Dissertation Defense - Omid Arfaie Khosrowshahi

Omid Arfaie Khosrowshahi – PhD. Mechanical Engineering

Asst. Prof. Ramazan Ünal – Advisor

Date: 26.03.26

Time: 13:00

Location: AB4- 318

“DESIGN, DEVELOPMENT, AND CONTROL OF AN ACTIVE TRANSFEMORAL PROSTHESIS”

Assistant Professor Ramazan Ünal, Özyeğin University

Associate Professor Barkan Uğurlu, Özyeğin University

Associate Professor Polat Şendur, Özyeğin University

Professor Volkan Polatoğlu, Sabancı University

Assistant Professor Elif Hocaoğlu, İstanbul Medipol University

Abstract:

This thesis presents the design, development, functional evaluation, and control of lightweight robotic lower-limb prostheses to support amputee users during walking and stair climbing. A new knee actuator concept integrating an elastic element with a ball-screw actuator and a four-bar mechanism is proposed and benchmarked against Direct Drive, Series Elastic Actuator, and Parallel Elastic Actuator approaches, achieving actuator weight reductions of up to 68% and higher overall concept scores in weight, complexity, ease of movement, and balance. Based on this concept, the Rnee (Robotic Knee Prosthesis) has been developed and evaluated. Using 3D printed composite material in manufacturing the Rnee yields a device that is at least 10% lighter than the lightest reported robotic knee prosthesis and more compact than the human knee, and energy analysis indicates daily usability (5000 steps and 100 stairs per charge). Experimental evaluations show that Rnee provides over 88% of the required walking torque and ~90% of the stair-climbing torque, with gait kinematics correlating with natural profiles (82%) and power delivery reaching 87% of the biological requirement. Integrated robotic systems (Ankatron robotic ankle prosthesis and RoboLeg robotic transfemoral prosthesis) further demonstrate strong agreement with natural biomechanics (94% correlation in knee angle and 86% in ankle angle) and improved torque/energy characteristics via elastic mechanisms, while experiments with transtibial amputees confirm Ankatron’s functionality. Finally, machine-learning models (1D CNN, LSTM, and CNN+LSTM) are implemented for ankle angle and moment prediction, with LSTM achieving the best overall performance for sequential biomechanical data.

Bio:

Omid Arfaie is a Ph.D. candidate in Mechanical Engineering at Özyeğin University, Istanbul, Türkiye, where he conducts research in the Human-Centered Design Laboratory. He received his M.Sc. degree from the Iran University of Science and Technology (IUST), Tehran, Iran. His doctoral research focuses on the design and control of lower-limb assistive devices, including passive and active exoskeletons and robotic prostheses. His research interests include biomechanics, biomechatronics, rehabilitation robotics, and human–robot interaction.