Inferring and Learning Multi-Robot Policies by Observing an Expert
This addresses the challenge of efficiently transferring expert knowledge to multi-robot systems without direct policy access, though it is incremental as it builds on existing imitation learning and filtering methods.
The paper tackles the problem of enabling an untrained multi-robot team to perform a mission by observing an expert system, using a multi-hypothesis filtering technique to estimate controllers and training a neural network policy, resulting in performances comparable to the expert system in a perimeter protection scenario.
We present a technique for learning how to solve a multi-robot mission that requires interaction with an external environment by observing an expert system executing the same mission. We define the expert system as a team of robots equipped with a library of controllers, each designed to solve a specific task, supervised by an expert policy that appropriately selects controllers based on the states of robots and environment. The objective is for an un-trained team of robots (i.e., imitator system) equipped with the same library of controllers, but agnostic to the expert policy, to execute the mission, with performances comparable to those of the expert system. From un-annotated observations of the expert system, a multi-hypothesis filtering technique is used to estimate individual controllers executed by the expert policy. Then, the history of estimated controllers and environmental states is used to train a neural network policy for the imitator system. Considering a perimeter protection scenario on a team of differential-drive robots, we show that the learned policy endows the imitator system with performances comparable to those of the expert system.