Mosaic: Runtime-Efficient Multi-Agent Embodied Planning
2026-07-10 • Multiagent Systems
Multiagent Systems
AI summaryⓘ
The authors studied how to speed up planning for teams of AI agents using large language models. They found that mistakes from losing track of the environment and poorly coordinated actions cause delays. To fix this, they created a system called Mosaic that keeps a simple but accurate memory of objects and uses mathematical optimization to coordinate actions efficiently. In tests, Mosaic made planning faster, reduced unnecessary steps, and improved success rates. Their work shows that good memory and careful coordination are key to faster multi-agent planning.
Large Language ModelsMulti-Agent SystemsEmbodied PlanningPartial ObservabilityState TrackingSemantic MemoryInteger Linear ProgrammingAI2-THORSearch-and-Rescue RoboticsCoordination Constraints
Authors
Kunjal Panchal, Saayan Mitra, Sunav Choudhary, Victor Bursztyn, Somdeb Sarkhel, Hui Guan
Abstract
LLM-based multi-agent embodied planning remains impractical due to prohibitively high execution latency. We identify failed actions as the dominant bottleneck, stemming from two core challenges: inaccurate state tracking under partial observability and inefficient coordination that produces redundant or conflicting actions. We introduce Mosaic, a runtime-efficient multi-agent planning framework that addresses both challenges. Mosaic maintains accurate yet lightweight state tracking through agent-centric semantic memory that stores objects in relative coordinates, enabling geometric transformations and coordination. It ensures efficient coordination through Integer Linear Programming that allocates actions at every planning step, enforcing physical feasibility and inter-agent coordination constraints. Across AI2-THOR and search-and-rescue benchmarks, Mosaic achieves 27-32% faster execution, 30-33% fewer LLM calls, 25-31% fewer steps, and 4-10% points higher success rates. These results demonstrate that efficient memory and constraint-guided coordination are critical for scalable, low-latency multi-agent planning.