Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Accessible Hardware Implementation for Multi-Agent Collective Construction
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes MACC-SLIM, a low-cost 2D hardware-style simulation of multi-agent collective construction that runs TERMES-style plans on Ozobot robots over an LCD screen, and uses it to expose synchronization delays in optimal MILP plans.
Problem
Multi-agent collective construction (MACC) research requires expensive, custom TERMES robots and large lab space, limiting access to real hardware implementation.
Approach
Replace custom TERMES robots with commodity line-following Ozobots driven by color-coded circle instructions shown on a flat LCD screen, which maps the 3D TERMES action set onto a 2D screen-based platform; border-cell queue counts handle agent entry and exit.
Key results
- Demonstrates 24 SLIM agents (Ozobots on an LCD screen) executing an optimal MILP-generated plan for MACC benchmark instance 1 in a demonstration video.
- Shows non-insignificant synchronization delays in execution of optimal plans, since agent movement durations vary with rotation and border entry, which current models do not capture.
Surprising findings
- Even optimal state-of-the-art MACC plans incur noticeable agent waiting time for synchronization due to varying movement durations.
Limitations
- Only one benchmark instance (24 agents) demonstrated; no quantitative evaluation across instances.
- Enter and leave actions with block reservoir interaction and multiple agents through one border cell are not fully handled.
- Pick-up and attach block actions are displayed but not physically executed by agents.
Experimental setup
Datasets: MACC benchmark instance 1
Benchmarks: MACC benchmark instance 1
Metrics: synchronization delays makespan
Builds on: TERMES Lam et al. 2020 exact MACC approaches Rameš and Surynek 2024 action duration generalization
Closest papers by meaning
- Debate2Create: Robot Co-design via Multi-Agent LLM Debate (ICML 2026 · similarity 0.83)
- MAS$^2$: Self-Generative, Self-Configuring, Self-Rectifying Multi-Agent Systems (ICLR 2026 · similarity 0.81)
- Multi-Agent Motion Planning for Differential Drive Robots Through Stationary State Search (AAAI 2025 · similarity 0.81)
- EMOS: Embodiment-aware Heterogeneous Multi-robot Operating System with LLM Agents (ICLR 2025 · similarity 0.81)
- Mosaic: Runtime-Efficient Multi-Agent Embodied Planning (ICML 2026 · similarity 0.81)
- Simulation-Informed Diffusion for Decentralized Multi-robot Motion Planning (NeurIPS 2026 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.