Data: Brain-machine interface control with artificial intelligence copilots

Dataset for "Brain-machine interface control with artificial intelligence copilots"

Abstract

Motor brain-machine interfaces (BMIs) decode neural signals to help people with paralysis move and communicate. Even with important advances in the last two decades, BMIs face key obstacles to clinical viability: BMI performance should strongly outweigh BMI costs and risks. We use shared autonomy, where artificial intelligence (AI) copilots collaborate with BMI users to achieve task goals, to significantly increase the performance of BMIs. We demonstrate this "AI-BMI" in a non-invasive BMI system decoding electroencephalography (EEG). We first contribute a hybrid adaptive decoding approach using a convolutional neural network (CNN) and ReFIT-like Kalman filter (KF), enabling healthy users and a paralyzed participant to autonomously and proficiently control computer cursors and robotic arms with EEG. We then demonstrate AI-BMIs that enable a paralyzed participant to (1) achieve 4.3× higher performance in a cursor control task and (2) control a robotic arm to sequentially move random objects to random locations, a task he could not do without an AI copilot. As AI copilots improve, BMIs designed with shared autonomy may achieve higher performance.