MSc.Thesis Defense:Cem Kaan Karataş

CONTROL OF A BIPEDAL HUMANOID ROBOT FOR TROLLEY PUSHING APPLICATIONS

Cem Kaan Karataş
Mechatronics Engineering, MSc. Thesis 2024

Thesis Jury

Assoc.Prof. Dr. Kemalettin Erbatur (Thesis Advisor),

Prof. Dr. Mustafa Ünel

Dr. Özkan Bebek

Date & Time: July  22nd, 2024 –13.30

Place: FENS L056

Keywords: Humanoid Robot, Zero Moment Point, Linear Inverted Pendulum Model, Bipedal Walk, Full-Body Motion Control, Trolley Pushing

Abstract

The state of art in humanoid robotics is gradually improving in the last 40 years. While the industrial robot became a standard production support tool within few decades after its introduction in 1950s, the development pace of human shaped counter parts towards an era of daily life human-robot coexistence is limited. This is primarily due to the challenges of the bipedal kinematic arrangement, balancing requirements, difficulties in its control and last, but not the least safety related problems.

Despite the drawbacks, the bipedal humanoid structure is popular because of its potential in operating in human environment, using tools and user interfaces meant for human personnel. A robot of with a humanoid kinematic arrangement can climb stairs, fit in a car and reach various instruments and appliances designed for human ergonomy. However, for the structure to be useful, the robot's skill level must match the advantages of the shape. A large class of physical work involves manipulation tasks through hands and walking or balancing on the feet. Such functions are treated in the framework of full-body motion control for humanoid robots.

Pushing a trolley is a common task in both industrial and everyday life circumstances. This thesis is centered on this motion. To this end, a combination of reference generation and control techniques are applied for the legs and arms of the humanoid. Walking reference trajectories are based on the Zero Moment Point Criterion. When hands are in contact with the trolley, a foot-ground interaction force planner enhances body posture control. Hands are subject to position control when not contacting and are controlled via hybrid position-force control when they push the trolley.

Reference generation and control system is tested on the dynamic simulation environment of the humanoid robot SURALP designed at Sabancı University.