TDMA-based exchange policies for multi-robot communication of world information

The aim of this paper is to present a distributed world model system for multirobot cooperation. Information is stored in semantic form in an ontology, allowing unlimited new concepts to be added. Automatic exchange of information is driven by the prioritised information needs of each vehicle, and is robust in the face of high latency, low bandwidth, lossy acoustic communications. The performances of different TDMA-based exchange policies were compared to a standard acknowledgement policy, through tests in simulation. Under certain conditions, particularly high packet loss, the matrix-based acknowledgement policies introduced here substantially reduced time to reliable exchange; the results suggest an adaptive exchange policy may perform best under varying conditions.