Domestic Rooms Dataset

Description

The Domestic Rooms (DR) Dataset contains robot observations of several rooms of the domestic environment. It provides a large scale testing environment for room classification approaches based on data obtained with a laser range finder.

Data acquisition

The following equipment has been used for data acquisition:

  • Robot: Pioneer P3-DX
  • Laser range-finder: Hokuyo URG
    • Sensor range: 5.6 meters
    • Field of view: 240 degrees
    • Sensor position on the robot: 30 cm above the floor
  • Odometry sensor

The robot was guided for a short tour through each room using joystick, at speeds between 0.1 and 0.5 m/s, while odometry readings and laser range data were being recorded with the frequency of 10 frames per second.

Scanned rooms

24 different homes have been scanned, consisting of houses and apartments from 4 different cities. There are a total of 90 rooms in the dataset, consisting of:

  • 21 living rooms
  • 6 corridors
  • 35 bathrooms
  • 28 bedrooms

Some examples of scans:

Ground plan images of a few example rooms, generated using a SLAM algorithm:

File organization

There are four directories, where each corresponds to one room type. In each directory there are as many files as there are examples of rooms that have been scanned for the corresponding room type. Therefore, each file contains all the data that has been gathered in a single room. Rows of the file represent consecutive observations, while each contains the following information:

  • Timestamp
  • Room type
  • Odometry x
  • Odometry y
  • Odometry phi
  • Start angle
  • Angle step
  • Number of beams
  • Min range
  • Max range
  • Laser scans

Download link

References

  • Peter Ursic, Matej Kristan, Danijel Skocaj, Ales Leonardis, Room Classification using a Hierarchical Representation of Space. IEEE/RSJ International Conference on Intelligent Robots and Systems, Portugal, October 2012