| EPSRC Reference: |
GR/J18125/01 |
| Title: |
EVOLUTION OF CONTROL ARCHITECTURES FOR VISUALLY GUIDED AUTONOMOUS AGENTS |
| Principal Investigator: |
Husbands, Professor P |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
Computing and Artificial Intelligence |
| Organisation: |
University of Sussex |
| Scheme: |
Standard Research (Pre-FEC) |
| Starts: |
01 October 1993 |
Ends: |
28 February 1997 |
Value (£): |
200,855
|
| EPSRC Research Topic Classifications: |
|
| EPSRC Industrial Sector Classifications: |
|
| Related Grants: |
|
| Panel History: |
|
|
Summary on Grant Application Form |
|
1) Use e specially developed robotic apparatus to explore the artificial evolution of control systems, based on continuous dynamic neural nets, for a sighted autonomous agent engaged in progressively more complex navigational tasks.2) Evaluate and analyse the results obtained.3) Investigate the ways in which this study may throw light on natural evolution.Progress:A specialised 'gantry' robot, allowing automatic evaluation of control systems in the real word, has been constructed. It has been successfully used to concurrently evolve control networks and visual morphologies for robots engaged in simple visually guided tasks.Using small converged populations, an incremental evolutionary process successfully developed robots capable of performing the following series of progressively harder tasks: approaching a large stationary target from a random position; approaching a small stationary target from a random position; following a moving target; distinguishing between two differently shaped targets. These were all achieved with extremely minimal vision and very small networks. More complex navigational tasks are currently being studied.Complimentary simulation work has studied coevolutionary predator-prey dynamics. It has also been shown that it is possible to evolve networks using a carefully constructed simulation which generate almost identical behaviours in the real robot. Various mathematical and visualisation tools have been developed to analyse the results of these experiments.A key issue in this research is the artificial genotype to phenotype mapping (artificial morphogenesis). A number of different schemes have been developed and tested and this has become a very active area of research. Several PhD students and visiting research fellows have become attached to this project. This has allowed more in-depth studies of certain aspects of the work, and the opening up of new directions, such as direct evolution of hardware. A large number of conference and journal papers have been produced, details of which can be obtained from the principal investigator.
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
|
| Further Information: |
|
| Organisation Website: |
http://www.sussex.ac.uk |