Learning Tabletop Object Manipulation by Imitation

We aim to enable robot to learn tabletop object manipulation by imitation. Given external observations of demonstrations on object manipulations, we believe that two underlying problems to address in learning by imitation is 1) segment a given demonstration into skills that can be individually learned and reused, and 2) formulate the correct RL (Reinforcement Learning) problem that only considers the relevant aspects of each skill so that the policy for each skill can be effectively learned. Previous works made certain progress in this direction, but none has taken private information into account. The public information is the information that is available in the external observations of demonstration, and the private information is the information that are only available to the agent that executes the actions, such as tactile sensations. Our contribution is that we provide a method for the robot to automatically segment the demonstration into multiple skills, and formulate the correct RL problem for each skill, and automatically decide whether the private information is an important aspect of each skill based on interaction with the world. Our motivating example is for a real robot to play the shape sorter game by imitating other’s behavior, and we will show the results in a simulated 2D environment that captures the important properties of the shape sorter game. The evaluation is based on whether the demonstration is reasonably segmented, and whether the correct RL problems are formulated. In the end, we will show that robot can imitate the demonstrated behavior based on learned policies.