In this library there are four key elements or groups of elements:
Sensor observations: All sensor observations share a common virtual base class (mrpt::slam::CObservation). There are classes to store laser scanners, 3D range images, monocular and stereo images, GPS data, odometry, etc. A concept very related to observations is a mrpt::slam::CSensoryFrame, a set of observations which were collected approximately at the same instant.
Rawlogs (datasets): A robotics dataset can be loaded, edited and explored by means of the class mrpt::slam::CRawlog. See also: http://www.mrpt.org/Rawlog_Format
Actions: For convenience in many Bayesian filtering algorithms, robot actions (like 2D displacement characterized by an odometry increment) can be represented by means of "actions". See mrpt::slam::CAction.
"Simple maps": In MRPT, a "simple map" is a set of pairs: "position", "sensory frames" (read above). The advantage of maintaining such a "simple map" instead a metric map is that the metric maps can be rebuilt when needed with different parameters from the raw observations, which are never lost.
| Page generated by Doxygen 1.6.1 for MRPT 0.9.0 SVN: at Mon Jun 7 06:47:58 UTC 2010 |
