Swarm experiments

Here is a small collection of experiments that I made in a Swarm Intelligence course at NTNU in Ålesund.

Arrival behavior

Two Seek implementations

The red one is smarter. Refresh the page multiple times to get new random scenarios.

Pack hunting 1

With single target per hunter and improved Pursue implementation. Suffers a bit from conflicts due to the (necessary) separation behavior.

Pack hunting 2

Experimental, using combined behaviors, weighted target points, PID controllers and online optimization. It suffers a bit from PID being more suited for docking than pursuing.

Simple PSO searching in a 2D plane

With interactive modification of PSO parameters

PSO dissonance

Experiment with PSO parameters sweeping over their range at frequencies related by the golden ratio, and respawning particles upon convergence to global best known.

PSO_Path

A simplistic short-sighted path planning using PSO, with interactive controls

PSO_Path_optimized

A variant of the above that uses the last PSO state in previous position as starting point in the new position and runs only a single iteration per decision, compensating with a more particles and larger space (still only selects a direction without building a search tree)