• upper-limb prostheses
• myoelectric
• multi-EMGs

This work aims at developing and evaluating of myoelectric control architectures for upper-limb prostheses. This kind of device uses electromyographic (EMG) signals acquired from muscle activations to control one or more motors on the artificial limb. Specifically, the goal consists in designing and testing proportional control algorithms that allow transradial amputees to command a prosthesis with 2 degrees of freedom (DOFs). The obtained systems use only two independent signals extracted from a larger group of EMGs by factorisation techniques like Non-negative Matrix Factorization (NMF).
The performance of the developed controllers have been evaluated by using a virtual reality environment and a commercial multi EMGs acquisition device. The two best controllers emerged from a first experiment on 12 subjects. The chosen algorithms were then tested on two different prosthetic devices. The first one uses the input signals to actuate two synergies on a robotic hand. The second one, instead, uses those inputs to move the first synergy of the hand and a wrist implementing pronation-supination.