Datasets that feed robot vision and SLAM.
Ten million 3D objects for populating simulators and training vision.
