Swarm robotics & programmable matter

From Claytronics vision to Kilobots, modular reconfigurable robots, particle robots, and 2024 surveys.

Story arc

1. Vision — Programmable Matter (Goldstein et al. 2005)

Goldstein, Campbell, Mowry, Programmable Matter, IEEE Computer
DOI  · Project page 

• Ensembles of tiny robots (“catoms”) form and reshape 3D objects.
• Goal is synthetic reality: physical artifacts that mimic shape, motion, appearance, touch — not teleportation of atoms.
• Hardware, algorithms, and power are co-equal challenges.
• Hub: CMU Claytronics 

Beginner takeaway: Clear, honest vision piece — aspirational north star.

2. Visible algorithmic self-assembly — Kilobots (2014)

Rubenstein, Cornejo, Nagpal, Programmable self-assembly in a thousand-robot swarm, Science
DOI  · Harvard news 

• 1,024 Kilobots form programmed 2D shapes using only local communication.
• Robots are simple, noisy, unreliable — collective algorithms still work.
• Seed robots define a coordinate origin; others edge-follow and fill the shape.
• Separates “smart global result” from “dumb local rules.”

Platform paper: Rubenstein, Ahler, Nagpal — DOI 
Designed for hundreds–thousands of units (~$14 parts originally); vibration locomotion + IR messaging.

Watch: Thousand-Kilobot demo  · Intro to Kilobot 

3. Modular reconfigurable robotics — Seo, Paik, Yim (2019)

Seo, Paik, Yim, Modular Reconfigurable Robotics, Annu. Rev. Control Robot. Auton. Syst.
DOI 

• Taxonomy: lattice, chain, mobile, hybrid.
• Promise: versatility, robustness, potentially lower cost vs fixed morphology.
• Hardware (connectors, actuators) and software (planning, control) are equally hard.
• Open bar: “Can it form a wrench that turns a bolt?”

4. Algorithms for self-reconfiguration — Thalamy et al. (2019)

Thalamy, Piranda, Bourgeois, A survey of autonomous self-reconfiguration methods…, Robotics and Autonomous Systems
DOI 

• Algorithm-first companion to hardware surveys.
• Current algorithms are far from visionary requirements (huge module counts, speed, reliability).
• Emphasizes distributed methods that scale.

5. Another philosophy — Particle robotics (2019)

Li, Batra, Brown, et al. (incl. Rus, Lipson), Particle robotics…, Nature
DOI  · Lab page 

• Loosely coupled units that only expand/contract.
• Collective locomotion emerges from statistical-mechanics-style coordination.
• Works even with ~20% malfunctioning particles in demos.
• Different from precise docking modular robots — embrace disorder + coupling.

6. 2024–2025 landscape

arXiv watch (verify before deep citation)

• Cubic modular robots with realistic constraints: arXiv HTML 
• Moving Matter (single robot rearranging blocks): arXiv HTML 
• MODUR modular dual-reconfigurable robot: arXiv HTML 

DNA sticky ends vs robot local messages

Practice on this track

• Experiment 5 — Kilobot / swarm simulation
• Experiment 8 — grid reconfiguration cost
• Experiment 1 or 7 — magnet / patchy physical analogues
• Week 2 Track B in Learn