University Of Pretoria Computer Science Department

MSc Presentation: Jade Abbott

Posted by mriekert on Mon 29 Jan 2018, 11:04:29 Mon 29 Jan 2018, 11:04:29

When: 1 February 2018 @ 13:00
Venue: IT 4-58
Topic: Nature Inspired Algorithms for Prioritized Foraging

Abstract: Foraging is a major problem in swarm robotics, which has been applied to many areas such as agriculture, and search and rescue. This dissertation defines a variation of swarm robotics foraging, called prioritized foraging. Prioritized foraging differs from other foraging problems, in that there are two types of items: prioritized items and non-prioritized items. Furthermore, a novel honey bee inspired foraging algorithm is developed. An empirical analysis of three foraging algorithms (a na¨?ve algorithm, and two nature inspired algorithms, namely a desert ant inspired algorithm and the novel honey bee inspired algorithm) is performed on the prioritized foraging problem. The analysis investigates each algorithm’s performance on the prioritized foraging problem in terms of the major swarm robotics characteristics of efficiency, scalability, flexibility, and robustness as well as the behaviours that enable those characteristics. The work concludes that the honey bee algorithm is highly efficient, highly flexible, highly scalable in terms of problem density, and most robust in terms of redundancy due to the algorithm’s division of labour between prioritized and non-prioritized items. The desert ant algorithm is almost as efficient as the honey bee algorithm. The na¨?ve algorithm has very poor efficiency, compared to the other algorithms, but was the most scalable in terms of swarm size. Both the desert ant and na¨?ve algorithm experience poor flexibility, scalability in terms of problem density, and robustness in terms of redundancy.

All content copyright © Department of Computer Science, School of IT, University of Pretoria, South Africa