Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.641615
Title: DRAMA, a connectionist model for robot learning : experiments on learning a synthetic language by initiation in autonomous robots
Author: Billard, A.
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 1999
Availability of Full Text:
Full text unavailable from EThOS. Please contact the current institution’s library for further details.
Abstract:
The present dissertation addresses problems related to robot learning from demonstration. It presents the mechanism for learning a synthetic language taught by an external teacher agent. This thesis considers three main issues: 1) learning of spatio-temporal invariance in a dynamic noisy environment, 2) symbol grounding of a robot's actions and perceptions, 3) development of a common symbolic representation of the world by heterogeneous agents. We build our approach on the assumption that grounding of symbolic communication creates constraints not only on the cognitive capabilities of the agent but particularly on its behavioural capacities. Behavioural skills, which allow the agent to coordinate its actions with those of the teacher, are required on top of general cognitive abilities of associativity, in order to provide an attentional mechanism. The agent's attention is then constrained to make relevant perceptions, onto which it grounds the teacher's words. In addition, the agent should be provided with cognitive abilities for extracting spatial and temporal invariance in the continuous flow of its perceptions. Based on these requirements, we develop a connectionist architecture, which provides the agent with the capacity to imitate and to learn spatio-temporal regularities. The model is a Dynamical Recurrent Associative Memory Architecture, called DRAMA. It is a fully connected recurrent neural network using Hebbian update rules. Learning is dynamic and unsupervised. The performance of the architecture is analysed theoretically and in numerical simulations, as well as in physical and simulated robotic experiments. Training of the network is computationally fast and cheap, which allows its implementation for real time computation and on-line learning in a computationally limited hardware system. Robotic experiments are carried out with different learning tasks, namely landmark recognition and prediction of perception-action sequences.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.641615  DOI: Not available
Share: