Towards Useful 3D Semantic Maps
|Radu B. Rusu||Aritificial Intelligence Center, SRI International||[Home Page]|
Date: Tuesday December 11, 2007 at 16:00
Location: EJ228 (SRI E building) (Directions)
Learning useful representations of 3d environments and automatically segmenting objects of interest is a challenging task. For the last decade, the area of robot mapping has been struggling to simultaneously localize and build maps of robots as they move in unknown environments. With a few exceptions, consistent point cloud maps can be acquired in a large variety of scenarios. On the other hand, people in robot planning have been developing algorithms that can operate robots, with the assumption that the underlying world model is known.
Our work is driven by the fact that a new area of research has been spawn, which connects the two, in the sense that meaningful representations have to be carefully extracted and higher level semantics discovered in the acquired world map, in order to be useful for higher level planning.
We investigate the problem of acquiring 3D semantic maps in a variety of scenarios, with a strong emphasis on our indoor distributed ubiquitous sensing project (the AwareKitchen). The objects modeled in these maps include cabinets, tables, drawers and shelves, and objects in the environment that have particular relevance for a household robotic assistant. The mapping approach is based on point cloud representations, and a set of sophisticated and robust interpretation methods operating on these representations, which lie at the intersection of various research fields like: computer graphics, statistics, photogrammetry and sensing, robotics, and optimization theory.
Several example applications for interpreting data coming from indoor, outdoor, and aerial scans, digitized models, and human motion segmentation are given as well.
Please arrive at least 10 minutes early in order to sign in and be escorted to the conference room. SRI is located at 333 Ravenswood Avenue in Menlo Park. Visitors may park in the visitors lot in front of Building E, and should follow the instructions by the lobby phone to be escorted to the meeting room. Detailed directions to SRI, as well as maps, are available from the Visiting AIC web page.