Many open source software packages have been developed at the Personal Robotics Lab in support of our research. This page describes our core packages, and also includes documentation on how to setup a development environment to use them.
Most of our public software is hosted on GitHub.
This dataset consists of high-definition RGB-D images for testing vision algorithms for manipulation. The entire 16 GB tarball is available here. There is data for 5 objects - Fuze Bottle, Hi-C Lemonade, Pop Tarts, Pop Cereal and Rice Pilaf. For each object, there are sets of images for varying environmental conditions - normal lighting, dim lighting, clutter with normal lighting and clutter with dim lighting. Each unit set for a particular condition in turn has images corresponding to 3 ranges of distance from the camera - near, medium and far. The raw data includes RGB images, aligned depth images, and intermediate data for unaligned depth images and mapping masks. We have also provided accurate ground truth pose estimates for validation, and detailed meshes for each object. Example code for accessing the dataset is available in a repository on the lab's Github.
We host a Debian Apt package repository for the LTS versions of Ubuntu (currently Trusty Tahr). To add this repository to your system, run the following commands.
$ sudo sh -c 'echo "deb http://packages.personalrobotics.ri.cmu.edu/public trusty main" > /etc/apt/sources.list.d/personalrobotics.list' $ curl https://www.personalrobotics.ri.cmu.edu/files/personalrobotics.gpg | sudo apt-key add - $ sudo apt-get update