摘要
为在统一的算法框架下,结合异构车辆特有的运动特性生成差异化的规划结果,提出一种依托于行为基元生成与选择的运动规划方法。首先,构建融合驾驶行为数据与车辆运动学模型的基元优化生成问题,通过建立具备时空耦合信息的驾驶行为约束条件,根据基元类别以及自身横纵向耦合规律,引导行为基元库的优化生成;其次,在统一的行为基元扩展与选择框架下,从行为基元库中选取基元,并关联为行为基元序列,生成轨迹与速度相耦合的运动规划结果。最后,在非结构化场景中,开展轮式车辆平台和履带式车辆平台,这两种异构平台的实车试验研究。结果表明,所提出的异构车辆统一运动规划方法,得益于行为基元库中基元的多样性以及所涵盖信息的丰富性,能够生成体现异构车辆特性差异的运动规划结果。
Under the unified algorithm framework,by incorporating the unique motion characteristics of heterogeneous vehicles to generate differentiated planning results,a motion planning method based on generating and selecting behavior primitives is proposed.First,a primitive optimization generation problem that integrates driving behavior data and vehicle kinematics model is constructed.By establishing driving behavior constraints with spatio-temporal coupling information,based on the primitive category and its own longitudinal and lateral coupling laws,the optimal generation of the behavior primitive library is completed.Secondly,under the unified behavior primitive extension and selection framework,the primitives are selected from the behavior primitive library,associating as the behavior primitive sequence,generating the motion planning results coupled with trajectory and velocity.Finally,in the unstructured scene,the actual experimental research of the wheeled and tracked vehicle platforms is carried out.The results show that the proposed unified motion planning method for heterogeneous vehicles can generate motion planning results,reflecting the differences in the characteristics of heterogeneous vehicles,benefiting from the diversity of primitives in the behavioral primitive library and the richness of the information covered.
作者
关海杰
王博洋
龚建伟
陈慧岩
GUAN Haijie;WANG Boyang;GONG Jianwei;CHEN Huiyan(School of Mechanical Engineering,Beijing Institute of Technology,Beijing 100081)
出处
《机械工程学报》
EI
CAS
CSCD
北大核心
2024年第18期288-298,共11页
Journal of Mechanical Engineering
基金
国家自然科学基金资助项目(52302489)。
关键词
智能车辆
运动规划
异构车辆
驾驶行为
intelligent vehicle
motion planning
heterogeneous vehicle
driving behavior