In autonomous robotics, there is still a trend to develop and tune controllers with highly explicit goals and environments in mind. However, this tuning means that these robotic models often lack the developmental and behavioral flexibility seen in biological organisms. The lack of flexibility in these controllers leaves the robot vulnerable to changes in environmental condition. Whereby any environmental change may lead to the behaviors of the robots becoming unsuitable or even dangerous. In this manuscript we look at a potential biologically plausible mechanism which may be used in robotic controllers in order to allow them to adapt to different environments. This mechanism consists of a hormone driven epigenetic mechanism which regulates a robot's internal environment in relation to its current environmental conditions. As we will show in our early chapters, this epigenetic mechanism allows an autonomous robot to rapidly adapt to a range of different environmental conditions. This adaption is achieved without the need for any explicit knowledge of the environment. Allowing a single architecture to adapt to a range of challenges and develop unique behaviors. In later chapters however, we find that this mechanism not only allows for regulation of short term behavior, but also long development. Here we show how this system permits a robot to develop in a way that is suitable for its current environment. Further during this developmental process we notice similarities to infant development, along with acquisition of unplanned skills and abilities. The unplanned developments appears to leads to the emergence of unplanned potential cognitive abilities such as object permanence, which we assess using a range of different real world tests.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:712772 |
| Date | January 2017 |
| Creators | Lones, John |
| Publisher | University of Hertfordshire |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hdl.handle.net/2299/17859 |
Page generated in 0.0019 seconds