Herding Sheep: With or without a sheperd?

Keywords
No Thumbnail Available
Issue Date
2014-07-17
Language
en
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Creating an algorithm to simulate animal behaviour is useful in many ways. This thesis investigates the performance of a number of algorithms that simulate flock control with a flock of sheep, and sheepdog. A new algorithm is created to find out whether it performs more effective than the already existing algorithms in the simulator. The created algorithm differs from other algorithms by using a user-controlled shepherd and sheepdog to control the flock. The shepherd is able to give commands to the sheepdog to herd the sheep. The performance of the algorithms is tested in an experiment where the flock has to move from a certain position to a goal position through a specific route. By introducing the shepherd who commands the sheepdog, the algorithm offers the possibility to adapt to changing circumstances in the simulator. Results of the experiment show that the algorithms without shepherd perform faster on a shepherding task than the algorithm with a shepherd. However, the results also indicate that with more practice the algorithm with shepherd can outperform the algorithms without shepherd.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen