Drone navigation study reports successful flights with intermittent camera observations

Researchers report that a drone navigation system completed all 10 indoor flight trials while using intermittent depth observations from a single camera. Their preprint describes ALONE, a system that updates an internal estimate of nearby space as the robot moves and requests another observation when its confidence becomes insufficient for navigation. Across two simulated scene families, it achieved 98% and 97% success at a 10 Hz decision rate. Among successful trials, the median shares of decision steps requiring fresh depth observations were 0.9% and 1.3%, respectively. The approach could free a shared sensor for other tasks between navigation observations; physical testing covered 10 indoor flights.