An offline approach to fNIRS-guided reinforcement learning for robot behavior
An offline approach to fNIRS-guided reinforcement learning for robot behavior
This research paper explores using functional near-infrared spectroscopy (fNIRS) brain signals to modulate robot learning in simulation. The authors compare agents trained on passive versus active interaction tasks and test methods for enhancing the RL algorithm with neural signals, focusing on parameter augmentation. Results show that the framework is effective, particularly when augmenting trajectory priorities and state-action q-targets, and it successfully learns from offline data, providing a practical alternative for settings where real-time BCI setups are impractical.
The study demonstrates that offline fNIRS signals can effectively augment reinforcement learning agents in simulation by modifying trajectory priorities and state-action q-targets, offering a practical alternative to real-time BCI setups.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S2
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
arXiv:2607.14393v3 Announce Type: replace Abstract: Human-in-the-loop Reinforcement Learning has become a popular approach for 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 (de
Open source S2
Source：arXiv Robotics — research abstracts · arxiv.org