Geospatial AI
Lunar Surface Navigation
Turning Chandrayaan-2 observations into safer paths across the lunar south pole.
The problem
Planning a route across the Moon’s south pole requires aligning different observations, identifying terrain hazards, and finding a navigable path through an unfamiliar landscape.
Features
- Terrain hazard detection from aligned lunar datasets
- Route planning with PRM-RL
- Interactive Three.js rover simulation
My contribution
Analyzed Chandrayaan-2 datasets, compared hazard-detection and path-planning methods, and developed an interactive 3D rover simulation to visualize planned routes.
The approach
01
Understand the terrain
Aligned datasets through georeferencing and explored YOLOv8, Digital Terrain Models, and Hough Transform with Canny edge detection for hazard identification. The latter produced the best results within the project comparison.
02
Find a navigable path
Compared more than five path-planning algorithms and selected PRM-RL based on its performance in the project evaluation.
03
Make the data explorable
Used the Imaging Infrared Spectrometer’s 3 μm hydration feature to identify potential water-ice sites, and built a Three.js rover simulation to communicate the planned route.
The outcome
The project earned first runner-up in its problem statement at ISRO Bharatiya Antariksh Hackathon 2025. The simulation illustrates route planning; it is not a deployed rover system.
Next exploration