simulation, using Model Predictive Control (MPC). This approach permits a virtual human to react online to unanticipated disturbances that occur in the course of performing a task. In particular, we predict the motion of a virtual human in response to two different types of real world disturbances: impulsive and sustained. This stands in contrast to prior approaches where all such disturbances need to be known a priori and the optimal reactions must be computed off line. We validate this approach using a planar 3 degrees of freedom serial chain mechanism to imitate the human upper limb. The response of the virtual human upper limb to various inputs and external disturbances is determined by solving the Equations of Motion (EOM). The control input is determined by the MPC Controller using only the current and the desired states of the system. MPC replaces the closed loop optimization problem with an open loop optimization allowing the ease of implementation of control law. Results presented in this thesis show that the proposed controller can produce physically realistic adaptive simulations of a planar upper limb of digital human in presence of impulsive and sustained disturbances.