$_{2}$ "/>

High-Fidelity and Curvature-Continuous Path Smoothing with Quadratic Bézier Curve

IEEE Transactions on Intelligent Vehicles(2023)

引用 0|浏览1
暂无评分
摘要
G $_{2}$ -continuous trajectories are crucial for vehicles under non-holonomic dynamic constraints to run smoothly. In this paper, we present a path smoothing method based on piece-wise quadratic Bézier curves, which can fit a set of sequential 2D collision-free path points generated by a higher-level path planner. The generated trajectories are G $_{2}$ -continuous except at the inflection points, and points of local curvature maxima appear exactly at path points and nowhere else. The local curvature control ensures that the trajectories keep high fidelity with the original paths, thus maintaining key properties of the original paths, such as free collision and gentle turns. Furthermore, we develop a joint optimization method for both fidelity and continuity, which has an efficient analytical solution for each iteration. Extensive experiments in both simulated and real scenarios validate that 1) the smoothed trajectories can deviate less than 10cm from the original paths, 2) the average curvature differences at junctures are under 1e-4 (1/m), 3) the smoothing time reduced by about 50% versus high-order bézier methods, showing superior performance of the proposed method in terms of fidelity, continuity, efficiency, and practicability.
更多
查看译文
关键词
Path smoothing,quadratic Bézier curve,G $_{2}$ -continuity,high-fidelity
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要