Thesis Project Form
Title (tentative): Machine Learning-Based Human Pose Estimation and Gait Analysis for the Evaluation of a Soft Exoskeleton| Thesis advisor(s): Sanguineti Vittorio, Spirito Mario, Traverso Simone (Rehab Technologies Lab, IIT - CCT) | E-mail: |
| Address: Via All'Opera Pia, 13 - 16145 Genova | Phone: (+39) 010 33 56487 |
Description
Motivation and application domain
Wearable robotics and assistive technologies are rapidly evolving fields with significant potential to improve healthcare, rehabilitation, and mobility assistance. Understanding how the human body
moves and how soft exoskeletons affect gait is crucial to design effective and user-friendly devices. Recent advances in deep learning and computer vision now allow us to estimate skeletal structures and joint angles directly from video, opening the door to noninvasive gait analysis.
This thesis will contribute to this exciting research area by combining artificial intelligence, biomechanics, and robotics to create innovative methods for motion analysis. The work is interdisciplinary by nature and offers the opportunity to gain practical experience in both algorithm development and biomedical applications.
moves and how soft exoskeletons affect gait is crucial to design effective and user-friendly devices. Recent advances in deep learning and computer vision now allow us to estimate skeletal structures and joint angles directly from video, opening the door to noninvasive gait analysis.
This thesis will contribute to this exciting research area by combining artificial intelligence, biomechanics, and robotics to create innovative methods for motion analysis. The work is interdisciplinary by nature and offers the opportunity to gain practical experience in both algorithm development and biomedical applications.
General objectives and main activities
The purpose of the thesis is to develop a markerless machine learning-based algorithm for analyzing human gait and to evaluate the effectiveness of a soft exoskeleton. The main steps of the project are:
1. Data acquisition: Collect video recordings of human walking trials on both treadmill and overground conditions.
2. Pose estimation: Extract skeletal landmarks using open-source deep learning libraries such as PyTorch, TensorFlow, OpenPose, or MediaPipe.
3. Angle estimation: Compute orientation angles of lower-limb segments from the extracted skeletal data.
4. Gait phase detection: Segment walking into gait phases, using joint angle trajectories and machine learning techniques.
5. Comparative analysis: Evaluate gait patterns before and after wearing the soft exoskeleton, performing statistical analysis to assess the efficiency of the solution.
1. Data acquisition: Collect video recordings of human walking trials on both treadmill and overground conditions.
2. Pose estimation: Extract skeletal landmarks using open-source deep learning libraries such as PyTorch, TensorFlow, OpenPose, or MediaPipe.
3. Angle estimation: Compute orientation angles of lower-limb segments from the extracted skeletal data.
4. Gait phase detection: Segment walking into gait phases, using joint angle trajectories and machine learning techniques.
5. Comparative analysis: Evaluate gait patterns before and after wearing the soft exoskeleton, performing statistical analysis to assess the efficiency of the solution.
Training Objectives (technical/analytical tools, experimental methodologies)
The student will:
• Gain practical experience with state-of-the-art AI and computer vision tools in Python.
• Learn to combine biomechanics, robotics, and data-driven methods to tackle real-world challenges.
• Contribute to an interdisciplinary research project with potential impact in rehabilitation engineering and assistive technology.
• Work in a supportive environment with the freedom to test new ideas and approaches.
• Gain practical experience with state-of-the-art AI and computer vision tools in Python.
• Learn to combine biomechanics, robotics, and data-driven methods to tackle real-world challenges.
• Contribute to an interdisciplinary research project with potential impact in rehabilitation engineering and assistive technology.
• Work in a supportive environment with the freedom to test new ideas and approaches.
Place(s) where the thesis work will be carried out: Rehab Technologies INAIL-IIT Lab, Via Morego 30, Genova
Additional information
Maximum number of students: 1