Using road topology to improve cyclist path prediction

Using road topology to improve cyclist path prediction

Pool, E.A.I. (Universiteit van Amsterdam)
Kooij, J.F.P. (TU Delft Intelligent Vehicles)
Gavrila, D. (TU Delft Intelligent Vehicles; Universiteit van Amsterdam)

Ioannou, Petros (editor)
Zhang, Wei-Bin (editor)
Lu, Meng (editor)

2017

We learn motion models for cyclist path prediction on real-world tracks obtained from a moving vehicle, and propose to exploit the local road topology to obtain better predictive distributions. The tracks are extracted from the Tsinghua-Daimler Cyclist Benchmark for cyclist detection, and corrected for vehicle egomotion. Tracks are then spatially aligned to local curves and crossings in the road. We study a standard approach for path prediction in the literature based on Kalman Filters, as well as a mixture of specialized filters related to specific road orientations at junctions. Our experiments demonstrate an improved prediction accuracy (up to 20% on sharp turns) of mixing specialized motion models for canonical directions, and prior knowledge on the road topology. The new track data complements the existing video, disparity and annotation data of the original benchmark, and will be made publicly available.

Roads
Tracking
Topology
Predictive models
Benchmark testing
Vehicle dynamics
Layout

http://resolver.tudelft.nl/uuid:809b23e2-ba73-4686-b9b9-4b521d4a7b8d

https://doi.org/10.1109/IVS.2017.7995734

IEEE, Piscataway, NJ, USA

978-1-5090-4804-5

Proceedings of the 2017 IEEE Intelligent Vehicles Symposium (IV)

28th IEEE Intelligent Vehicles Symposium (IV2017), 2017-06-11 → 2017-06-14, Redondo Beach, United States

Accepted Author Manuscript

Institutional Repository

conference paper

© 2017 E.A.I. Pool, J.F.P. Kooij, D. Gavrila