►OpenSpaceRoiDeciderConfig ►OpenSpaceROITest OpenSpaceRoiUtil ►OpenSpaceStatus ►OpenSpaceTrajectoryOptimizer ►OpenSpaceTrajectoryOptimizerConfig OpenSpaceTrajectoryOptimizerUtil ►OpenSpaceTrajectoryPartition ►OpenSpaceTrajectoryPartitionConfig ►OpenSpaceTrajectoryProvider ►OpenSpaceTrajectoryProvider...
Open Space算法框架 论文DL-IAPS and PJSO: A Path/Speed Decoupled Trajectory Optimization and its Application in Autonomous Driving是Apollo在2020年发表的Open Space场景中的一种路径和速度解耦的开放空间轨迹规划算法,整个Open Space轨迹规划算法主要由三部分组成。 Hybrid A*:生成Coarse path,作为下游路径平滑的W...
TDR-OBCA(Temporal and Dual warm starts with Reformulated Optimization-based Collision Avoidance (TDR-OBCA) algorithm in Open Space Trajectory Planning)是Apollo开源的openspace场景下的轨迹规划算法,相比较于之前介绍的openspace下的时空解耦的轨迹规划算法DL-IAPS(关于DL-IAPS的介绍可以看这篇文章:Apollo Open ...
PULL_OVER_RETRY_PARKING apollo::planning::PullOverStageRetryParking 执行openspace的轨迹规划,主车位置、航向达到阈值后,退出该stage 1. PULL_OVER_APPROACH PullOverStageApproach : 该阶段用于主车规划接近靠边停车点,获取靠边停车轨迹,并检查是否完成靠边停车。 Process: 该阶段处理的主函数,输入为规划初始点 pl...
主框架分为两个线程，子线程ReferenceLineProvider以20HZ的频率运行，用于计算planning中最重要的数据结构reference_line；主线程上还是基于场景划分的思路，多数场景下还是采用基于ReferenceLine的规划算法，对于泊车相关场景，则利用open space算法。目前Apollo的场景划分为了16种，在proto文件中可以查看到。在Apollo 7.0中...
有了场景之后,会基于不同的规划器(Planner)进行规划,Apollo中目前有俩个Planner,一个是On Lane Planning Tasks,一个是Open Space Planning Tasks。 On Lane Planning Tasks是默认的规划器,接下来会重点介绍。如果大家对Open Space Planning Tasks感兴趣,可以在GitHub找相关的论文和代码。 有了规划器之后,会通过Path...
Currently Open Space Planner is used for 2 parking scenarios in the planning stage namely: Valet This scenario is designed to park the ego car in a designated parking spot using a zig-zag trajectory. Pull Over This scenario is designed to park the ego car along the curb, either normally or...
OPEN_SPACE_TRAJECTORY_PROVIDER 通过OpenSpaceTrajectoryProvider类实现, 主要提供状态机(stop/replan 等)以及多线程计算的功能. 里面会去调用OpenSpaceTrajectoryOptimizer类, 具体实现规划算法, 是核心部分, 其中又分为四个计算模块: HybridAStar: Hybrid A* 搜索获取粗轨迹. ...
Currently Open Space Planner is used for 2 parking scenarios in the planning stage namely: Valet This scenario is designed to park the ego car in a designated parking spot using a zig-zag trajectory. Pull Over This scenario is designed to park the ego car along the curb, either normally or...
On Lane Planning Tasks是默认的规划器，接下来会重点介绍。如果大家对Open Space Planning Tasks感兴趣，可以在GitHub找相关的论文和代码。有了规划器之后，会通过Path decision，把道路边界测出来。主要是通过不同脚本的task拆分完成的，比如Path decision，可以拆分成Lane Borrow和Lane Change。当decision做好之后，还会...