An Experience-Based Approach to Mobile Push-Manipulation

We present an experience-based pushmanipulation method, where the mobile robot learns through experimentation how the pushable real world objects with complex 3D structures move in response to various pushing actions. These experimentally acquired models are then used as building blocks for constructing achievable push plans via a Rapidly-exploring Random Trees variant planning algorithm we contribute. We test our method in a realistic 3D simulation environment and demonstrate safe and successful transportation and placement of a variety of passively mobile pushable objects.