Communication
Data & Control Flow
End-to-end flow between the web frontend, Python backend, ESP-NOW radio bridge, and the drones.
Low-budget platform for drone swarm research and education
Drone swarm research faces significant barriers that limit access for students and researchers.
Drone swarm research typically requires expensive commercial platforms and hardware, putting it out of reach for most academic labs.
Commercial motion-capture systems and specialized swarm platforms cost tens of thousands of dollars, limiting access to well-funded institutions.
Existing platforms rarely offer the flexibility to define and test custom swarm algorithms without deep embedded programming expertise.
Students and researchers need a low-cost, flexible, programmable platform to conduct meaningful swarm experiments without prohibitive investment.
The Drone Swarm Research Platform is a low-budget, programmable testbed that lets students and researchers run custom swarm algorithms on real hardware.
The platform separates high-level intelligence from drone execution. A central PC or server handles all the heavy lifting — localization, swarm state management, algorithm execution, and command generation. The drones act as physical agents, receiving motion commands and sending back telemetry.
This architecture makes the platform easy to extend: you write your swarm algorithm in a familiar language on the server, and the drones execute it in real time.
Write and upload your own swarm behavior; the platform executes it on live drones.
PC/server runs localization, state management, and command scheduling in one loop.
Drones focus on execution — receive commands, fly, send telemetry. No complex onboard logic required.
Monitor and control experiments through a browser-based interface — no special software needed.
Built from off-the-shelf components to keep the cost accessible for student labs and small research groups.
Five interconnected layers form the complete swarm research platform.
End-to-end flow between the web frontend, Python backend, ESP-NOW radio bridge, and the drones.
Top-mounted cameras provide a bird's-eye view of the entire flight arena with minimal occlusion.
Drones carry colored LEDs that are easily detectable by the cameras and distinguishable from one another and from the background.
Multiple cameras improve coverage and accuracy through triangulation and cross-validation.
Achieves indoor positioning without expensive commercial motion-capture systems.
System overview image
Place your diagram at docs/images/localization.png
Six modules that together form the server-side of the research platform.
Main coordination layer. Orchestrates all other modules and handles communication with drones and the web dashboard.
Stores the latest position and telemetry data for each drone. Provides a unified view of swarm state to all other modules.
Runs user-defined swarm algorithms each control-loop tick. Receives current state, produces target positions or velocities.
Converts algorithm outputs into drone commands. Handles command timing, sequencing, and priority to ensure safe execution.
Records drone positions, telemetry, issued commands, and experiment metadata. Data is available for post-experiment analysis.
Browser-based interface for monitoring and controlling experiments. No special client software required.
Off-the-shelf components chosen for affordability and availability.
Small, low-cost quad rotors as physical swarm agents
Onboard controller for stabilization and command execution
Wireless bridge — sender on PC and receiver on each drone, linked via ESP-NOW
Top-mounted USB cameras capturing the full flight arena
Runs localization, swarm engine, and dashboard backend
Visible-color LEDs on each drone for camera-based position detection
The tools and frameworks that power the platform.
What the platform will deliver upon completion.
Live position tracking of multiple drones with low-latency updates to the swarm state manager.
A centralized, deterministic control loop that closes the perception-action cycle for the full swarm.
User-defined swarm algorithms run in real time on live drone hardware without modifying firmware.
A browser-based interface for launching, monitoring, and stopping experiments from any device.
Structured logs of positions, telemetry, and commands for post-experiment analysis and reproducibility.
A working testbed built from affordable hardware, usable by student labs and small research groups.
Department of Computer Engineering, University of Peradeniya · E/21 Batch
