This project seeks to evaluate alternative means of controlling a prosthetic (in this case, a hand) using electroencephalographic control. The project consists of four methods; an unsure-feedback neural network, a neural network which lets the user know where it assumes the user wants to go, if unsure; a neutrally-iterated tree, which stores a preset list of locations that the user moves between based on how intently they focus on a task; a continuously-trained neural network, which tries to assume the user's hand position and trains relative to that; and a direct neural network, as described above. The selected methods will be compared to determine training efficiency, accuracy, and response time relative to each other on a universal platform.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Polley, Patrick
• Gallati, Walter Gage

Publisher

• Worcester Polytechnic Institute

Identifier

• E-project-043018-083606

Advisor

• Beck, Joseph E.
• Popovic, Marko B.

Year

• 2018

Date created

• 2018-04-30

Resource type

• Major Qualifying Project

Major

• Robotics Engineering

Rights statement

• In Copyright