An efficient road representation for autonomous vehicles using arc-splines with application to trajectory planning

Date

2024-7

Author

Bolat, Atakan Salih

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

259
views

4059
downloads

Cite This

Autonomous driving technology is the basis of future transportation, particularly on highways, where conditions are optimal due to steady traffic and minimal interruptions. Implementing autonomous driving on highways enhances safety, traffic flow, and environmental impact while benefiting long-distance travel and logistics. Trajectory planning for autonomous vehicles (AVs) on highways is commonly based on high-definition (HD) maps, which aim to provide a highly accurate road representation with low memory requirements. Highways are typically designed with clothoid curves, which are characterized by a linear change in road curvature and can be followed by vehicles at high speeds. Since clothoids lack an analytical representation, this study develops an algorithm combining arc-splines and straight-line segments to approximate clothoid curves for a novel memory-efficient HD map representation. Using real road data from OpenStreetMap and HERE maps, we demonstrate that an analytical road representation using arc-splines and straight segments achieves approximately 3 centimeters of accuracy. A further benefit of our method is the computation of all lanes on a highway by parallel shifting a reference lane. To showcase the usability of our road representation, we perform trajectory planning for AVs using Bezier curves and arc-splines under typical highway conditions. Our findings highlight that arc-spline trajectories are superior to Bezier curves in terms of controllability and computational efficiency. Overall, this research demonstrates that accurate road information, including geometric properties like position, heading, and curvature, can be achieved with low memory requirements and computational effort.

Subject Keywords

Path planning, Autonomous vehicles, Highway modeling, Bézier Curves, Arc-splines

URI

https://hdl.handle.net/11511/110452

Collections

Graduate School of Natural and Applied Sciences, Thesis