Helping Humans Become Better Teachers for Robots with Augmented Reality
Keywords: Augmented-Reality, Data-Visualization, Reinforcement-Learning
Abstract: We demonstrate TRAinAR, an augmented reality (AR)-based tool that is designed to improve sim2real reinforcement learning (RL) for robots. TRAinAR aims to enable users to train a robot by quickly prototyping complex environments in a virtual training environment with constraints to match the real-world. TRAinAR} also allows a user to visualize the robot's training data in context of the environment which potentially can provide insights into ways to improve the robot's training process. In this paper, we propose a human-participant study to evaluate TRAinAR as a valuable training tool. The proposed user study will help humans better identify ways to teach and improve a robot's learning process. In a technical demonstration, our application enabled a robotic arm manipulator to learn how to navigate its end-effector toward a goal object while implicitly learning to avoid obstacles.
Submission Number: 1
Loading