Development of Humanoid Walking Robot with a Bio-inspired Knee Joint

Abstract

In recent centuries, the development of robots has greatly improved human’s life. The most advanced robots still show a large difference compared to humans with a natural waking gait. Although the rigid structure with simplified joints reduces the control complexity of humanoid robots, it decreases energy efficiency and walking stability as well. The compliant components integrated into the human musculoskeletal system enable simple control with high robustness and stability. Thus, a humanoid walking robot with a bio-inspired knee joint was raised in this thesis. Firstly, bio-inspired joints were purposed and constructed, especially for the knee joint. To replicate the mechanical intelligence design in human joints, the design of a bio-inspired knee joint used compliant artificial ligaments with an irregular articular surface. The artificial ligaments were braided by fishing line, and the mechanical properties were determined by a mathematical model. Simultaneously, the layout of artificial ligaments in a bio-inspired joint was set by a 2D mathematical joint model. The bio-inspired knee joint assembly demonstrated highly flexible 6 degrees of freedom (DOFs) and a human-like motion range. The humanoid walking robot was assembled, including the bones, artificial muscles, and joints. The robot was controlled by a motor that was driven as a hip joint in an open loop, aiding with passive artificial muscles. The artificial muscles were fabricated by the Nylon line and the stiffness was determined by a simulation model. The finished humanoid walking robot is 1.15 m high and weighs 7.5 kg with a total of 52 DOFs. As a result, the experiments of the humanoid walking robot exhibit a natural walking gait in three dimensions. The kinematics of the bio-inspired knee joint showed a high similarity to the bi-plane x-ray data of the human knee. The ankle joint trajectories were also comparable to the human data. With a user-friendly control system, the humanoid walking robot can adapt to different speeds from 2.0 km/h to 4.5 km/h. The stability and energy cost experiments were evaluated and showed excellent performance. At least 30 minutes of walking with a comparable energy cost to humans was conducted by the humanoid walking robot. It also exhibited the ability to walk on different slopes and resist obstacles. In conclusion, the humanoid walking robot gives a guide to achieve a human-like walking gait and plays the role of a test platform to understand human biomechanics underlying the walking principles in future experiments.

Date of Award	1 Aug 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Lei Ren (Supervisor) & Marco Domingos (Supervisor)