Combining Implicit and Explicit Communication in Object Manipulation Tasks Between Two Robots

The number of automated environments where multiple robots must cooperate with one another and humans on object manipulation is increasing. From the rapidly expanding automated warehouse industry, to research investigating search and rescue applications, the need for effective and robust communication between agents is essential to facilitate cooperation.

Human cooperative mechanisms, such as Joint Action, are known to enable effective cooperation between humans. In particular, humans are able to use a combination of explicit communication (the direct transfer of information, such as speech) and implicit communication (the indirect transfer of information, where information is inferred through an action, such as force-feedback) to cooperate on tasks. This behaviour could be used as inspiration for a robotic mechanism that improves communication between two robots when manipulating an object together.

This thesis presents a hybrid system that combines explicit and implicit communication in object manipulation tasks between two robots. The system combines communication strategies for explicit communication and implicit communication, which use wireless messaging and force information, respectively, using a weighted sum. In a Leader-Follower configuration, the Follower uses the hybrid system to coordinate its movements with the Leader, and the two robots jointly move an object along a pre-determined trajectory.
The hybrid system presented in this thesis is found to capitalise on the advantages of both strategies that have been identified by evaluating their performances when operating in isolation. In addition, different weightings for the hybrid system are identified that offer the best compromise in performance and robustness across four experimental environments.