Researchers at the Massachusetts Institute of Technology (MIT) and the National University of Ireland at Maynooth have written a mapping algorithm that creates real-time, detailed three-dimensional (3D) maps of indoor and outdoor environments.
Using videos taken with a Kinect camera of the halls and stairways of MIT’s Stata Center, the team used a mapping method to develop 3D maps. Significantly, the algorithm was able to quickly merge images to "close the loop" when the camera returned to its starting point, forming a continuous, realistic 3D map and solving a major problem in robotic mapping known as "loop closure" or "drift."
The problem occurs when a camera, for example, pans across a room and introduces slight errors in the estimated path taken, perhaps shifting a doorway to the right or elongating a wall. These errors compound over long distances, creating maps with walls and stairways that fail to line up.
The new mapping technique determines how to connect a map by tracking a camera's position in space throughout its route, and when a camera returns to a place it has already been, the algorithm determines which points within the 3D map to adjust. The technique could help guide robots through potentially hazardous or unknown environments.
From MIT News
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found