openDD: A Large-Scale Roundabout Drone Dataset

Abstract

Analyzing and predicting the traffic scene around the ego vehicle has been one of the key challenges in autonomous driving. Datasets including the trajectories of all road users present in a scene, as well as the underlying road topology are invaluable to analyze the behavior of the various traffic participants. The interaction between the traffic participants is especially high in intersection types that are not regulated by traffic lights, the most common one being the roundabout. We introduce the openDD dataset, including 84,774 accuratelytracked trajectories and HD map data of seven different roundabouts. The openDD dataset is annotated using images taken by a drone in 501 separate flights, totalling in over 62 hours of trajectory data. As of today the openDD is by far the largest publicly available trajectory dataset recorded from a drone perspective, while comparable datasets span 17 hours at most. The data is available, for both commercial and noncommercial use, at: http://www.l3pilot.eu/openDD.

Imagining The Road Ahead: Multi-Agent Trajectory Prediction via Differentiable Simulation

“…This issue will most likely be overcome by using additional training data with extensive coverage of road topologies. In recent years, many suitable datasets have been publicly released [31], [32], [33], [34], [35] and training ITRA on them is one of our directions for future work. We note, however, that those datasets come in substantially different formats, particularly regarding the map specification, which requires a significant amount of work to incorporate them into our differentiable simulator.…”

MONA: The Munich Motion Dataset of Natural Driving

“…Their advantage is that the field of view usually reduces occlusions compared to onboard vehicle setups. Similarly, a drone camera also provides a clear topview recording, e.g., highD [6] of highways, inD [7] of urban intersections, openDD [8] and roundD [13] of roundabouts, or INTERACTION [5] of urban intersections and highways. Due to the limited drone battery capacity it is required to interrupt the recording after some time to recharge or exchange the battery.…”

“…INTERACTION [5] highD [6] inD [7] openDD [8] Waymo [2] Argoverse [3] 4) Data extraction: Due to the amount of data to be processed, all datasets rely on automatic labeling. The recorded data is usually processed offline.…”

“…5) License: Apart from openDD [8], no dataset can be used for commercial use free of charge. To facilitate the transfer from research to industry, we argue that restricting the dataset to non-commercial use limits its impact.…”