For Engineers

Build Robots That Think Like Ants

You Build Things That Work

Algorithms are nice. Proofs are elegant. But you care about real systems.

Systems that handle noise. Systems that fail gracefully. Systems that scale.

This is your challenge: make stigmergy work in the physical world.

The Digital System

We’ve built a colony that works:

• 101 agents on Agentverse
• TypeDB Cloud for shared state
• STAN algorithm for coordination
• Real mission: hunting a $700K Bitcoin puzzle

It works. But it’s digital.

The Physical Challenge

Can robots implement stigmergy?

The Problems

1. Pheromone Deposition Digital agents write to a database. Physical robots can’t.

Options:

• Chemical deposits (actual pheromones)
• RFID tags embedded in environment
• Light patterns (UV, IR)
• Virtual pheromone (shared database via mesh network)
• Sound/ultrasonic markers

2. Pheromone Sensing Digital agents query a database. Physical robots need sensors.

Options:

• Chemical sensors (actual pheromone detection)
• RFID readers
• Light sensors
• Localization + database lookup
• Acoustic sensors

3. Pheromone Decay Digital pheromone decays via algorithm. Physical markers don’t auto-decay.

Options:

• Time-based electronic decay
• Evaporating chemicals
• Fading light sources
• Active removal robots (“cleaners”)

4. Latency and Noise Digital agents see instant, perfect pheromone. Physical sensors are noisy and delayed.

Options:

• Filtering algorithms
• Redundant sensors
• Conservative thresholds
• Probabilistic models

Robot Swarm Design

Minimal Viable Swarm

┌─────────────────────────────────────────────────────────────────┐
│                      ROBOT SPECIFICATIONS                        │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  Hardware per robot:                                             │
│  ─────────────────────                                           │
│  • Microcontroller (ESP32 or similar)                           │
│  • Motor controller + wheels                                     │
│  • Pheromone sensor (TBD)                                        │
│  • Pheromone depositor (TBD)                                     │
│  • Short-range communication (Bluetooth/Zigbee)                 │
│  • Battery                                                       │
│  • Optional: GPS/localization                                    │
│                                                                  │
│  Software per robot:                                             │
│  ─────────────────────                                           │
│  • STAN algorithm (adapted for physical constraints)             │
│  • Sensor processing                                             │
│  • Motor control                                                 │
│  • Optional: Mesh networking                                     │
│                                                                  │
│  Swarm requirements:                                             │
│  ─────────────────────                                           │
│  • 5-20 robots for proof of concept                              │
│  • Arena with trackable paths                                    │
│  • Pheromone infrastructure                                      │
│  • Observation/logging system                                    │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

Virtual Pheromone Option

Don’t deposit physical pheromone. Use mesh network + shared database:

Robot → Publishes location + deposit to mesh → Database updates → Other robots read

Pros: Simpler, cleaner, more controllable Cons: Requires reliable networking, not “true” stigmergy

Hybrid Option

Physical markers + digital augmentation:

• Robots deposit RFID tags or light beacons
• Central system tracks and decays virtual pheromone
• Robots query system for local pheromone levels

Engineering Questions

1. Sensor Design

What’s the best way to sense pheromone in physical space?

2. Deposit Mechanism

What’s the most reliable way to leave markers?

3. Robustness

How does the swarm handle robot failures? Communication loss?

4. Scalability

What limits scale? How do we push past those limits?

5. Energy Efficiency

Stigmergy should reduce communication overhead. Does it reduce energy use?

Hardware Acceleration

For the Bitcoin puzzle hunt, we need fast elliptic curve operations.

Challenge: Can you accelerate STAN on:

• GPU (CUDA, OpenCL)
• FPGA
• Custom ASIC (design only)
• Specialized chips (TPU, IPU)

Current benchmark:

• CPU: ~1M ops/sec
• RTX 3080: ~500M ops/sec
• What can you achieve?

What We Provide

Codebase

• Complete Python implementation
• STAN algorithm (easily portable)
• Agent behaviors
• TypeDB integration

Documentation

• Algorithm specifications
• Data formats
• API documentation

Support

• Hardware advice from robotics faculty
• Algorithm support from project team
• Testing infrastructure for virtual pheromone

Funding

• Component budget for hackathon projects
• Access to lab equipment (if local)
• Post-hackathon development funding for promising projects

Hackathon Challenges for Engineers

Challenge: Design a Pheromone Robot

Design a robot that can deposit and sense pheromones.

Deliverables:

• Hardware specification
• Bill of materials
• CAD files (optional)
• Sensing/depositing strategy

Prize: $1,000 for best design

Challenge: Build a Virtual Pheromone Swarm

Implement stigmergy using mesh networking.

Deliverables:

• Working prototype (simulation or physical)
• Mesh networking implementation
• STAN adaptation

Challenge: Hardware Acceleration

Accelerate STAN for GPU/FPGA.

Deliverables:

• Optimized implementation
• Benchmark results
• Speedup analysis

Prize bonus: $500 for 10x speedup

Challenge: Sensor Array Design

Design a sensor system for pheromone detection.

Deliverables:

• Sensor selection
• Array design
• Signal processing pipeline
• Noise analysis

Your Heroes Built Real Systems

Rodney Brooks built robots that worked in the real world by abandoning classical AI.

Vijay Kumar built swarms that coordinate through simple rules.

Raffaello D’Andrea made drones dance. Could they forage?

James McLurkin built practical swarm robots for warehouses.

Publication/Impact Opportunities

Why Engineering?

Everyone else is theorizing. You’re building.

Biologists describe behavior. Mathematicians prove theorems. Economists model incentives.

You make things work.

If stigmergy is going to change the world—warehouses, drones, exploration—engineers will do it.

Register Your Team

[REGISTER NOW]

Include at least one non-engineering team member (we recommend CS or Biology).

The best systems are informed by theory and biology.

“In theory, there is no difference between theory and practice. In practice, there is.”

— Yogi Berra (often misattributed)

You’ve built robots, drones, and IoT systems.

Now build a swarm that thinks.

[JOIN THE HACKATHON]