← Back to Projects
Lost Target Search
Modified Particle Swarm Optimization — visualized in the browser
Overview
A web-based simulation and visualization of a Modified Particle Swarm Optimization (MPSO) algorithm applied to the lost target search problem. The goal is to coordinate multiple search agents to locate a lost target as efficiently as possible — using swarm intelligence rather than exhaustive grid search. The simulation runs in the browser with a Python/Flask backend driving the algorithm.
Stack
| Layer | Technology |
|---|---|
| Backend | Python / Flask |
| Frontend | JavaScript, HTML, CSS |
| Algorithm | Modified Particle Swarm Optimization (MPSO) |
| Simulation | MATLAB-style logic (ported to JS) |
| Deployment | Render |
How It Works
The simulation models a swarm of search agents deployed over a map area. Each agent moves according to the MPSO update rules — influenced by its own best known position, the swarm's global best, and a randomised component that prevents premature convergence.
- —Agents update velocity and position each iteration based on personal and global best positions
- —The modification to standard PSO improves convergence behaviour in search-and-rescue style scenarios
- —The full simulation is visualized in real time in the browser — three panels showing agent positions, search coverage, and convergence