Following Leaders in Byzantine Multirobot Systems by Using Blockchain Technology

【Author】 Ferrer, Eduardo Castello; Jimenez, Ernesto; Luis Lopez-Presa, Jose; Martin-Rueda, Javier

【Source】IEEE TRANSACTIONS ON ROBOTICS

【影响因子】6.835

【Abstract】Interest in multirobot systems is rising rapidly both in academia and in industry. The use of multiple robots working in a coordinated way, rather than a single robot, has several advantages in a diverse range of applications. However, few studies have focused on operations in which robots can behave maliciously and alter the outcome of the collective mission. In this article, we present a set of Byzantine Follow The Leader (BFTL) problems, in which a subset of robots in the system shows unintended or inconsistent behavior (i.e., Byzantine robots). In the BFTL problems, leaders discover routes from their starting positions to specific destinations and guide the followers, while Byzantine robots try to hinder the leaders and mislead the followers. In this research, blockchain technology is used as a communication tool within multirobot systems, for leaders to broadcast directions to the whole group. We propose algorithms to tackle the BFTL problems, prove their correctness, and validate them in simulated experiments of realistic scenarios. Results show that the proposed algorithms mitigate the impact of Byzantine robots in multirobot systems conducting a BFTL mission. Our analysis provides minimum and maximum boundary calculations for important metrics including number of robots reaching their destination, number of steps taken, and weight requirements of the chain used during the mission. Our results provide a path toward the deployment of byzantine-resistant real-world multirobot systems.

【Keywords】Robots; Blockchains; Robot sensing systems; Robot kinematics; Multi-robot systems; Tools; Computer crashes; Blockchain-aided robotics; byzantine fault-tolerance; distributed systems; multirobot systems; resilient robotics; verifiable robotics

【发表时间】

【收录时间】2022-01-01

【文献类型】

【主题类别】

--