We present a system for visual robotic docking using an omnidirectional camera coupled with the actor critic reinforcement learning algorithm. The system enables a PeopleBot robot to locate and approach a table so that it can pick an object from it using the pan-tilt camera mounted on the robot. We use a staged approach to solve this problem as there are distinct sub tasks and different sensors used. Starting with random wandering of the robot until the table is located via a landmark, and then a network trained via reinforcement allows the robot to turn to and approach the table. Once at the table the robot is to pick the object from it. We argue that our approach has a lot of potential allowing the learning of robot control for navigation removing the need for internal maps of the environment. This is achieved by allowing the robot to learn couplings between motor actions and the position of a landmark.

www.his.sunderland.ac.uk This paper describes an approach in the field of reinforcement learning for robot control and a new Modular Actor-Critic architecture (MAC) which supports platform-independent robot control. The architecture is tested on a landmark approaching task using movable pan/tilt cameras which successfully control both, a large PeopleBot and a small Sony Aibo robot to perform the navigation task, with no retraining required. The architecture provides insight into the skills transfer between different robotic platforms and the modularisation of the architecture derived from splitting the control tasks into its component parts. The architecture and underlying principles could be used in rapid prototyping of new robotic platforms, where an already functioning control system can by used to allow more sophisticated navigation.