An energy and memory-efficient distributed self-reconfiguration for modular sensor/robot networks

Abstract

Self-reconfiguration for mobile microrobots currently needs a positioning system and a map of the target shape. Traditional positioning solutions, such as GPS or multilateration are not applicable in the micro-world, and maps sharing does not scale. In the literature, if we want a self-reconfiguration of microrobots to a target shape that consists of millions of positions, each microrobot should have a memory capacity of at least million positions. Therefore, this is not scalable. In this paper, nodes do not record any position. We present self-reconfiguration methods where nodes are unaware of their positions and where they do not have the final coordinates of each microrobot. In other words, nodes do not store the coordinates that build the target shape. Therefore, memory usage for each node is hugely reduced to \(O(1)\) and communications are limited to neighboring nodes. These algorithms aim to improve the logical topology of a set of microrobots by restructuring their physical topology. To that end, we consider here the case of restructuring a set of microrobots from a chain to a square and we study two algorithms: the first algorithm ensures the connectivity of the network at the end of the algorithm, where the second guarantees the connectivity of the network through the execution time. The paper presents both analytical and experimental assessments of the algorithms performances using the declarative language \(Meld\) and executed under the Dynamic Physical Rendering Simulator (DPRSim).