Sparse Online Locally Adaptive Regression using Gaussian Processes for Continual Robot Model Learning and Control
Machine learning methods have been explored more recently for robotic control, though learning the inverse mapping from sensor outputs to joint inputs in real time remains challenge for real-time control, particularly in the task of teleoperation, where commands and sensor data are received in sequential streams. In this thesis, we explore recent advances in Gaussian Processes and aim to achieve online model-learning and control for a teleoperation task on a real robot. We combine sparse, local, and streaming methods to form Sparse Online Locally Adaptive Regression using Gaussian Processes (SOLAR GP), which trains streaming data on localized sparse Gaussian Process models and infers a weighted local function mapping of the robot sensor states to joint states. The resultant prediction of the teleoperation command is used for joint control. The algorithm was adapted to perform on arbitrary link manipulators including the Baxter robot, where modifications to the algorithm are made to run training and prediction in parallel. This framework allows for a user-defined cap on complexity of generated local models while retaining information on older regions of the explored state-space.