Radish: The Robotics Data Set Repository

Introduction:

Radish is an American initiative created by Andrew Howard and Nick Roy and intended as a repository of standard data sets for the robotics community. It started in May 2003. The goal is to facilitate the development, evaluation and comparison of robotics algorithms. The current focus is clearly on localization and mapping, although they expected that Radish would ultimately expand to reflect the interests of the broader robotics community.

Researchers are invited to download and make use of the available data sets, and also to make their own contributions to the repository. New data sets can be submitted through a submissions page and there also exists a mailing list for subscribers to receive news and notification of new file uploads.

Any file format is acceptable as long as it is sufficiently documented, but the preferred format seems to be that of CARMEN, the Carnegie Mellon Robot Navigation Toolkit which is an open-source collection of software for mobile robot control that provides basic navigation primitives including: base and sensor control, logging, obstacle avoidance, localization, path planning, and mapping.

Some examples of available data sets at Radish

Presently, the Radish repository contains a collection of some 39 data sets falling into four categories (the figure shows a sample):

 

Some examples of available data sets at Radish

 

The only exception is a data set from the NASA Robonaut during human teleoperation to perform 45 grasps trials. The provided data contains readings from various force and tactile sensors on the arm and hand on Robonaut used for grasping. The data is in MATLAB format.

Although no information about downloads is available, there is evidence that suggests that Radish is not being very active, with the number of uploads per year (i.e. contributed data sets) consistently decreasing, with only a few persons contributing since Jan. 2005 and a mailing list made up of 11 people.

 

Further information:

More information can be found at the official website.