DescriptionHuman walking is a fundamental motor skill that is developed at an early stage in our lives. Maintaining stable walking capability demands a substantial effort and requires synchronization and coordination of many neurological, sensorimotor and musculoskeletal systems. Moreover, disturbances such as foot slip require even more demanding walking control strategies for successful balance recovery and fall prevention. However, it is challenging to capture and model human motion and reaction to foot slip. Most of the existing slip-and-fall studies focus on clinical human experiments and few use control systems approaches to analyze the slip dynamics and human recovery mechanisms. Further challenges arise as few real-time sensing and robotic assistive technologies are currently available for reliably detecting foot slips and assisting human balance for slip-induced fall prevention. The goal of this dissertation is to advance the understanding and knowledge of slip dynamics with emphasis on four interlaced topics: (i) analyzing and modeling the shoe-floor interaction during foot slip, (ii) developing a novel bipedal modeling framework to capture human walking locomotion with foot slip, (iii) developing a novel linear inverted pendulum (LIP) modeling framework for balance recovery control, and (iv) developing new wearable sensing and robotic assistive devices for real-time detection of foot slip and effective prevention of slip-induced falls. In the first part, we present modeling of foot slip evolution and development based on a quasistatic friction force model. We present a model to obtain the normal force distribution on the shoe-floor contact patch. In addition, we extend the previously developed beam-spring network model and integrate it with the LuGre dynamic friction model. In the second part of the dissertation, we present a new bipedal modeling approach, where we relax the non-slip assumption used in the existing literature. We develop a hybrid bipedal model and the gait controllers to capture and predict human walking with foot slip. In the third part, we present a new two-mass LIP model for human balance control during walking and walking with foot slip. We extend the capture point based control approach and incorporate time-varying locations of the zero moment point and the LIP pivoting location. In the fourth part, we propose a novel real-time foot slip detection method using only wearable inertial measurement units. The developed slip-prediction algorithm is built on a dynamic model for bipedal walking and is also integrated with the human locomotion constraints. A slip indicator is introduced into the algorithm to detect the foot slip shortly after the heel-strike event. All of the above mentioned models, control strategies and devices are validated through the extensive experiments and simulations. In addition, we further design and fabricate a wearable robotic knee assistive device for slip balance recovery and slip-induced fall prevention. This device prototype serves as an enabling tool for future testing of possible robotic assistive fall prevention strategies.