Byzantine Fault Detection in Swarm-SLAM Using Blockchain and Geometric Constraints

Effective methods for Simultaneous Localisation And Mapping (SLAM) are key to enabling autonomous robots to navigate unknown environments. Multi-robot collaborative SLAM (C-SLAM) offers the opportunity for higher performance thanks to parallel execution of mapping and localisation by a distributed team of robots but it also introduces challenges in system scalability and consistent data aggregation, exposing the system to potential security risks. In particular, we show that the state-of-the-art decentralised C-SLAM framework for swarm robotics is vulnerable to Byzantine robots, which are robots that behave incorrectly, possibly due to malfunctioning or hacking. We propose a solution that uses a blockchain to achieve data consistency and a smart contract that manages robots’ reputations to identify and neutralise Byzantine robots. Each robot’s contribution to collaborative mapping is peer-reviewed by other robots by verifying its correctness through geometric constraints. Our multi-robot simulation results show the existence of a trade-off between fault tolerance and efficiency in terms of map generation speed. With this work, we also release open-source research software that interfaces a custom blockchain with the ROS 2 framework.