Projects
Brain-Machine Interfaces
Developing a Flexible and Robust Software Environment for Brain-Machine Interface Research
Richard Andersen (Division of Biology and Biological Engineering)
SASE: Jennifer Yu, Scholar
The Andersen Lab is developing Brain-Machine Interfaces (BMIs) in humans for both basic and translational research. A BMI in its most basic form records brain activity while users perform tasks, such as reaching or grasping. Brain activity associated with the tasks is identified by a computer in real time. Thus, the user’s intent can be decoded and replicated in external devices such as interaction with a computer or robotic prosthetic. To date BMI devices are mostly ‘open-loop’, there is no direct feedback from the device to the brain other than information from the user’s vision. This is unlike the natural sensorimotor system which relies heavily on sensory feedback of both the movement (proprioceptive) and contact (cutaneous).
To replicate sensory feedback in the framework of BMIs, the lab is exploring the use of intracortical microstimulation to deliver sensory feedback directly into the brain. Small currents delivered through electrodes implanted in the primary sensory cortex can evoke various cutaneous and proprioceptive sensations. This is called a ‘closed-loop’ device (see a schematic below).
In order to realize a closed-loop BMI several individual devices for recording the neural signal, stimulating and presenting various visual scenes to the user’s must all be synchronized and communicate between one another. Closed-loop devices thus require a complex hardware and software architecture to function.
To address this, the Lab developed their own in-house Framework package for running multimodal experiments. The Schmidt Academy collaborated with the Andersen lab to explore the integration of this array of devices into a single software management package, providing insight on migrating the existing codebase to the Python programming language and documenting the technologies for the lab’s existing research pipeline.