MLFC: From 10 to 50 Planners in the Multi-Agent Programming Contest

In this paper, we describe the strategies used by our team, MLFC, that led us to achieve 2\(^{nd}\) place in 15\(^{th}\) edition of Multi-Agent Programming Contest. The scenario contest is an extension previous (14\(^{th}\)) “Agents Assemble” wherein two teams agents move around a 2D grid and compete assemble complex block structures. We discuss languages tools during development team. Then, summarise main were carried over from participation 14\(^{th}\) list limitations (if any) using these latest edition. also developed new made specifically for extended scenario: cartography (determining size map); formal verification map merging protocol (to provide assurances it works when increasing number agents); plan cache (efficiently scaling planners); task achievement (forming groups tasks); bullies (agents focus on stopping opposing team). Finally, give brief overview performance what believe shortcomings.