Work
SlimeHive — Swarm Robotics Platform
Project Type
Open-Source Swarm Intelligence Framework — Hardware + Software
Overview
SlimeHive is an open-source experimental framework for controlling and visualizing decentralized drone swarms. Inspired by slime mold (Physarum polycephalum) and ant colony optimization, it uses a "Pheromone Grid" to map the world, creating a shared memory that enables emergent behavior in simple agents.
Most drone swarms rely on precise GPS and rigid formation algorithms. SlimeHive goes the opposite direction — decentralized sensing, gravity physics positioning, and emergent intelligence from simple rules.
How It Works
The Biological Architecture
The system is modeled on biological components:
- Queen Brain (queen_brain.py): The cortex. Runs the physics loop, decays pheromones, manages state. Central processing but not central intelligence — the intelligence emerges from the swarm.
- The Ears (hive_ear.py): Distributed BLE sensor nodes that listen for drone signals and forward them via MQTT. Like the nervous system's input layer.
- The Eyes (dashboard_hud.py): Flask-based HUD visualizing the swarm in real-time with flight trails and "Ghost Fog" long-term history.
- The Scribe (hive_logger.py): Records every heartbeat and position to CSV for post-analysis.
Gravity Positioning (No GPS)
Instead of triangulation, position is calculated using a weighted "Gravity Model" where BLE sensor nodes pull the drone based on signal strength (RSSI). Robust and noise-tolerant — works indoors, underground, anywhere GPS fails.
Pheromone Grid
A 50×50 grid represents the shared memory of the swarm. Drones deposit "scent" at their location. Scent decays over time (configurable decay rate). Other drones follow scent gradients. Structures and paths emerge from these simple rules — nobody programs the path, it appears.
Hybrid Swarms
Real physical drones (Raspberry Pi Zero/Pico) and virtual simulated drones coexist in the same system. The Queen Brain doesn't distinguish between them. This allows testing algorithms with virtual agents before deploying to hardware.
Biomimetic Moods
The hive responds to environmental light sensors, shifting between behavioral states:
- SLEEP (Night) — conservation mode, minimal activity
- FRENZY (Day/Sunrise) — full exploration and activity
Technical Details
- Hardware: Raspberry Pi Zero 2W, Raspberry Pi Pico 2W, BLE beacons
- Communication: MQTT (Mosquitto broker), BLE mesh
- Processing: Python, numpy, scipy (convolution for pheromone diffusion)
- Visualization: Flask + real-time HUD with trail rendering
- Logging: CSV flight logs, JSON state snapshots, session archiving
- Configuration: JSON-based hive configuration for live parameter tuning
Connection to BSM Patent
SlimeHive is the working prototype of BSM Claim 4 — Autonomous Drone Habitat & Swarm Coordination. The same NLM (No Language Model) architecture that governs the BSM tree's drone flock is proven at prototype scale in SlimeHive. The consumer product version is AL (Artificial Life terrarium).
What Makes It Novel
- Gravity-based positioning using BLE RSSI — no GPS dependency
- Pheromone grid as collective swarm memory
- Hybrid real/virtual drone coexistence
- Biomimetic behavioral states tied to environmental conditions
- Open-source design — anyone can build compatible drones
Status
Working platform with physical hardware. Queen Brain, Ears, Eyes, and Scribe all operational. Dashboard HUD with real-time visualization. Session recording and replay capability. Web-based SlimeHive simulation version built for NLM 101 course.