Abstract
Humanoid robots are considered as complex and challenging platforms, and the state of the art in robotics. Humanoid robots are naturally expected to perform a wide variety of tasks using the same tools as humans, to operate in unconstrained environments, and to interact with other robots and humans in the same way we do. Humanoid robots are envisaged to be used in hazardous environments and as assistants to humans in the home or work place. In space, these platforms are considered as pre-human explorers operating in dangerous environments. The suitability of humanoid robots for space exploration has been acknowledged by NASA with the launch of the humanoid Robonaut to the International Space Station. As the expectations of these platforms continues to grow, many challenges still exist on how to control and manipulate such systems to perform the tasks expected humans. For example, maintaining the robots balance under different perturbation, as well as generating a stable, fast and efficient walking gait, is an important requirement that has to be naturally inherited in these platforms. However, the large number of degrees of freedom, and the non-linear chaotic nature of robot dynamics, result in the increased difficulty in manipulating the full body behaviours of these robots. The main goal of this research is to develop an efficient model that captures the full body behaviour accurately, while restoring balance and controlling the locomotion system. The Spherical Inverted Pendulum (SIP) concept was developed to model the biped robot centre of mass motion using the ankle joints. A novel balancing control law based on the principles of dissipative systems is developed and presented. It has been demonstrated that this controller restores balance by dissipating the kinetic energy introduced in the system as a result of disturbances. The SIP model is later used in the development of a balance and locomotion control framework using the concept of passive dynamic walking, and full body inverse kinetics, to achieve efficient and robust locomotion gait for the biped robot. Simulations were used to validate the SIP model and the new control framework for balance restoration and walking. Hardware validation of the multi-task manipulation and the simultaneous execution of tasks is also developed and presented in this thesis using the Nao humanoid robot.