Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Exploring the impact of novelty and objective-directed evolution in company with MAP-Elites and HyperNEAT
Collective robotics refers to the field of robotics that focuses on the coordination and collaboration of multiple agents to perform a task or solve a problem. The ability to automatically design controllers for such agents in a collective system is an attractive proposition. In this thesis we inves...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | Eng |
| Published: |
Department of Computer Science
2025
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613213867114496 |
|---|---|
| access_status_str | Open Access |
| author | Breytenbach, Jeremy |
| author2 | Nitschke, Geoff Stuart |
| author_browse | Breytenbach, Jeremy Nitschke, Geoff Stuart |
| author_facet | Nitschke, Geoff Stuart Breytenbach, Jeremy |
| author_sort | Breytenbach, Jeremy |
| collection | Thesis |
| description | Collective robotics refers to the field of robotics that focuses on the coordination and collaboration of multiple agents to perform a task or solve a problem. The ability to automatically design controllers for such agents in a collective system is an attractive proposition. In this thesis we investigate the impact on performance of combining MAP-Elites with HyperNEAT while varying the evolutionary search directive between an objective and non-objective search, and a hybrid approach. Objective search refers to evolutionary algorithms that explicitly optimize a predetermined performance metric, whereas nonobjective search refers to evolutionary approaches that primarily focuses on exploration and diversity within the search space. HyperNEAT is an evolutionary method that makes use of indirect encoding to evolve agents. Whereas in typical evolutionary methods, only the fittest agents survive to future generations, the inclusion of MAP-Elites allows not only the fittest agents but also those that demonstrate unique behaviour to survive (the elites). MAP-Elites is referred to as an illumination algorithm because by retaining these elite agents in the population, we expect to increase the chances of exploring and thus illuminating novel, yet potentially high-performing regions of the search space. To evaluate these methods, we use Keep-away, a simulated collective robotics task within the RoboCup football framework as a case study. In Keep-away, a team of "keeper" robots attempt to maintain possession of the football while opposing "taker" robots try to intercept it. For this study, we produced controllers for the keeper agents. This research report sheds light on how the combination of these methods affects the agents' performance and their ability explore the behaviour search space. The insights gained from this study will be valuable for researchers working to understand the value and applicability of combining illumination algorithms such as MAP-Elites with objective and non-objective search for gaining performance in Keep-away and similar tasks. |
| format | Thesis |
| id | oai:open.uct.ac.za:11427/41491 |
| institution | University of Cape Town (South Africa) |
| language | Eng |
| last_indexed | 2026-06-10T12:32:34.479Z |
| license_str | Not specified — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from UCTD — University of Cape Town Open Access Repository |
| publishDate | 2025 |
| publishDateRange | 2025 |
| publishDateSort | 2025 |
| publisher | Department of Computer Science |
| publisherStr | Department of Computer Science |
| record_format | dspace |
| source_str | UCTD — University of Cape Town Open Access Repository |
| spelling | oai:open.uct.ac.za:11427/41491 Exploring the impact of novelty and objective-directed evolution in company with MAP-Elites and HyperNEAT Breytenbach, Jeremy Nitschke, Geoff Stuart Computer Science Collective robotics refers to the field of robotics that focuses on the coordination and collaboration of multiple agents to perform a task or solve a problem. The ability to automatically design controllers for such agents in a collective system is an attractive proposition. In this thesis we investigate the impact on performance of combining MAP-Elites with HyperNEAT while varying the evolutionary search directive between an objective and non-objective search, and a hybrid approach. Objective search refers to evolutionary algorithms that explicitly optimize a predetermined performance metric, whereas nonobjective search refers to evolutionary approaches that primarily focuses on exploration and diversity within the search space. HyperNEAT is an evolutionary method that makes use of indirect encoding to evolve agents. Whereas in typical evolutionary methods, only the fittest agents survive to future generations, the inclusion of MAP-Elites allows not only the fittest agents but also those that demonstrate unique behaviour to survive (the elites). MAP-Elites is referred to as an illumination algorithm because by retaining these elite agents in the population, we expect to increase the chances of exploring and thus illuminating novel, yet potentially high-performing regions of the search space. To evaluate these methods, we use Keep-away, a simulated collective robotics task within the RoboCup football framework as a case study. In Keep-away, a team of "keeper" robots attempt to maintain possession of the football while opposing "taker" robots try to intercept it. For this study, we produced controllers for the keeper agents. This research report sheds light on how the combination of these methods affects the agents' performance and their ability explore the behaviour search space. The insights gained from this study will be valuable for researchers working to understand the value and applicability of combining illumination algorithms such as MAP-Elites with objective and non-objective search for gaining performance in Keep-away and similar tasks. 2025-06-25T13:18:38Z 2025-06-25T13:18:38Z 2025 2025-06-25T13:15:22Z Thesis / Dissertation Masters MSc http://hdl.handle.net/11427/41491 Eng application/pdf Department of Computer Science Faculty of Science University of Cape town |
| spellingShingle | Computer Science Breytenbach, Jeremy Exploring the impact of novelty and objective-directed evolution in company with MAP-Elites and HyperNEAT |
| thesis_degree_str | Master's |
| title | Exploring the impact of novelty and objective-directed evolution in company with MAP-Elites and HyperNEAT |
| title_full | Exploring the impact of novelty and objective-directed evolution in company with MAP-Elites and HyperNEAT |
| title_fullStr | Exploring the impact of novelty and objective-directed evolution in company with MAP-Elites and HyperNEAT |
| title_full_unstemmed | Exploring the impact of novelty and objective-directed evolution in company with MAP-Elites and HyperNEAT |
| title_short | Exploring the impact of novelty and objective-directed evolution in company with MAP-Elites and HyperNEAT |
| title_sort | exploring the impact of novelty and objective directed evolution in company with map elites and hyperneat |
| topic | Computer Science |
| url | http://hdl.handle.net/11427/41491 |
| work_keys_str_mv | AT breytenbachjeremy exploringtheimpactofnoveltyandobjectivedirectedevolutionincompanywithmapelitesandhyperneat |