Swarm Intelligence in Autonomous Heterogeneous Robotic Navigation Over Land and Water Using a Single Algorithm.
Exploration, Surveillance operations, Search and Rescue, and Security are becoming more
important to both civilian and military endeavors everyday. However, several
applications are dangerous for humans, which points to a critical role for robots.
Teamwork is often carried out by someone commanding the operations as a central leader,
but if contact with this person is lost, the team ... read moreoften has to quickly regroup and
review options under high pressure which can potentially be dangerous and a threat to
success. Such situations may be resolved by employing autonomous agents to work together
to accomplish a goal in which there is no full-time central leader. Agents employing
swarm intelligence offers a significant advantage over a group with an assigned leader
because there is no single point of failure. This allows for continuation of the
mission, even if agents are lost during it. For my research, I developed a single
algorithm for a scalable heterogeneous swarm of agents to reach a beacon located on land
or water. Agents collectively determine which is the closet to it and sends that agent
forward. They make decisions on which terrain they can transverse using local surface
reflectivity as a cue to check knowledge of their own physical capabilities or message
another if it is more fit to traverse. The swarm consists of a water-agent and multiple
land-agents that were able to distinguish land from water with 100% accuracy in
experiments performed. Results of various tested scenarios demonstrated that while the
dynamic ad-hoc mesh network I created allowed the swarm to succeed in broadcast
communication with each other 78% of the time, the swarm had an overall success rate of
84% in reaching the beacon in experiments performed. These results demonstrated that the
heterogeneous swarm was robust and capable of overcoming several challenges it faced in
the scenarios. This research can provide directions for future research in swarm
intelligence using heterogeneous agents with numerous real-world
Thesis (M.S.)--Tufts University, 2012.
Submitted to the Dept. of Mechanical Engineering.
Advisor: Chris Rogers.
Committee: Jason Rife, and Ethan Danahy.
Keyword: Engineering.read less
- Component ID:
- To Cite:
- TARC Citation Guide EndNote