Centralised Collaborative SLAM in the Lunar Environment: Implementation of a CSLAM method of mapping lunar topology using multiple rovers
Authors
Fabricius, Hector Gabriel ; Langfeldt, Marcus ; Nielsen, Carsten Harald Pinderup
Term
4. semester
Education
Publication year
2026
Submitted on
2026-06-04
Pages
64
Abstract
This thesis develops a centralized collaborative SLAM system (simultaneous localization and mapping, CSLAM) that enables two or more rovers on the lunar surface to determine their position and build a shared map. On the Moon, the terrain has few visual features and lighting can be extremely high-contrast, which makes camera-only methods unreliable. To address this, each rover combines LiDAR (a laser distance scanner), an RGB-D camera (color plus depth), and an IMU (motion sensors), with RTAB-Map as the main SLAM and map-merging software. In the centralized design, every rover runs local SLAM and periodically sends its map database to a base station. The base station merges all incoming maps into one global map and distributes the updated result back to the fleet. The system was built in ROS 2 Humble and containerized with Docker. It was first tested in an Isaac Sim lunar-analogue simulation, then on AAU Space Robotics’ rover Gorm at the Aalborg Portland Rørdal limestone quarry. To integrate the new sensors, Gorm was fitted with a soft motor starter, a USB hub, and an Ethernet switch. Field tests confirmed that each rover can map on its own and that maps from multiple rovers can be merged into a single coherent global map.
Dette projekt udvikler et centraliseret, samarbejdende SLAM-system (simultan lokalisation og kortlægning, CSLAM), som gør det muligt for to eller flere rovere på måneoverfladen at finde deres position og opbygge et fælles kort. På Månen har underlaget få visuelle detaljer, og lyset kan være ekstremt højkontrast, hvilket gør rene kamerabaserede metoder upålidelige. For at løse dette kombinerer hver rover LiDAR (en laserafstandsscanner), et RGB-D-kamera (farve plus dybde) og en IMU (bevægelsessensorer), med RTAB-Map som den centrale software til SLAM og sammenfletning af kort. I den centraliserede arkitektur kører hver rover lokal SLAM og sender med jævne mellemrum sin kortdatabase til en basestation. Basestationen fletter alle kort til ét globalt kort og sender den opdaterede version tilbage til flåden. Systemet er udviklet i ROS 2 Humble og containeriseret med Docker. Det blev først afprøvet i en måneanalog simulation i Isaac Sim og derefter testet på AAU Space Robotics’ rover Gorm i Aalborg Portland Rørdal kalkbrud. For at integrere den nye sensorsuite fik Gorm en blød motorstarter, en USB-hub og en Ethernet-switch. Feltforsøg bekræftede, at hver rover kan kortlægge selvstændigt, og at kort fra flere rovere kan sammenfletes til ét sammenhængende, globalt kort.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords
SLAM ; multi robot ; lunar ; mapping