Reducing Communication Load on Contract Net by Case-Based Reasoning -- Extension with Directed Contract and Forgetting

Takuya Ohko, Kazuo Hiraki and Yuichiro Anzai, Keio University, Japan

This paper describes the communication load reduction on the task negotiation with Contract Net Protocol for multiple autonomous mobile robots. We have been developing LEMMING, a task negotiation system with low communication load for multiple autonomous mobile robots. For controlling multiple robots, Contract Net Protocol(CNP) is useful, but the broadcast of the Task Announcement messages on CNP tends to consume much communication load. So LEMMING learns proper addressees for the Task Announcement messages with Case-Based Reasoning so as to suppress the broadcast. The learning method is called Addressee Learning. In this paper, we extend LEMMING with "directed contract" to reduce the communication load more effectively. Moreover, we extend LEMMING with "forgetting" to restrict the number of cases, since it is impossible to have enough memory to keep all the cases. The efficiency of LEMMING is evaluated in a simulated multi-robot environment to show that these extension are effective for LEMMING.


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.