navigation进阶.ppt

Shedding Light on the ROS Navstack 西北工业大学ROS开发者与学习者廖欣祎 Obstacles Layer Inflation Layer Static Layer Range Sensor Layer Proxemic Layer Claustrophobic Layer class Layer { public: void initialize( LayeredCostmap* parent, std::string name, tf::TransformListener *tf ); virtual void updateBounds( double robot_x, double robot_y, double robot_yaw, double* min_x, double* min_y, double* max_x, double* max_y) {} virtual void updateCosts(Costmap2D master_grid, int min_i, int min_j, int max_i, int max_j) {} /ros-planning/navigation /wg-perception/people /DLu/navigation_layers Overview of ROS NavStack Weighted Sum = oscillation_cost + costmap_cost class TrajectoryCostFunction { ... virtual double scoreTrajectory(Trajectory traj); ... } /DLu/navigation /tree/groovy_plugin_planner /DLu/navigation /tree/groovy_mbsplit /ros-planning/navigation DWA Plugin Planner and MoveBase2 Cornerstone of ROS Open Source Platform Dozens of Supported Hardwares Robots Using ROS Overview of ROS Navigation Overview of ROS Navigation Overview of ROS Navigation 如何理解 ROS navigation nav_core: 使用 pluginlib 定义了 各个组件的接口, 框架重用性的来源,使用dynamic_reconfigure实现 组件行为动态变化 . costmap: 作为 planner 的信息输入, 通过不同的costmap cost 取值方法, 不同的传感器输入, 不同的layer 叠加可以最大程度上定制navigation stack planners: 针对不同的应用场景可以使用不同的planner 或者自己定制 Costmap Data Sources Obstacle Inflation Global Costmap and Plan Standard Navigation Social Navigation Monolithic Costmap Costmap Overlap Limited Update Information Layered Costmaps Two PassUpdate Process Layer Combination + = + = Costmap Layers Implementing a Layer Layered Costmaps global_planner BaseLocalPlanner vs. DWALocalPlanner Scoring Trajectories + goal_distance_cost + path_distance_cost + goal_alignment_cost + path_alignment_cost Scoring Different Trajectories Vary x position and x velocity Implementing a Cost Function navigation 逻辑部分---move_base