Distributed Shape Recognition Algorithm for Lattice-Based Modular Robots

Abstract

Lattice-based modular robots are composed of modules arranged on a lattice and forming 3D shapes. This paper presents a fully distributed algorithm executed by each module of a large lattice-based modular robot to determine their current shape using a small memory footprint and fast convergence. The algorithm consists of finding overlapping boxes that cover the entire robot configuration using message-passing. Thus, allowing robots to determine a representation of their current shape. The discovery of a distributed representation of the current shape of modular robots provides valuable information that can be either used to facilitate the distributed self-reconfiguration planning or to efficiently send the current shape to a connected computer. The algorithm is executed in a simulated environment and compared with exhaustive coordinates collection in order for modules to send their current global shape to a computer to be used by an interactive CAD software. The obtained results show the efficiency of our algorithm in detecting the current shape of the robot, while also outperforming the coordinates collection algorithm.