#Project title:"Networks and Swarms: The effect communication constraints in swarm robotics"
This project is created by Su,heng-ming. Any point views expressed in the dissertation are those of the author and dissertation may be made freely available immediately for academic purposes while it still needs citation to respect author's right.
In this paper, robot swarms based on three-valued model and distributed decision-making has been investigated. The project focuses on the different effect communication constraints in swarm robot. In addition, random network without evidence model and random network with evidence model has been compared. With both simulation experiments of models, the results show that random network with evidence model has appropriate evidence rate and error rate. Random network with evidence model is not only robust but also has faster speed of convergence.
There are two algorithm to help robot swarms reach the consensus and make the decision. Compare to "Practical Byzantine Fault Tolerance" the random network with evidence algorithm can decrease converence time as 200% and more robust to reach the consensus. It is believed to be use at the field as blockchain, cloud services and swarm intelligence. There are four factor to affect random network with evidence model algorithm. 1.Number of agents 2.possibility of connection 3.evidence rate 4.error rate. Each factor have difference impact about algorithm especially evidence rate and eror rate.