27-08-2013, 04:28 PM
The New College Vision and Laser Data Set
College Vision and Laser.pdf (Size: 505.48 KB / Downloads: 132)
Abstract
In this paper we present a large dataset intended for use in mo-
bile robotics research. Gathered from a robot driving several kilo-
meters through a park and campus, it contains a five-degree-of-
freedom dead-reckoned trajectory, laser range/reflectance data and
20 Hz stereoscopic and omnidirectional imagery. All data is carefully
timestamped and all data logs are in human readable form with the
images in standard formats. We provide a set of tools to access the
data and detailed tagging and segmentations to facilitate its use.
Data Description
Data collection was performed using the vehicle shown in Fig-
ure 2. The drive unit is that of a RMP200 base from Segway.
This unit also provides the roll and pitch data recorded in the
odometry data stream. The data was gathered over a 2.2 km
traverse of the college grounds using the sensor configuration
described in Table 4. All data was logged using the MOOS
software infrastructure (Newman 2003).
Data Access Methods
The data is available for download from http://www.
robots.ox.ac.uk/NewCollegeData/. This site also goes into
much greater detail when describing the dataset than is appro-
priate for this paper. To avoid the requirement to download the
full 30 GB just to evaluate the data, the data is available com-
pressed as single events (periods of a few seconds) and chunks
of the entire dataset. All non-image data is stored in plain text
files with an “alog” suffix. Each line contains a single log en-
try which is a comma-separated list of token–value pairs. The
format of all log entries is supplied in detail on the website.
Supplied Tools
Ease of use is important and while having plain text logs re-
sults in OS-independence and a degree of immediacy, it has
the disadvantage that a parsing step is required to extract nu-
merical data. We have provided some tools to help with this
task and expedite data access.