I have worked at AABL since early 2021. AABL seeks to design algorithms for Human-Robot Interaction (HRI). More specifically, the lab is working towards creating better assistive technology by creating more helpful robots.
I have worked to develop the HRI Sandbox, which is an interactive reinforcement learning environment. This environment allows for robot tasks to be simulated visually. It allows for humans to give feedback live as reinforcement learning models are learning to perform a task. I have worked on incorporating this human feedback back into the reinforcement learning models to improve their performance. To accomplish this, I implemented the TAMER framework, which compensates for the delay when humans give feedback.
