A collection of outdoor robotic datasets with centimeter-accuracy ground truth

Abstract

The lack of publicly accessible datasets with a reliable ground truth has prevented in the past a fair and coherent comparison of different methods proposed in the mobile robot Simultaneous Localization and Mapping (SLAM) literature. Providing such a ground truth becomes specially challenging in the case of visual SLAM, where the world model is 3-dimensional and the robot path is 6-dimensional. This work addresses both the practical and theoretical issues found while building a collection of six outdoor datasets. It is discussed how to estimate the 6-d vehicle path from readings of a set of three Real Time Kinematics (RTK) GPS receivers, as well as the associated uncertainty bounds that can be employed to evaluate the performance of SLAM methods. The vehicle was also equipped with several laser scanners, from which reference point clouds are built as a testbed for other algorithms such as segmentation or surface fitting. All the datasets, calibration information and associated software tools are available for download

Purposive Sample Consensus: A Paradigm for Model Fitting with Application to Visual Odometry

“…Málaga 2009 Robotic Dataset Collection PARKING-6L is used in this paper for featured based VO testing [20]. A section of the dataset is selected from the images captured by a camera mounted on a test vehicle which runs a closed loop in a car park.…”

ParallaxBA: bundle adjustment using parallax angle feature parametrization

“…Publicly available Malaga 2009 Robotic Dataset Collection (Blanco et al 2009) is used for demonstrating that the singularity issues also occur in a real-life situation. This dataset was collected using an electric car equipped with laser scanners, cameras, IMU, and GPS receivers.…”

A Multi-sensor Traffic Scene Dataset with Omnidirectional Video

“…Blanco et al present in [1] a dataset with a focus on "centimeter-accuracy ground truth" of position data for testing SLAM algorithms designed for small mobile robots. Their dataset is composed of synchronized data coming from multiple sensors (color cameras, laser scanners, different kinds of GPS receivers and IMU sensors) and provided in a structured, plain text file based format together with an API to handle the data.…”

“…The data is, however, taken at slow speeds from a steady and controlled environment such as a university campus and a parking lot. In a later extension of their work, data has been gathered also for a longer sequence taken in a urban scenario 1 . Their setup misses, however, the omnidirectional visual data that we consider of fundamental importance for full surround sensing.…”