Reaching New Heights in Multi-Agent Collective Construction
Martin Rameš, Pavel Surynek
TL;DR
The paper addresses multi-agent collective construction (MACC) by introducing ReRamp, a reversible ramp-based planning approach that trades a path-footprint for a tree-footprint to support higher structures within polynomial time. ReRamp defines ramps as central paths with reversible side ramps and uses the Ramp from Tree (RFT) primitive plus an add_edge operation to grow ramps and incorporate target columns, achieving scalable construction plans. Theoretical results establish NP-hardness for unrestricted ramp use via a Hamiltonian-circuit construction, but show that ReRamp attains a practical height bound of $\Omega(\sqrt{n})$ when reversibility and side ramps are employed, yielding polynomial-time planning. Empirical evaluations against exact MILP/CP solvers and 3D decomposition demonstrate substantial speedups while producing competitive plans, including a discretized single-story house built in under 4 hours with a 24-agent plan, illustrating real-world feasibility and potential for distributed multi-agent construction.
Abstract
We propose a new approach for multi-agent collective construction, based on the idea of reversible ramps. Our ReRamp algorithm utilizes reversible side-ramps to generate construction plans for ramped block structures higher and larger than was previously possible using state-of-the-art planning algorithms, given the same building area. We compare the ReRamp algorithm to similar state-of-the-art algorithms on a set of benchmark instances, where we demonstrate its superior computational speed. We also establish in our experiments that the ReRamp algorithm is capable of generating plans for a single-story house, an important milestone on the road to real-world multi-agent construction applications.
