Myoelectric controlled hand prosthetics are artificial hands controlled by electrical signals generated naturally by the muscles in a residual limb. These prosthetics play a pivotal role for users in their life, enabling them to do daily activities, as well as feeling accepted as part of society. However, there is a significantly high abandoned rate which is as high as 44% in some countries [1]. This demonstrates that currently available commercial prosthetics are not meeting user needs or expectations, despite recent technologic advances.
One significant factor in myoelectric prosthetics abandonment is the control system. It is not seen to be reliable and the hand grip selection as being a slow and complicated process [2]. Therefore, implementing a novel robust control system for hand control, based on a computer vision approach can address this issue.
The aim of this project is to develop a computer vision-based control system for a novel low-cost myoelectric prosthetic to provide improved capability and reliability in gripping objects, together with hand dexterity. In addition, a myoband combined with sEMG electrodes will be employed to assess whether EMG calibration time during user trials can be minimised, while ensuring reliable control signals. A user centred design approach will be taken, ensuring users needs are addressed to maximise acceptance by the community, and improvements to the quality of life for users. Specific objectives include;
O1: Development of computer vision algorithm featuring object detection and grip mode classification, capable of real-time performance.
O2: Integration into the myoelectric prosthetic and combining with EMG actuation.
O3: Engaging with end users through testing, discussion and feedback sessions.
The projects deliverable will be a myoelectric prosthetic with integrated computer vision control approach.