maplab

Extensively tested on actual robotic systems and backed by numerous research publications, as evidenced by the citation list and real robot images in the README.

Designed as a modular and extensible framework, allowing easy integration and testing of new algorithms, highlighted in its philosophy as advancing research in visual-inertial mapping.

Capable of merging maps from multiple sessions and enabling collaborative mapping across robots, demonstrated with gifs and descriptions in the features section for large-scale and multi-robot operation.

Includes robust visual-inertial odometry, dense reconstruction, and optimization tools, providing a full pipeline from sensor data to 3D maps, as shown in the README's feature breakdown.