Abstract: This thesis contains contributions to contact simulation and human motion analysis. Effects of the foot and ankle modelling techniques on the foot kinematics and dynamics are investigated. The analyses are carried out based on experimental data obtained using a motion capture system. The appropriateness of modelling the human ankle joint based on a stationary axis of rotation is investigated and a technique is also proposed which is capable of predicting the directional changes of the ankle axis during the foot flexion. Furthermore, two main modelling assumptions related to the number of the foot segments and the dimension of the foot model were the subject of the foot dynamics analyses. Effects of these modelling assumptions on the ankle joint torque and power are determined. A framework was developed which quantifies the gait abnormality of multiple sclerosis (MS) patients using a Kinect camera. The reliability of such a framework in assessing gait parameters in MS patients is evaluated based on captured data by Kinect. Also, a novel set of MS gait indices based on the concept of dynamic time warping is introduced whichcan characterize a patient's gait pattern and quantify the subject's gait deviation from the healthy population. In the second part of the thesis, two algorithms, namely, the accelerated-box and the generalized inverse-based algorithms, were developed for contact dynamics simulation. The accelerated-box algorithm improves the simulation of rigid body contact problems, in particular when the system under consideration has redundant constraints. The mathematical formulation is expressed in terms of a mixed linear complementarity problem (MLCP). The accelerated-box approach is partly motivated by the box friction model which is one of the existing approaches to solve contact problems. The original box friction model suffers from certain drawbacks in the presence of a large number of contact points such as long computational time, divergence problems, and instability. On the other hand, the accelerated-box approach developed in this thesis overcomes such drawbacks by taking advantage of the sparse structure of the lead matrix of the MLCP. This new method reduces the sensitivity of the solution to the constraint relaxation terms and decreases the number of required pivots to obtain the solution, and hence, shorter computational times result. This approach accordingly suggests a more reliable method for real-time simulation of multibody systems. A method based on the use of the Moore-Penrose generalized inverse was developed to deal with systems with redundant contacts. This approach omits the necessity of relaxing the constraints when redundancy exists in the system. To develop such a method, the generalized inverse is incorporated inside the pivoting steps of the MLCP solver. The method is very stable and robust, and its computational time is considerably smaller than the counterpartmethods, specially for highly redundant systems. Finally, a novel complementarity problem formulation is introduced. In…