Back to search

National Technical University of Athens (NTUA)

Design and control of anthropomorphic and energy-efficient robotic systems with applications in lower-limb prostheses

Abstract

dc:description

This thesis investigates the dynamics of bipedal locomotion through the modeling, simulation, and control of passive and active bio-inspired mechanical walkers, with emphasis on applications in robotics and lower-limb prosthetics. The research follows a structured progression, beginning with a bipedal walking base model that captures key aspects of human walking using compliant legs and semicircular feet. Despite its simplicity, this initial model exhibits stable passive walking cycles that resemble human gait. A nondimensional formulation enables sensitivity analysis, fixed-point identification, and energetic evaluation across a wide range of design parameters. The model’s predictions are validated through multibody simulations using a digital twin and through the development and use of an experimental prototype. Building on this foundation, the design of the model is gradually refined to incorporate more anatomically inspired features. The study advances through a series of footshape investigations, beginning with a semielliptical design that approximates the evolving curvature of the human foot during rollover. This model allows a parametric investigation of rollover dynamics, allowing the fine-tuning of gait descriptors and the optimization of energetic efficiency. The model is subsequently expanded to support any convex foot geometry, enabling the integration of biomimetic profiles extracted from experimental human walking data. These geometries allow the replication of biomimetic ankle trajectories and prediction of ground contact behavior with increasing fidelity, offering a clear pathway for prosthetic foot design that mirrors natural biomechanics. The work continues through the study of knee joint morphology. An initial passive kneed model is developed that uses simple pin joints. Passive gaits for this walker are identified, and the effect of parameter variations is investigated. The model is later compared with a novel biomimetic four-bar kneed model, inspired by the human cruciate ligament system. The four-bar configuration exhibits superior compliance, impact absorption and energetic efficiency, which are critical for the design of functional prosthetic knees. Comparative simulations reveal the importance of joint design in achieving both gait stability and energetic performance. To extend these passive behaviors to level-ground locomotion, a novel Under-actuated Virtual Gravity (UVG) control framework is introduced. This controller analytically replicates the gravitational torques experienced during downhill passive walking, enabling highly efficient locomotion with minimal actuation. Implemented on a bipedal robot model that features a counterweight torso, the UVG control scheme achieves dynamic behaviors closely aligned with the passive model, while requiring low motor effort, offering strong potential for integration into active prosthetic systems. Moreover, the UVG control output is compared to an alternative actuation approach based on trajectory optimization and stabilization. The UVG is found to outperform the alternative scheme, due to its passive dynamics-based approach. In parallel, a markerless, stereo-camera-based gait analysis system is developed to extract full-body kinematic and kinetic data from human walking trials. Using neural network-based pose estimation and inverse dynamics analysis, this system provides joint torque and ground reaction force estimates without requiring markers or force plates. These human gait data are used to inform and calibrate the robotic models, ensuring that the simulated behaviors converge to real human biomechanics. The outcome of this research is a comprehensive framework for bipedal walking analysis and robotic or prosthetic design. The combination of detailed modeling, parametric tuning and efficient control provides valuable tools for the development of lower-limb prostheses that closely emulate human motion. The work contributes to the understanding of energy-efficient gait generation, design-driven dynamics, and controller synthesis, offering new perspectives in both robotics and assistive device engineering.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2025

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Smyrli, Aikaterini
  • Σμυρλή, Αικατερίνη

Subjects

dc:subject × 20

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/60169

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Smyrli, Aikaterini; Σμυρλή, Αικατερίνη. Design and control of anthropomorphic and energy-efficient robotic systems with applications in lower-limb prostheses. National Technical University of Athens (NTUA), 2025. http://hdl.handle.net/10442/hedi/60169