Abstract
This paper describes a highly modular hierarchical behavior-based control system for robots. Key features of the architecture include: easy addition/removal of behaviors, easy addition of specialized behaviors, easy to program hierarchical structure, and ability to execute nonconflicting behaviors in parallel. The architecture uses a unique reward based combinator to arbitrate amongst competing behaviors such as to maximize reward. This behavior system was successfully used in our Sony Legged League entry in RoboCup 2000 where we came in third losing only a single game. © 2002 Springer-Verlag Berlin Heidelberg.
Cite
CITATION STYLE
Lenser, S., Bruce, J., & Veloso, M. (2002). A modular hierarchical behavior-based architecture. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 2377 LNAI, pp. 423–428). Springer Verlag. https://doi.org/10.1007/3-540-45603-1_54
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
