Ubuntu 14.04, ROS indigo	Ubuntu 16.04, ROS kinetic

• Load camera poses from different formats (such as PIX4D, COLMAP)
• Generates a dense point cloud from raw images, camera poses and camera intrinsics
• Generates Digital Surface Models (DSMs) from raw point clouds and exports e.g. to GeoTiff format
• Different methods to generate (ortho-)mosaics from raw images, camera poses and camera intrinsics

• aerial_mapper: Meta package
• aerial_mapper_demos: Sample executables.
• aerial_mapper_dense_pcl: Dense point cloud generation using planar rectification.
• aerial_mapper_dsm: Digitial Surface Map/Model generation.
• aerial_mapper_google_maps_api: Wrapper package for Google Maps API.
• aerial_mapper_grid_map: Wrapper package for grid_map.
• aerial_mapper_io: Input/Output handler that reads/writes poses, intrinsics, point clouds, GeoTiffs etc.
• aerial_mapper_ortho: Different methods for (ortho-)mosaic generation.
• aerial_mapper_thirdparty: Package containing thirdparty code.
• aerial_mapper_utils: Package for common utility functions.

• How to install
• How to run (+ sample dataset)
• How to contribute

Dense point cloud (from virtual stereo pair, 2 images)	Digital Surface Map (DSM, exported as GeoTiff)	(Ortho-)Mosaic (from homography, 249 images)

Raw images	Dense point cloud	Digital Surface Map
Observation Angle (red: nadir)	Grid-based Orthomosaic (Cell resolution: 0.5m)	Textured DSM

If you use this work in an academic context, please cite the following publication:

T. Hinzmann, J. L. Schönberger, M. Pollefeys, and R. Siegwart, "Mapping on the Fly: Real-time 3D Dense Reconstruction, Digital Surface Map and Incremental Orthomosaic Generation for Unmanned Aerial Vehicles" [PDF]

@INPROCEEDINGS{fsr_hinzmann_2017,
   Author = {T. Hinzmann, J. L. Schönberger, M. Pollefeys, and R. Siegwart},
   Title = {Mapping on the Fly: Real-time 3D Dense Reconstruction, Digital Surface Map and Incremental Orthomosaic Generation for Unmanned Aerial Vehicles},
   Booktitle = {Field and Service Robotics - Results of the 11th International Conference},
   Year = {2017}
}

This work was partially funded by the European FP7 project SHERPA (FP7-600958) and the Federal office armasuisse Science and Technology under project number 050-45. Furthermore, the authors wish to thank Lucas P. Teixeira from the Vision for Robotics Lab at ETH Zurich for sharing scripts that bridge the gap between Blender and Gazebo.