With the purpose of improving the accuracy of vision-based lane change assist system (LCAS), automatic lane change model under uncertainty of vehicle motion in blind zone (BZ) is researched. Firstly, we propose RGBD-flownet, which fused monocular depth network, ego-motion estimation and flownet. They are coupled by scenario reduction and learned in an end-to-end manner. Secondly, lane change decision database is built, which categorize the driving environment based on the observation grid and propose different lane change behavior to handle the different scenarios. Thirdly, optimal trajectory generation model for lane change in a Frenet Frame is proposed, which combile different lateral and longitudinal cost functionals for lane change as well as to mimic human-like driving behavior. Lastly, we compare experienced driver and the proposed automatic lane change model in the same scenarios. It is showed that the proposed model is able to handle complicated lane change scenarios.
Automatic Lane Change Model under Uncertainty of Vehicle Motion in Blind Zone
01.10.2019
2262467 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Lane change behavior with a side blind zone alert system
Online Contents | 2008
|Lane change behavior with a side blind zone alert system
Online Contents | 2008
|Blind zone risk warning method for vehicle lane changing working condition
Europäisches Patentamt | 2025
|