Simulation
Flocking Birds
A boids sketch where local steering rules, spatial lookup, and live tuning create convincing flock behavior in the browser.
Employer signal
What This Project Shows
This project shows that I can turn an algorithmic idea into an interactive system that stays readable while many independent agents are moving at once.
Problem
What Needed To Be Solved
A naive flocking simulation compares every boid with every other boid. That is easy to write, but it scales poorly and makes tuning feel random because performance noise hides the behavior of the rules.
Approach
How I Built The Solution
I kept the steering model explicit: separation, cohesion, and alignment are separate forces that can be inspected and weighted. For neighbor discovery, the sketch uses a quadtree query so each boid works with nearby agents instead of scanning the full population.
Outcome
What It Demonstrates
The result is a compact simulation that demonstrates agent-based modeling, performance-aware rendering, and the ability to make abstract rules understandable through motion.
Evidence From Source
Source signal
The old p5 source uses `qt.query(...)` for nearby boids and combines `cohesion_weight`, `alignment_weight`, and separation into each velocity update.
Engineering decision
Keeping the force terms visible makes this more useful than a black-box animation because the same structure can be tuned, debugged, or explained.