Adaptive Learning in Brain-Robot Interactions

This project aims to develop a non-invasive brain-machine interface (BMI) that allows a user to direct a semi-autonomous robot to perform different tasks through brain signals. For this purpose, we aim to employ a two-way co-adaptation paradigm where both the user and robot adapt to each other such that the likelihood of committing the same error in future is reduced. Importantly, a limiting factor in the current BMI technology is a high mental workload required for controlling the robot. To reduce the mental workload of the user, we are interested in using principles of the adaptive shared control, such that the robot adaptively learns to anticipate the user’s mental intent based on a number of sensory readings. Thus, the user will address the task at a high level and all the low level details are handled automatically by the robot.

This project has a large number of potential applications in healthcare and rehabilitation. This research involves developing novel intelligent/adaptive algorithms, offline and online data analysis, conducting experimental research, and online evaluation of the developed adaptive strategies with a robotic application. The prospective students can work on one or a number of these aspects.

Students with good degrees on robotics, electrical engineering, computer science, mathematics, cognitive science or subjects where signal processing and artificial intelligence/machine learning may be applied are encouraged to apply. If you are interested in research in brain-machine interfaces, and are unsure about whether you have the right background, please get in touch.