Project Firefly is a novel localization method using a webcam to calculate the position of a LED fixture. It combines a fixed array of LEDs to with closed-loop LED control to reliably calculate position in 3D. Everything is open source.
An example application is provided, which navigates a small rover through a sequence of waypoints.
A technical breakdown is available on my Youtube channel.
Connects to rovers and executes waypoint mission. Communicates with other scripts over ZMQ. Needs DataProc_ZMQ to be running in parallel for processing Run python3 navigate.py cal to set up waypoint mission by moving the rover by hand through each desired position.
Processes localization data from navigate.py. Needs to be run in a Linux enviroment, I use WSL. To work with WSL update the nameserver at the bottom of roverConfig.py. The localization system is written in C++ and is wrapped using SWIG using the Makefile. It may be able to run on Windows as well with some tweaks.
Plots waypoints and rover motion over time.
Records and processes motion history. Finds relationship between rover commands and actual motion and sends improved system back to navigate.py.
Firefly uses the default OpenCV2 camera calibration system. Start by printing off calibrationCheckerBoard.pdf and taping it to something flat. Run calibrateCamera.py with the camera connected. Take several images of the checkerboard at different angles by pressing space. Press escape to process and save the calibration data. This is saved to cameraCal.yaml for later use.
I would recommend building the wooden rover. It is laser cut out of 3mm material, and the files are available in /CAD. The wiring information can be found on page 56 of FireFly_Technical_Report.pdf. We use Platformio to flash the Arduino, although the default IDE should work as well.