原创文章,转载请注明: 转载自慢慢的回味
Apollo里面最重要的模块就是感知和规划模块,规划模块最重要的是路径规划和速度规划任务,对应ROS机器人里面的局部规划。Apollo的规划模块首先根据当前的情况进行多场景(scenario)调度,每个场景又包含一个或者多个阶段(stage),每个阶段又由多个具有独立功能的小模块任务(task)完成。这篇文章就主要解读一下路径规划piecewise jerk path optimizer这个任务。任务最终会生成轨迹(trajectory):每个点的位姿和速度信息,进而输出给控制模块去控制车辆。
任务代码调用入口
比如通过测试用例可得到如下的堆栈:
apollo::planning::PiecewiseJerkPathOptimizer::Process(apollo::planning::PiecewiseJerkPathOptimizer * const this, const apollo::planning::SpeedData & speed_data, const apollo::planning::ReferenceLine & reference_line, const apollo::common::TrajectoryPoint & init_point, const bool path_reusable, apollo::planning::PathData * const final_path_data) (piecewise_jerk_path_optimizer.cc:61) apollo::planning::PathOptimizer::Execute(apollo::planning::PathOptimizer * const this, apollo::planning::Frame * frame, apollo::planning::ReferenceLineInfo * const reference_line_info) (path_optimizer.cc:45) apollo::planning::scenario::lane_follow::LaneFollowStage::PlanOnReferenceLine(apollo::planning::scenario::lane_follow::LaneFollowStage * const this, const apollo::common::TrajectoryPoint & planning_start_point, apollo::planning::Frame * frame, apollo::planning::ReferenceLineInfo * reference_line_info) (lane_follow_stage.cc:167) apollo::planning::scenario::lane_follow::LaneFollowStage::Process(apollo::planning::scenario::lane_follow::LaneFollowStage * const this, const apollo::common::TrajectoryPoint & planning_start_point, apollo::planning::Frame * frame) (lane_follow_stage.cc:116) apollo::planning::scenario::Scenario::Process(apollo::planning::scenario::Scenario * const this, const apollo::common::TrajectoryPoint & planning_init_point, apollo::planning::Frame * frame) (scenario.cc:76) apollo::planning::PublicRoadPlanner::Plan(apollo::planning::PublicRoadPlanner * const this, const apollo::common::TrajectoryPoint & planning_start_point, apollo::planning::Frame * frame, apollo::planning::ADCTrajectory * ptr_computed_trajectory) (public_road_planner.cc:38) apollo::planning::OnLanePlanning::Plan(apollo::planning::OnLanePlanning * const this, const double current_time_stamp, const std::vector<apollo::common::TrajectoryPoint, std::allocator<apollo::common::TrajectoryPoint> > & stitching_trajectory, apollo::planning::ADCTrajectory * const ptr_trajectory_pb) (on_lane_planning.cc:572) apollo::planning::OnLanePlanning::RunOnce(apollo::planning::OnLanePlanning * const this, const apollo::planning::LocalView & local_view, apollo::planning::ADCTrajectory * const ptr_trajectory_pb) (on_lane_planning.cc:417) apollo::planning::PlanningTestBase::RunPlanning(apollo::planning::PlanningTestBase * const this, const std::__cxx11::string & test_case_name, int case_num, bool no_trajectory_point) (planning_test_base.cc:229) apollo::planning::SunnyvaleBigLoopTest_keep_clear_02_Test::TestBody(apollo::planning::SunnyvaleBigLoopTest_keep_clear_02_Test * const this) (sunnyvale_big_loop_test.cc:191) |
继续阅读“Apollo二次规划算法(piecewise jerk path optimizer)解析”本作品采用知识共享署名 4.0 国际许可协议进行许可。