Taking inspiration from an old sea faring navigation using devices as astrolabe and sextant we’ve prepared a new project which aims to develop our own system for celestial navigation to complement available GPS systems and mainly to use it as an excuse to get our hands on some Machine Learning.
Following goals to be achieved:
- Prepare a training & testing set from 20 brightest stars – Achieved
- Train PyTorch to identify presented stars with 80%+ success rate – 93% Achieved
- Develop a code which will use identified stars + accurate local time to identify observer’s GPS coordinates with < 1 degree error.
Post being added as project develops:
TRAINING STAR IDENTIFICATION AI USING PYTORCH
TRAINING SET WITH STELLARIUM II
TRAINING SET WITH STELLARIUM
