Neural Boids: Using Small Neural Networks to Simulate Natural Flocking Behavior
By
ecto
Toasted golden, schmeared with insight. Top of the rack.
Summary
The article introduces 'noids' (neural boids), which are flocking agents controlled by small neural networks instead of hand-written rules. The author explains how real birds flock in murmurations without leaders or choreography, then contrasts this with traditional 'boids' simulation models that use explicit rules. The article describes how neural networks can learn flocking behavior through training, creating more natural and emergent behaviors than rule-based systems. The neural network approach uses only 1,922 parameters per agent and allows for complex, coordinated movements that mimic real bird flocking dynamics.
Key quotes
· 5 pulledI'm calling them noids - neural boids. No hand-written rules. A small neural network takes in what each agent can see and outputs a steering force.
A starling murmuration can have 300,000 birds. No leader. No choreography. No bird knows the shape of the flock.
Those aren't boids. I'm calling them noids - neural boids. No hand-written rules.
And yet the whole mass moves as one, rippling and folding, evading a falcon in coordinated waves that propagate faster than any single bird can fly.
1,922 learned parameters. That's the entire program!
You might also wanna read
Werld: Open-Source Artificial Life Simulation with Evolving Neural Network Agents
Werld is an open-source artificial life simulation project on GitHub where agents with NEAT neural networks evolve in a computational ecosys
Reflections on DwarfStar 4's rapid rise in local AI inference
The author reflects on the unexpected popularity of DwarfStar 4 (DS4), a local AI inference project. They attribute its success to the conve
Reflections on DwarfStar 4's rapid rise in local AI inference
The author reflects on the unexpected popularity of DwarfStar 4 (DS4), a local AI inference project. They attribute its success to the conve
Building a Personal AI Agent with Markdown-Based Skills and Local Models
The article describes a personal AI agent built on Pi that manages the author's inbox, calendar, deal pipeline, blog publishing, and researc
StepFun Releases Step 3.5 Flash: 196B Sparse MoE Model for OpenClaw Agents
StepFun has released Step 3.5 Flash, a 196B sparse Mixture of Experts (MoE) model that activates only 11B parameters per token for high effi
Anthropic Releases Claude Opus 4.7 AI Model with 1M Context Window and Enhanced Coding Capabilities
Anthropic announces Claude Opus 4.7, their latest AI model featuring a hybrid reasoning architecture with a 1 million token context window.
