- Introduction
- Features
- Technologies Used
- Usage
- Scientific Stops
- Pathfinding
- Visualization
- Images
- Contributors
Welcome to the Radicals Lunar Exploration Project, developed for the Bharatiya Antariksh Hackathon 2024. Our mission is to explore the lunar South Pole using advanced image processing, crater and boulder detection, and pathfinding algorithms. With the help of cutting-edge technologies such as YOLOv5 and the Ant Colony Optimization (ACO) system, we aim to assist future moon missions in optimizing traversal routes while avoiding obstacles.
- 🌑 Georeferenced OHRC Imagery: Reprojects Orbiter High-Resolution Camera (OHRC) images for better precision.
- 🟡 Crater & Boulder Detection: Parallel processing through YOLOv5 and traditional computer vision techniques like Hough Transforms.
- 🐜 Ant Colony Optimization: Innovative ACO-based pathfinding to traverse lunar surfaces efficiently, avoiding craters and boulders.
- 🛰️ Simulated Rover Path Visualization: Pathfinding and rover movement simulation using Unity3D on Digital Terrain Model (DTM) data.
- Python: Used for georeferencing, detection, and pathfinding.
- GDAL: For georeferencing OHRC images.
- YOLOv5: For crater and boulder detection.
- Ant Colony Optimization (ACO): To optimize pathfinding.
- Unity3D: Simulating rover movement on the lunar surface.
- QGIS: For terrain data extraction.
- OHRC images are georeferenced using GDAL and reprojected to the Lunar South Polar Stereographic CRS. The calibrated images are clipped to focus on landing sites.
- After preprocessing (normalization, denoising), images are fed into two parallel detection pipelines: YOLOv5 and traditional computer vision methods.
We implemented two approaches for detection:
- Approach A: Parallel processing using YOLOv5 and traditional methods (Hough Transform, Blob Detection).
- Approach B: Separate YOLOv5 and traditional CV models for flexibility between accuracy and speed.
During the mission, we have a total of 10 scientific stops, with significant analysis at each stop:
- Initial Landing Site Characterization at coordinates 85.36161° S, 31.41594° E.
- Surface Composition: 85.36162° S, 31.41252° E.
- Navigation & Mobility Test Site.
Our ACO algorithm optimizes the rover's path by avoiding obstacles like craters and boulders. The process involves pheromone-based pathfinding:
- Ants search for food and return to the colony, leaving pheromone trails that guide others. We simulate this behavior in ACS to identify optimal rover paths.
- Distance: Changes based on elevation.
- Speed: Adjusted based on terrain steepness.
- Heuristic Value: Dynamically adjusted for optimal path calculation.
We used Unity3D to simulate the rover's movements based on the optimized traversal path. The rover starts at the Landing Site and navigates through the 10 scientific stops on the lunar surface.
- Extract terrain mesh using QGIS from DTM/DEM data.
- Import the mesh into Unity.
- Load the optimal path from the ACO results into the Unity script.
- Simulate rover traversal across 10 scientific stops.
Here are some sample visualizations from the project:
Initial landing site for the rover at coordinates 85.36161° S.
Detected craters and boulders using YOLOv5 and traditional CV methods.
ACO-based optimized path on lunar terrain, considering elevation and obstacles.
Unity3D simulation showing the rover traversing the optimal path.
This project is licensed under the MIT License - see the LICENSE file for details.