Curriculum Guided Massive Multi Agent System Solving For Robust Long Horizon Tasks
This addresses the problem of computational inefficiency and instability in multi-agent systems for long-horizon tasks like robotic manipulation, though it appears incremental as it builds on existing curriculum and multi-agent methods.
The paper tackles long-horizon reasoning tasks in multi-agent systems by introducing a hierarchical architecture with a 64*64 grid of lightweight agents and a spatial curriculum, resulting in improved stability, reduced oracle usage, and stronger distributed reasoning on a Tower of Hanoi benchmark.
Large Language Models and multi-agent systems have shown promise in decomposing complex tasks, yet they struggle with long-horizon reasoning tasks and escalating computation cost. This work introduces a hierarchical multi-agent architecture that distributes reasoning across a 64*64 grid of lightweight agents, supported by a selective oracle. A spatial curriculum progressively expands the operational region of the grid, ensuring that agents master easier central tasks before tackling harder peripheral ones. To improve reliability, the system integrates Negative Log-Likelihood as a measure of confidence, allowing the curriculum to prioritize regions where agents are both accurate and well calibrated. A Thompson Sampling curriculum manager adaptively chooses training zones based on competence and NLL-driven reward signals. We evaluate the approach on a spatially grounded Tower of Hanoi benchmark, which mirrors the long-horizon structure of many robotic manipulation and planning tasks. Results demonstrate improved stability, reduced oracle usage, and stronger long-range reasoning from distributed agent cooperation.