Note: this dataset is now inactive. It was collected in 2011 and released in 2013. While we found it very useful, we have found that the various changes in the robot hardware and calibration as well as minor issues like a broken inter-computer time synchronisation lead to this dataset being less consistent than we would have liked. Please use it if interested, but please beware.

The MIT Stata Center Data Set is a vast scale data set collected over a multi-year period in a 10 storey academic building. It contains sensor data collected since January 2011. As of September 2012 the data set comprises over 2.3TB, 38 hours and 42 kilometres (the length of a marathon) . The dataset is of particular interest to robotics and computer vision researchers. The authors have specifically used this dataset to develop Visual SLAM algorithms, however it is expected to be useful in a wide-variety of other research areas - change detection in indoor environments, human pattern analysis and learning, long-term path planning.

Of particular importance, this dataset also includes ground truth position estimates of the robot at every instance (to typical accuracy of 2cm) using the as-built floorplans.

About the Building

The Ray and Maria Stata Center or Building 32 is a 720,000-square-foot (67,000 m2) academic complex designed by Pritzker Prize-winning architect Frank Gehry for the Massachusetts Institute of Technology (MIT).

Wikipedia Entry

About the PR2

The PR2 is a two-armed wheeled robot designed by Willow Garage. In June 2010 the robot was loaned to MIT (along with 10 other universities). We proposed mapping the building using it's multiple LIDAR sensors and cameras.

About the PR2

About the Collection

Starting in January 2011, researchers at MIT explored the Stata Center with the PR2 and logged LIDAR, stereo camera, odometry and other system information. The robot was teleoperated for all the data presented here, otherwise the dataset represents typical cluttered indoor environments.

Dataset Summary