Skip to content

Master Thesis: Improved prosthetic control leveraging deep learning on time-series

By Kirstin Ahmed


A prosthetic limb is commonly controlled by interpreting myoelectric signals from the muscles left after amputation. Currently we use different deep learning approaches to predict which movement the prosthesis user is intending to execute to accomplish a task (check out this video of a patient using one of the current algorithms). A promising, yet unexplored, avenue to improve movement predictions is to consider the chronology of the acquired data, i.e. to consider the acquired data as a time series. In this project, you’d be implementing, testing, and comparing different time-series prediction algorithms with the aim to improve prosthetic control accuracy.

Required field of study:

Computer science and/or engineering

Required skills:

Experience in machine learning (specifically deep learning)

Matlab and/or Python

Project duration:

30 or 60 ECTS