Search papers, labs, and topics across Lattice.
This study investigates the integration of brain signals from functional near-infrared spectroscopy (fNIRS) into reinforcement learning (RL) for robot behavior modulation, comparing passive and active interaction tasks. The research demonstrates that augmenting RL algorithms with neural signals enhances learning outcomes, particularly in adjusting trajectory priorities and state-action q-values. Notably, the framework successfully operates with offline data, providing a viable solution for scenarios where real-time brain-computer interface (BCI) implementations are not feasible.
Augmenting reinforcement learning with brain signals can significantly enhance robot learning, even when relying solely on offline data.
Human-in-the-loop Reinforcement Learning has become a popular approach to training, finetuning, and aligning robot behavior with user preferences. Our paper explores the feasibility of using brain signals via functional near-infrared spectroscopy (fNIRS) to modulate robot learning in simulation. We compare agents trained on passive (observational) versus active (demonstrative) interaction tasks, and test multiple methods for enhancing the RL algorithm with the neural signal, focusing on parameter augmentation rather than replacement. We further examine how model granularity and noise affect agent learning. Our results show that this framework is effective: the neural signal improves learning when augmenting trajectory priorities and state-action q-values. Additionally, the framework learns successfully from offline data, offering a practical alternative for settings where real-time BCI setups are impractical or only limited data is available.
