move_base 模块

Published: 2020-03-15

整体模块 #

1. odometry source: 里程计用来提供高频率的概略初始位姿；
2. sensor sources: 机器人或无人车中，包含大量可用于定位的传感器：相机、激光、毫米波雷达等；
3. map_server: 用激光SLAM手段，如 LOAM、GMapping 建立高精度地图；
4. amcl: 结合传感器与高精度地图，实现 蒙特卡洛定位。

规划和控制 #

1. Global Planner

   • global_planner: 可用来取代过时的navfn。 使用A*或Dijkstra算法进行全局路径规划。
   • carrot_planner: 简单的测试算法。

2. Local Planner

   • base_local_planner: 实现Dynamic Window Approach(DWA) and Trajectory Rollout两种算法。其中DWA算法被dwa_local_planner取代。
   • eband_local_planner: Implements the Elastic Band method on the SE2 manifold
   • teb_local_planner: Implements the Timed-Elastic-Band method for online trajectory optimization
   • mpc_local_planner: Provides several model predictive control approaches embedded in the SE2 manifold

3. Recovery Behavior

   • clear_costmap_recovery: A recovery behavior that reverts the costmaps used by move_base to the static map outside of a user-specified range
   • rotate_recovery: A recovery behavior that perfroms a 360 degree rotation of the robot to attempt to clear out space.

其他 #

1. Cost Map

2. Robot Pose EKF

Next: IMU标定与解算
Previous: Kd Tree