Description

The concept of employing ground swarm robotics to accomplish tasks has been proposed for future use in humanitarian de-mining, plume monitoring, searching for
survivors in a disaster site, and other hazardous activities. More importantly in the military context, with the development of advanced explosive detectors, swarm robotics with autonomous search and detection capability could potentially address the improvised explosive device (IED) problem faced by foot patrols, and aid in the search for hidden
ammunition caches and weapons of mass destruction (WMDs). The intent of this research is to leverage on agent based simulation to model a ground robotic swarm on a search and detection mission in a semi-urban environment rigged with stationary IEDs.
Efficient design of experiment (DOE) techniques and data farming are engaged to help identify controllable factors and capabilities that have the most impact on overall effectiveness. The focus of this thesis is to explore agent based simulation applied to swarm robotics; the technological and algorithmic aspects are not delved on. Results
from the simulations provide several insights on the impact of both decision and noise factors on the performance of the swarm. Incorporation of virtual pheromones as a shared memory map is modeled as an additional capability that is found to enhance the robustness and reliability of the swarm.