Davis: LiDAR-Vison Campus Dataset

Yash Turkar, Karthik Dantu
Timothy Chase Jr, Kartikeya Singh, Pranay Meshram, Christo Aluckal, Charuvahan Adhivarahan
Center for Embodied Autonomy and Robotics (CEAR)
University at Buffalo
Paper (Coming Soon) Code (Coming Soon) Dataset (Coming Soon)

Davis consists of multiple sequences of indoor and outdoor environments collected at the University at Buffalo. Figure shows ground truth point cloud constructed using a Robotic Total Station and a LiDAR. The dataset is collected using a Boston Dynamics Spot robot equipped with a Ouster OS1-128 LiDAR, a ZED stereo camera, and an IMU. The dataset includes ground truth trajectories using GPS for outdoor sequences while using Aruco markers for indoor sequences. The dataset is collected in various weather conditions and lighting conditions to provide a diverse set of data for training and testing.

Abstract

We introduce Davis, a comprehensive LiDAR-Vison Campus Dataset, collected at the University at Buffalo using a Boston Dynamics Spot robot equipped with an Ouster OS1-128 LiDAR, ZED stereo camera, and IMU. The dataset includes ground truth point-clouds of the environment (building) collected using a Robotic Total Station and a LiDAR, as well as ground truth trajectories using GPS for outdoor sequences and Aruco markers for indoor sequences. To provide a diverse set of data, we have collected the dataset in various weather conditions and lighting conditions, with each day's data collected at different times of the day over multiple consecutive days. Additionally, the dataset includes all internal data from the Spot robot, such as camera images, IMU data, joint information, and more, making it a rich resource for training and testing various robot perception tasks.

BibTeX

@inproceedings{turkar2024davis,
            title={Davis: LiDAR-Vison Campus Dataset},
            author={Yash Turkar, Kartikeya Singh, Pranay Meshram, Christo Aluckal, Charuvahan Adhivarahan, Karthik Dantu},
            year={2024}
          }