Learning 6DoF Grasping Using Reward-Consistent Demonstration
This addresses the complexity of robot motion implementation for high-degree-of-freedom systems, but it is incremental as it builds on existing imitation and reinforcement learning methods.
The study tackled learning 6DoF grasping motions by dividing them into multiple tasks and combining imitation and reinforcement learning, resulting in a reduction in the steps required to learn the grasping motion.
As the number of the robot's degrees of freedom increases, the implementation of robot motion becomes more complex and difficult. In this study, we focus on learning 6DOF-grasping motion and consider dividing the grasping motion into multiple tasks. We propose to combine imitation and reinforcement learning in order to facilitate a more efficient learning of the desired motion. In order to collect demonstration data as teacher data for the imitation learning, we created a virtual reality (VR) interface that allows humans to operate the robot intuitively. Moreover, by dividing the motion into simpler tasks, we simplify the design of reward functions for reinforcement learning and show in our experiments a reduction in the steps required to learn the grasping motion.