Swarm Intelligence: Particle Swarm Kinematics, Ant Colony Stigmergy & Decentralized AI
Swarm Intelligence (SI) investigates the emergent collective problem-solving capabilities of decentralized, self-organized multi-agent systems. Without any central leader or global clock, simple autonomous agents interacting locally through environmental feedback solve complex global optimization problems.
1. Particle Swarm Optimization (PSO): Mathematical Kinematics
Developed by Kennedy and Eberhart (1995), PSO simulates bird flocking. Each particle i possesses a position x_i and velocity v_i in high-dimensional continuous search space, updated iteratively:
Controls exploration vs exploitation by scaling momentum from previous velocity vectors.
Pulls the particle toward its own historical personal best position (pbest_i).
Pulls the particle toward the global best position found by the entire collective swarm (gbest).
2. Ant Colony Optimization (ACO) & Chemical Stigmergy
Pioneered by Marco Dorigo (1992), ACO models foraging ants that coordinate via stigmergy—indirect communication through chemical pheromone deposits left in the environment:
Probabilistic Transition Rule
The probability of ant k moving from node i to node j:
Where τ_ij is pheromone intensity and η_ij = 1/d_ij is heuristic visibility.
Pheromone Evaporation & Deposition
Prevents premature convergence to sub-optimal local loops:
Where ρ ∈ (0, 1] is the evaporation rate ensuring exploration remains active over time.
