• magnet
• Levenberg–Marquardt
• prosthetic
• limb
• upper
• embedded
• myokinetic
• MYKI
• localization
• controller

Deprivation of a limb, mainly due to vascular problems or trauma, strongly affects the amputee’s quality of life. This work is part of the MYKI project, a research project funded by the European Research Council which aims to develop and clinically evaluate a dexterous hand prosthesis with tactile sensing which is naturally controlled and perceived by the amputee through a Human-Machine Interface based on the use of magnets implanted into residual muscles.
In this thesis a myokinetic embedded controller has been realized which, through the continuous localization of Magnetic Markers placed into the muscles, has the purpose of controlling independent DoFs of the prosthesis in real time. This is achieved by acquiring the magnetic field with a matrix of 32 3-axial sensors and solving the localization of dipoles non-linear least squares problem using the Levenberg – Marquardt algorithm. Parallelization and optimization techniques were used to speed up code execution.
The performance of the myokinetic interface have been evaluated in terms of localization accuracy and precision, as well as in time taken by the algorithm to reach convergence, using both a 3D positioning platform and a physiologically relevant forearm mockup.