Hardware accelerator for evolutionary robotics

  • Authors:
  • Masaya Yoshikawa;Hidekazu Terai

  • Affiliations:
  • Department of Information Engineering, Meijo University, Nagoya, Japan;Department of VLSI System Design, Ritsumeikan University, Shiga, Japan

  • Venue:
  • WSEAS Transactions on Circuits and Systems
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Evolution robotics is one of the autonomous learning and its approaches have been applied to various fields. In this paper, we discuss new evolutionary robotics technique. The proposed evolutionary robotics acquires their behavior using genetic-based machine learning. It adopts new if-then rules for learning algorithm. Moreover, it introduces novel hardware accelerator in order to reduce simulation time for calculation, since the genetics-based machine learning requires very long computational time. Experiments using quasi-ecosystem demonstrate not only the effectiveness of the proposed learning algorithm, but also that of the proposed hardware accelerator.