Image Sensor Classification System for Prosthesis Control

Abstract

This paper proposes the integration of an image sensor classification system into the control scheme for prosthetic limbs. The goal is to minimize the mental strain experienced by brain-computer interface (BCI) users when performing routine tasks. In particular, the focus is to investigate the capabilities of image classification as a method for identifying and classifying objects. Incorporating this process into prosthesis control will allow more fluid and natural movements while minimizing the mental strain of multiple commands by the user. A popular imaging and classification software package was evaluated by exposing it to various images under different conditions. Results indicate that an image sensor classification system is suitable for prosthetic limb control in many situations; however, the system performance is reduced in low light and clustered object conditions. It is therefore feasible to consider the integration of image sensor classification and camerain-hand technology into EMG- and BCI-controlled prosthetics.