US: Researchers at the Massachusetts Institute of Technology and the National University of Ireland (NUI) have developed an algorithm to create highly detailed real-time indoor and outdoor 3D maps by using videos taken by low-cost cameras.
In robotic mapping, when a camera pans across a location, it invariably introduces slight errors in the estimated path taken. This is known as a ‘drift’. The researches shot some videos with a low-cost camera. When the camera returned to a position where it had already been, the algorithm determined which points within the 3D map to adjust, based on the camera’s previous poses. Then it stitched all the images together to “close the loop” to create a continuous 3D map in real-time. The researchers believe that their technique can be used to guide robots through potentially hazardous or unknown environments.