Development and control of an intelligent social robot

Abstract

Humanoid robots face a significant challenge in achieving stable movement due to complex leg movements and limited processing power. The NAO robot developed by Aldebaran Robotics, highlights these difficulties with its rigid walking system, which restricts improvements in control and speed. These restrictions result in unnatural movements, decreased energy efficiency, and a higher possibility of falls, particularly in unpredictable circumstances. The importance of dynamic stability and balance control for effective robot movement on different surfaces has led to a growing interest in the control of human-robot movement. Extensive research has focused on discovering walking patterns that enable humanoid robots to maintain dynamic stability and adapt to different surfaces, including creating static gaits for stability and adaptability. The study of dynamic stability has been crucial for understanding robot walking, emphasizing balance control as essential for bipedal locomotion. This knowledge is directly applied to humanoid robots, highlighting the necessity of fast response, balance maintenance, and rapidly walking for effective real-world interaction. Thus, stability advances theoretical comprehension and underpins practical humanoid robot functionality in dynamic environments. Despite these advancements, several challenges persist, including addressing unknown dynamics, energy efficiency, control complexity, terrain adaptability, and integrating responsive interactions with changing environments. The first step of this thesis involved conducting extensive simulations to analyze the dynamic models of the NAO robot. Moreover, our methodology was put to the test in various real-life scenarios, showcasing the NAO robot’s ability to achieve stable locomotion at high velocities and on inclined surfaces. In addition, the system converts torque into executable positional commands, allowing for precise and efficient control of the NAO robot’s movements. This ensures that the robot can perform complex tasks accurately and reliably, enhancing its operational effectiveness. This thesis improves the NAO robot’s walking stability with three innovative control strategies. First, introducing Modified Function Approximation Techniques (MFAT) with adaptive sliding mode control allows for precise motor control of the vritual NAO robot . Second, the Nonlinear Disturbance Observer-based Fixed-Time Terminal Sliding Mode (NDO-FTSM) approach effectively manages torque inputs and converge tracking errors to zero under uncertain conditions. Finally, Fixed-Time Terminal Sliding Mode control with Fixed Time Observer (FTO) enhances stability amid disturbances and limited computational resources of the NAO robot.

Date	10 Oct 2024
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Maarouf Saad (Supervisor), Mohammad H. Rahman (Co-supervisor) & Roberto Erick Lopez Herrejon (Co-supervisor)