Over 60,000 km in a year: remotely collecting large-volume high-quality data from a logistics truck
Tóm tắt
After the first successful large-scale demonstration of eleven self-driving vehicles at the DARPA Urban Challenge in 2007, research results from the competing teams found their way into advanced driver systems (ADAS) that support typical driving tasks like adaptive cruise control and semi-automated parking. However, as of today, SAE Level 4 vehicles are not commercially available yet, which would allow the driver to be inattentive for longer periods. Hence, SAE Level 3, which represents partial automation yet continuously monitored by a human operator, may provide a step towards a viable SAE Level 4 product especially for commercial freight logistics. However, large amounts of data from such freight operations is needed to study the unique challenges in such use cases. In this paper, we present the system and software architecture of an end-to-end data logging solution, which is capable of recording large volumes of high-quality data. The system is installed in a commercial truck that is in daily operation by a logistics company and hence, the recorded data is only accessible remotely (i.e., over-the-air). We report about the fail-safe system design, initial findings from over one year of operation, as well as our lessons learned. During its first year of operation, the truck was used for 210 days by the logistics company, out of which 193 days were logged resulting in more than 4.5 TB of data from five cameras, two GNSS–IMU sensors, and six on-board vehicle controller area networks (CAN) busses. We demonstrate the value of the proposed end-to-end approach for traffic and driver behavior research by analyzing the uploaded data in the cloud to spot critical events such as unexpected harsh braking maneuvers caused by lane merging operations.
Tài liệu tham khảo
Buehler M, Iagnemma K, Singh S (2009) The DARPA urban challenge: autonomous vehicles in city traffic, vol 56. Springer, New York
Berger C, Rumpe B (2012) Engineering autonomous driving software. In: Rouff C, Hinchey M (eds) Experience from the DARPA Urban Challenge. Springer-Verlag, London, UK, pp 243–271. https://arxiv.org/pdf/1409.6579
Bojarski M, Del Testa D, Dworakowski D, Firner B, Flepp B, Goyal P, Jackel L. D, Monfort M, Muller U, Zhang J, Zhang X, Zhao J, Zieba K (2016) End to end learning for self-driving cars. NVIDIA Corporation, Holmdel. http://arxiv.org/pdf/1604.07316v1.pdf
Yin H, Berger C (2017) When to use what data set for your self-driving car algorithm: an overview of publicly available driving datasets. In: Proceedings of the 20th IEEE international conference on intelligent transportation systems (ITSC). Yokohama
Kang Y, Yin H, Berger C (2019) Test your self-driving algorithm: an overview of publicly available driving datasets and virtual testing environments. IEEE Trans Intell Veh 4(2):171–185
Geiger A, Lenz P, Stiller C, Urtasun R (2013) Vision meets robotics: the KITTI dataset. Int J Rob Res 32(11):1231–1237
Maddern W, Pascoe G, Linegar C, Newman P (2017) 1 year, 1000km: the Oxford RobotCar Dataset. Int J Robot Res (IJRR) 36(1):3–15
Caesar H, Bankiti V, Lang A. H, Vora S, Liong V. E, Xu Q, Krishnan A, Pan Y, Baldan G, Beijbom O (2020) nuscenes: a multimodal dataset for autonomous driving
Sun P, Kretzschmar H, Dotiwalla X, Chouard A, Patnaik V, Tsui P, Guo J, Zhou Y, Chai Y, Caine B, et al (2020) Scalability in perception for autonomous driving: Waymo open dataset. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 2446–2454
Wieringa RJ (2014) Design science methodology for information systems and software engineering. Springer, New York
Kamla J, Dawson A, Parry T (2017) Feasibility of using truck harsh braking incidents for predicting all vehicle accidents at roundabouts. In: Transportation Research Board 96th Annual Meeting, Jan, p 16