In my research paper, I examine how a collaborative autonomous Multi-Robot-System in an intralogistic environment can be designed and implemented so that it is practically viable. Robot collaboration, the act of completing an order with at least two robots being involved in fulfillment, is a so far unused practice in intralogistics. However, it has high potential for intralogistics as a range of new strategies can be explored that may be more efficient than traditional single-robot task fulfillment. To further explore the research question, Linus Väthröder, a fellow student, and I developed a proof-of concept system at the TUM Chair of Materials Handling, Material Flow, Logistics. The system development was approached by designing a concept for two robots where one of them goes to pick up the warehouse item, moves it to an exchange location where it is handed over to the second robot which then delivers it to the drop off location. All the mentioned locations were defined by an actor over a Graphical User Interface. The process was then coordinated by a central server that communicated with the GUI and the robots. Naturally, the robots should be assigned tasks in the most efficient way so that the process takes as little time as possible. For that, an optimization-based task allocation algorithm was used computing costs for both possible task distributions. In addition to that, the general process had to be efficiently coordinated to achieve robot collaboration without errors.

Implementation was carried out using two mobile robots of the chair as well as JavaScript Node.js. We could demonstrate a functional system in our tests, though we encountered a few obstacles along the way. Several conclusions were drawn from the results, one of the most important being that time coordination is a challenging problem in collaborative Multi-Robot Systems, as minor uncertainties can lead to significant malfunctions during the execution.