Title

The Distributed Simulation Of Intelligent Terrain Exploration

Abstract

In this study we consider the coordinated exploration of an unfamiliar Martian landscape by a swarm of small autonomous rovers, called Swarmies, simulated in a distributed setting. With a sustainable program of return missions to and from Mars in mind, the goal of said exploration is to efficiently prospect the terrain for water meant to be gathered and then utilized in the production of rocket fuel. The rovers are tasked with relaying relevant data to a home base that is responsible for maintaining a mining schedule for an arbitrarily large group of rovers extracting water-rich regolith. For this reason, it is crucial that the participants maintain a wireless connection with one another and with the base throughout the entire process. We describe the architecture of our simulation which is composed of HLA-compliant components that are visualized via the Distributed Observer Network tool developed by NASA. Additionally, a well-known terrain exploration algorithm, which takes the constraint of a mobile ad hoc network into account, is summarized and then extended by using a trainable genetic algorithm to determine the movement of the robotic swarm at every time step of the simulation. The integration of this extended algorithm into the distributed simulation is discussed and the empirical results of a comparison between the original and extended versions are given. Our results suggest that the genetic algorithm serves as a useful aid in the simulation of coordinated exploration and provides a layer of flexibility, offered by the trainable parameters its fitness function depends upon, that allows for the introduction of new constraints while maintaining compatibility with dynamic shifts in priority.

Publication Date

10-30-2018

Publication Title

SAE Technical Papers

Volume

2018-November

Issue

November

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

DOI Link

https://doi.org/10.4271/2018-01-1915

Socpus ID

85057321272 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/85057321272

This document is currently not available here.

Share

COinS