IROS 2015 Paper Abstract


Paper ThAP.13

Tarapore, Danesh (University of York), Cully, Antoine (Université Pierre et Marie Curie (UPMC)), Mouret, Jean-Baptiste (Université Pierre et Marie Curie (UPMC))

Late Breaking Abstract: Intelligent Trial and Error for Damage Recovery with a Hexapod Robot and Single-Unit Pattern Generators (SUPGs)

Scheduled for presentation during the Poster session "Late Breaking Posters" (ThAP), Thursday, October 1, 2015, 09:45−10:00, Saal G1

2015 IEEE/RSJ International Conference on Intelligent Robots and Systems, Sept 28 - Oct 03, 2015, Congress Center Hamburg, Hamburg, Germany

This information is tentative and subject to change. Compiled on July 20, 2019

Keywords Robot Learning, Failure Detection and Recovery, Evolutionary Robotics


Autonomous robots operating in complex, natural environments have to adapt to unforeseen damages. In previous work, we developed an intelligent trial and error (ITE) algorithm that allows robots (tested on hexapod robot and multi-joint robotic arm) to adapt to damage in less than two minutes, without requiring self-diagnosis or pre-specified damage-recovery plans (Cully et al., Nature 2015). However, an open-loop oscillator was used to generate periodic signals for the hexapods gait, thus preventing the robot to rapidly adapt to sudden changes in its environment. Furthermore, robot movement was constrained by the specific type of periodic signal, prefixed a priori. In this poster paper, we present our recent results in the use of a type of neural network oscillator, the single-unit pattern generator (SUPG), that allows for the creative, open-ended encoding of gait signals, and a close-loop control. Our results demonstrate that the ITE+SUPG is able to rapidly discover compensatory gaits that work in spite of the damage to the physical robot. Additionally, the walking gaits appear dynamic, and seem to utilize the capabilities of the SUPG for unconstrained encoding of oscillatory signals.



Technical Content © IEEE Robotics & Automation Society

This site is protected by copyright and trademark laws under US and International law.
All rights reserved. © 2002-2019 PaperCept, Inc.
Page generated 2019-07-20  00:27:16 PST  Terms of use