Event

Proposes a resource allocation framework for heterogeneous multi-team altruistic collaboration

1 reports · 1 sources

Riwa Karam et al. propose a resource allocation framework for heterogeneous multi-team robot collaboration, treating robots as transferable resources, drawing on Hamilton's rule from ecology to design an altruistic decision-making mechanism, and using graph neural network policies to approximately solve altruistic allocation under a centralized training, decentralized execution setting. The authors prove that the generalized heterogeneous resource allocation problem is combinatorial optimization and NP-hard. The method is evaluated in a firefighting scenario against exact optimization and multi-step simulation, and five paired trials are conducted on a real-world platform with 15 robots, comparing learned reallocation with fixed team membership: the learned policy reached the firefighting threshold set by the model in all five trials, while fixed membership did so in only two.