Adjustably Autonomous Robotic Agents

Jason Garforth, Sue McHale, and Anthony Meehan

In seeking to explore architectures that help manage capture errors we attempt to identify how humans might help the autonomous robot recover from the capture error (without, of course, resorting to repair). We have drawn upon the model of Norman and Shallice which was developed to explain how people control attention and avoid (frequent) capture errors. This model proposes architecture for understanding (neurologically-based) agent systems which are capable of 'high level behaviours which require planning or decision making, involve trouble shooting, are ill-learned or contain novel sequences, are dangerous or technically difficult, require overcoming a strong habitual response.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.