This thesis proposes a new method of dividing a task between members of a robot team. This method is dynamic territoriality. Territoriality is an emergent method of distributing resources between animals- each animal independently marks and defends their own area. Territoriality has previously been used on robot teams with varying results; these robot territorial systems did little or no adaptation to the environment or task, unlike natural territorial systems. / A dynamic territorial system adapts the territories to fit the environment, and the task the robot team must perform. The algorithm used to generate the territories is a novel extension of the ant clustering algorithm. Given a topological map of the space, it divides a space into a pre-specified number of territories such that the territories have minimal contact and near-equal area. The dynamic territorial algorithm was tested using hand-generated topological maps, then on physical environments. A robot system was developed to traverse an area and generate a topological map usable by the algorithm. / Dynamic territoriality can be used by simple robot teams to organize tasks such as multi- robot cleaning, monitoring and surveillance. The territories divide a complex environment into several simpler environments; this makes many tasks easier to perform. As the dynamic territorial algorithm is designed to minimize connections between territories, robot surveillance teams can capture or track intruders most easily at the territorial boundaries. Predator-prey and robot simulations were used to demonstrate the effectiveness of this territorial system in multi-robot surveillance. / This thesis extends previous work in territorial robotics and biologically inspired algorithms to create a new multi-robot control system. This control system has been implemented in hardware using a new topological mapping system. The thesis shows that this new multi-robot control system can effectively survey an area. In particular, it can control the movement of large groups of targets, or targets that move faster than the robots. / Thesis (PhD)--University of South Australia, 2005.
| Identifer | oai:union.ndltd.org:ADTP/267420 |
| Creators | Richer, Toby. |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | copyright under review |
Page generated in 0.0017 seconds