Automated turning and merging for autonomous vehicles using a Nonlinear Model Predictive Control approach.

Abstract

Accidents at intersections are highly related to the driver's misdecision while performing turning and merging maneuvers. This paper proposes a merging/turning controller for an automated vehicle, called the ego vehicle, which avoids collisions with surrounding (target) vehicles. An optimization-based control problem is defined based on receding horizon control, that parameterizes the system trajectory with the control input and employs a nonlinear model on the ego vehicle dynamics. Most existing solutions focus on 1-D (longitudinal) motion for the vehicles. In this paper, the 2-D motion of the turning/merging vehicle is considered instead. The intersection is modeled under realistic traffic conditions, a probabilistic model is used to predict the trajectories of the target vehicles, and is integrated within a novel collision avoidance model. These models allow our controller to perform both line following when turning/merging, and collision avoidance, while simulations of several scenarios validate its performance.

Bib

@inproceedings{DBLP:conf/amcc/HuangP17a,
  author       = {Lixing Huang and
                  Dimitra Panagou},
  title        = {Automated turning and merging for autonomous vehicles using a Nonlinear
                  Model Predictive Control approach},
  booktitle    = {2017 American Control Conference, {ACC} 2017, Seattle, WA, USA, May
                  24-26, 2017},
  pages        = {5525--5531},
  publisher    = {{IEEE}},
  year         = {2017},
  url          = {https://doi.org/10.23919/ACC.2017.7963814},
  doi          = {10.23919/ACC.2017.7963814},
  timestamp    = {Fri, 03 Dec 2021 13:04:31 +0100},
  biburl       = {https://dblp.org/rec/conf/amcc/HuangP17a.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}