Academic
Publications
Metrics for the Evaluation of DisCSP: Some Experiments on Multi-robot Exploration

Metrics for the Evaluation of DisCSP: Some Experiments on Multi-robot Exploration,10.1109/WI-IAT.2010.71,Pierre Monier,Arnaud Doniec,Sylvain Piechowia

Metrics for the Evaluation of DisCSP: Some Experiments on Multi-robot Exploration   (Citations: 1)
BibTex | RIS | RefWorks Download
Many algorithms to solve Distributed Constraint Satisfaction Problems (DisCSP) have been introduced in the literature. In this paper, we propose to compare three different algorithms to solve DisCSP. Contrary to algorithms of the literature which are evaluated on graph coloring problems or uniform random binary DisCSPs, we use a multi-robot exploration problem. We show that, for this real world application, the comparison of algorithms may be improved by using additional metrics than those used in the literature. We will define other metrics that can be used for measuring different aspects of the multi-robot exploration problem. The aim of our attempt for defining metrics is to analyze and compare different aspects of complexity of this multi-robot problem. We will observe that using both classical and real world metrics is interesting to obtain a better and more precise comparison.
Conference: Web Intelligence - WI , pp. 370-373, 2010
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...• DBS [5, 6] (for Distributed Backtracking with Sessions) does not use a list of couples (variable, value) to manage the message context but only an integer called session...
    • ...We decided to use six criteria [5] to evaluate ABT , AW C and DBS algorithms: the length of the travelled path, the number of checked constraints, the number of cycles [11], the CPU time, the number of exchanged messages and the number of DCSP solved...
    • ...This is due to the dynamic priority in AWC .A t each step of the simulation, the highest priority agent, which impose its favourite direction to lower priority agents (see [5]) changes...
    • ...Others experiments were done and explained in [5] such as range of the numbers of agents or explored area by the highest priority agent...
    • ...We can see that it is not the case here because the dynamic priority order for agents is not adapted to this multi-agent exploration problem (see [5])...

    Pierre Monieret al. Comparison of DCSP Algorithms: A Case Study for Multi-agent Exploratio...

Sort by: