Autonomous Vehicle Sensor Suite Data with Ground Truth Trajectories for Algorithm Development and Evaluation

This paper describes a multi-sensor data set, suitable for testing algorithms to detect and track pedestrians and cyclists, with an autonomous vehicle\u27s sensor suite. The data set can be used to evaluate the benefit of fused sensing algorithms, and provides ground truth trajectories of pedestrians, cyclists, and other vehicles for objective evaluation of track accuracy. One of the principal bottlenecks for sensing and perception algorithm development is the ability to evaluate tracking algorithms against ground truth data. By ground truth we mean independent knowledge of the position, size, speed, heading, and class of objects of interest in complex operational environments. Our goal was to execute a data collection campaign at an urban test track in which trajectories of moving objects of interest are measured with auxiliary instrumentation, in conjunction with several autonomous vehicles (AV) with a full sensor suite of radar, lidar, and cameras. Multiple autonomous vehicles collected measurements in a variety of scenarios designed to incorporate real world interactions of vehicles with bicyclists and pedestrians. Trajectory data for a set of bicyclists and pedestrians was collected by separate means. In most cases, the real-time kinetic receivers on the bicyclists and pedestrians achieve RTK (Real Time Kinematic)-fixed, or RTK-float accuracy, resulting in errors on the order of a few centimeters, or a few decimeters, respectively; position accuracy on the instrumented interaction vehicles is on the order of 10 cm. We describe the data collection campaign at the University of Michigan\u27s Mcity Test Facility for connected and automated vehicles, the interaction scenarios and test conditions, and will show some visualizations of the test as well as initial evaluation results. These data will serve as a global-frame, multi sensor/multi actor canonical dataset which can be used for the development and evaluation of extended-object tracking algorithms for autonomous vehicles