The Learning Robot Laboratory at CMU recently started distributing real-world data collected from it's mobile robot. The goal of this action is two-fold; to make robot data available to researchers without access to a mobile robot and to encourage cross comparison of different methods by running them on the same data. We present a description of our robot and the type of data it provides. We briefly describe a selection of utilities that have been created for reviewing this data and show how these data sets have been used in practice. As a result of our experience, we make some recommendations as to requirements for a common robot learning dataset made available over the Internet. Appendixes describe how to access the data and tools, and summarize our existing data sets.