In our vehicle demo “Interaction Self-Driving Vehicle with Pedestrian” at the Intelligent Vehicles 2019 Symposium in Paris, last month in Paris, we showed that by the use of contextual visual information (e.g. has the pedestrian seen the approaching vehicle, what is the pedestrian distance to the curb side, is the situation dangerous?) the self-driving vehicle can anticipate road user behavior better (will the pedestrian stop or cross?) rather than when using kinematic information only (pedestrian position and velocity).
This demo is based on following publications:
- L. Ferranti, B. Brito, E.A.I. Pool, Y. Zheng, R.M. Ensing, R. Happee, B. Shyrokau, J.F.P. Kooij, J. Alonso-Mora and D.M. Gavrila. SafeVRU: A Research Platform for the Interaction of Self-Driving Vehicles with Vulnerable Road Users. Proc. of the IEEE Intelligent Vehicles Symposium (IV), Paris (France), 2019
- J.F.P. Kooij, F. Flohr, E.A.I. Pool, and D. M. Gavrila. Context-Based Path Prediction for Targets with Switching Dynamics. International Journal of Computer Vision, vol. 127, nr. 3, pp. 239-262, 2019
Many thanks to the team that made it happen: Ewoud, Yanggu, Fabian, Frank, Laura, Ronald, Thomas (4activeSystems), Barys, Javier, Julian and … Angela (the dummy)!
We also thankfully acknowledge the support of the Dutch Science Foundation (NWO-TTW), 4activeSystems and Daimler.