I'm pretty excited to see this project released. This is from the researchers who made LSD-SLAM[0], a widely used monocular SLAM system, and DSO is showing some

I'm pretty excited to see this project released. This is from the researchers who made LSD-SLAM[0], a widely used monocular SLAM system, and DSO is showing some very impressive results in terms of accuracy, robustness and efficiency. For convenience, here are a few more resources related to this release: * Video demo (with ORB-SLAM and LSD-SLAM comparisons): https://www.youtube.com/watch?v=C6-xwSOOdqQ https://www.youtube.com/watch?v=C6-xwSOOdqQ * DSO research paper: https://arxiv.org/abs/1607.02565 https://arxiv.org/abs/1607.02565 * A small ROS wrapper for DSO was also released yesterday: https://github.com/JakobEngel/dso_ros https://github.com/JakobEngel/dso_ros It'd be interesting to try and integrate DSO as a visual "sensor" for use with Google Cartographer [1]. This might allow people without a LIDAR or stereo/RGB-D/FOV cameras to perform fairly accurate and robust SLAM that's quite possibly efficient enough to run in realtime on a smartphone (although the rolling shutter cameras usually found in smartphones might complicate that a bit). [0] http://vision.in.tum.de/research/vslam/lsdslam http://vision.in.tum.de/research/vslam/lsdslam [1] https://github.com/googlecartographer https://github.com/googlecartographer