ROVR is releasing an open-source dataset of synchronized camera and LiDAR clips to make real-world 3D data widely accessible for spatial AI, robotics, and autonomous driving. It covers highways, cities, and rural roads in day, night, and rain with tightly aligned sensors and clean calibration so models can learn from motion rather than single frames. Systems that look great on older datasets often stumble here, which shows how narrow training can hide real-world weaknesses. Training on ROVR h...
