Control Oriented System Modelling and Instrumentation of Intelligent Walker-Human Systems

Abstract

Active intelligent walkers (i-walkers) are promising to provide stable and efficient motion to people with walking disabilities. There are three focuses in this thesis: system modelling, controller design, and instrumentation of an active type i-walker system considering interaction with human user as well. Two different control oriented system models and one high-fidelity model are proposed. All system models are designed as having two-body kinematics and consider the physical human-walker interaction (pHWI) based on the user gait dynamics and characteristics. The two-body kinematics of the systems define the relative motion of the user body with the i-walker. The dynamic models of control oriented systems are developed as single-body i-walker dynamics with human effect. These single-body models are classified as symmetric vs. asymmetric, and designed considering the vertical force components of pHWI on the i-walker. The symmetric model is developed for the center of gravity (CG) displacement of the i-walker human system only along the symmetric axis. The asymmetric model is more comprehensive than the symmetric one including the lateral CG displacement as well. The high-fidelity model considers the effect of human user as a separate dynamic body and covers the horizontal and vertical force components of pHWI during walking. Different control schemes are designed for each of the models with single-body dynamics, demonstrating the characteristics and efficiency of each model. All of these control designs are based on the same inverse kinematic controller, which utilizes two-body kinematic model and provides the desired i-walker velocities regarding the user motion intention. Each of the aforementioned dynamic controllers is designed to generate the torques required to track these desired velocities. Two different dynamic control schemes are proposed for dynamic controller: Proportional-integral-derivative (PID) and sliding mode controllers. The designed controllers are simulation tested in MATLAB/Simulink for the control oriented models. The asymmetric model parameters are also set for two different type of users with symmetric and asymmetric gait patterns, respectively. Finally, instrumentation of i-walkers for implementing the designed controllers is discussed, including presentation of a new human motion detection technique involving laser range finder (LRF) and encoders, and instrumentation is performed regarding the designed controllers.