SPICE: Swarm Production Implemented in a re-Configurable Environment
Authors
Andersen, Jonas ; Clement, Anders ; Lauritsen, Oliver Beyer
Term
4. semester
Education
Publication year
2023
Submitted on
2023-06-01
Pages
78
Abstract
This work takes the first steps toward implementing the novel production paradigm of swarm production—an approach where many mobile robots coordinate to move parts and supply workstations. The system includes 5 PolyBot mobile robots acting as carrier robots for material handling, 4 virtual workstations for part processing, and a central swarm manager that orchestrates operations. We designed and tested three traffic management strategies—rules for routing, prioritizing, and queuing robots at stations—including a new concept called dynamic queue positions, where waiting spots are adjusted on the fly based on current conditions. Tests show that strategies that consider the temporal dimension (such as predicted arrival times and the timing of potential conflicts) are better suited to swarm production, consistent with solutions from multi-agent path finding and pickup-and-delivery research. Dynamic queue positions were also found to potentially improve the system’s scalability. Future work will scale up the number of robots and further iterate on the system.
Dette arbejde tager de første skridt mod at implementere den nye produktionsparadigme sværmproduktion—en tilgang, hvor mange mobile robotter samarbejder om at flytte emner og forsyne arbejdsstationer. Systemet består af 5 PolyBot-mobilerobotter, der fungerer som bærerobotter til materialehåndtering, 4 virtuelle arbejdsstationer til emnebearbejdning og en central sværm-manager, der koordinerer helheden. Der er udviklet og testet tre forskellige strategier for trafikstyring—dvs. hvordan robotter rutes, prioriteres og køer ved stationerne—herunder et nyt koncept med dynamiske køpositioner, hvor ventepositioner tilpasses løbende efter situationen. Testene viser, at strategier, der inddrager den tidslige dimension (fx forventede ankomsttidspunkter og tidsmæssige konflikter), er bedre egnet til sværmproduktion, i tråd med løsninger kendt fra multi-agent ruteplanlægning og afhentning-og-levering. Desuden kan dynamiske køpositioner potentielt forbedre systemets skalerbarhed. Fremtidigt arbejde omfatter at øge antallet af robotter og videreudvikle systemet.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords