Abstract:
Swarm intelligence systems, wherein robotic devices are created with abilities
executed at individual level to create swarm level emergent behaviour, are
appealing to industries such as nanotechnology, amorphous computing, and
engineering. They are often considered suitable for determining solutions to
computationally complex problems within a context where large, sophisticated
robots are too expensive to procure. For instance, bee colony models, fish
schooling solutions, and ant colony systems have been proposed for solving
complex problems by imitating the behaviours of related natural species. Ant
colony systems, in particular, have been used to offer solutions to complex
optimisation issues such as the traveling salesman problem, the vehicle routing
problem, and the bridge crossing task. Compelling robust and fault tolerant
solutions have been produced and reported. However, formal representation of
the knowledge of the ant-like robotic devices thereof has not been fully explored.
In this context, ant-like robotic devices are interchangeably referred to as antbots.
An ant-bot, on its own, may not have capabilities to accomplish notable
solutions to problems. However, as a swarm, ant-bots can produce compelling
emergent behaviour more than the sum of the actions of the individual ant-bots
in the swarm. This study proposed the development of an ontology for
characterising the low-level behaviours and capabilities of simulated ant-like
robotic devices tasked to path find. First, we considered the distinctive qualities
of ant-bots and found out that foraging is key. Ant-bots require five key actions;
dropping pheromone-like indicators, flipping between different internal states,
orientating based on the amount of the virtual pheromone within their local
environment, making informed movements when it becomes necessary, and no
action as self-explanatory. The key meta needs include the environment that
holds information about the targets, and pheromone dissipation attributes. Data
flow diagrams, entity relationship diagrams, and Warnier-Orr figures were derived
as part of the ingredients for the proposed ontology. The study investigated the
validity of the ontology that arose from the study through a triangulated mix of
[Date] V
simulations, experimentation, and data visualisation. Precisely, three
experiments were administered to evaluate the usefulness of the proposed
ontology. Speed of emergence was the main metric for this assessment. Visual
emergent behaviour augmented the outcomes obtained from evaluating speed of
emergency. Also, assessment of the resource demand from using the ontology
sealed the evaluations. Results indicated that the proposed ontology captured
and represented the knowledge required by ant-bots to achieve path finding. The
ontology provided a knowledge representation approach for swarms of ant-like
robotic devices, providing a roadmap to the low-level behaviours of ant-bots
towards convergence. Representation of swarm knowledge in the form of an
ontology offers the potential to reshape the field of swarm robotics. Such formal
knowledge representation technique brings about effectiveness, adaptability, and
reliability in swarm systems.