Thesis Project Form

Accurate and reliable reconstruction of upper-limb movement is essential for quantitative motor assessment, rehabilitation monitoring and VR-supported motor training. While laboratory motion capture systems provide high accuracy, their use is often limited by cost, space requirements and the need for supervised acquisition protocols. Wearable visuo-inertial trackers represent a promising alternative because they provide six-degree-of-freedom estimates of sensor position and orientation in a portable and immersive setup. However, the pose of a body-worn tracker does not directly correspond to anatomical segment motion. Reliable upper-limb reconstruction therefore requires calibration procedures able to estimate the relationship between tracker frames, body segments and functional joint centers.

This problem is particularly relevant in rehabilitation, where patients may present reduced range of motion, weakness, pain or compensatory motor strategies that limit their ability to perform standardized calibration movements. A fixed calibration protocol may therefore be unfeasible or suboptimal for some individuals. This thesis addresses the need for patient-specific calibration strategies by developing and testing a simulation framework able to identify, from the residual motor capabilities of an individual, the calibration movements that minimize upper-limb reconstruction error.

The main objective of the thesis is to develop and evaluate a subject-specific calibration framework for upper-limb motion reconstruction using wearable visuo-inertial trackers. The work aims to determine which static postures or dynamic movements provide the most reliable estimation of shoulder and elbow functional joint centers, and how calibration performance changes when the available range of motion is reduced.

The main activities will include the development of a kinematic model of the upper limb, the implementation of algorithms for functional joint-center estimation, and the design of a MATLAB-based simulation framework to test candidate calibration protocols under controlled conditions. The simulator will generate subject-specific reachable workspaces from range-of-motion constraints and will compare different calibration strategies using a known kinematic ground truth. Realistic tracker noise will be included to assess the robustness of each protocol.

The project will also involve the integration of visuo-inertial tracker data with hand-tracking information and the implementation of a reconstruction pipeline suitable for real-time or near-real-time use in VR environments. Candidate calibration protocols will be compared in terms of reconstruction error, with particular attention to the identification of minimal but informative movements that remain feasible for individuals with limited motor abilities.

The thesis will provide training in biomechanical modeling, upper-limb kinematics and wearable motion capture. The student will learn how to model the upper limb as a kinematic chain, how to represent body segments and tracker reference frames, and how to estimate functional joint centers from movement data.

From a technical perspective, the student will gain experience in MATLAB programming, simulation-based method development, signal processing and quantitative error analysis under realistic noise conditions. The work will also include exposure to Unity/C# development for VR applications and to the integration of wearable visuo-inertial trackers with hand-tracking systems.

From an experimental and methodological perspective, the student will learn how to design calibration protocols, evaluate their robustness, compare static and dynamic calibration strategies, and interpret reconstruction accuracy in relation to rehabilitation constraints. The thesis will also provide training in experimental validation, responsible human-participant testing and the translation of motion-analysis methods toward clinical and rehabilitative applications.