Navigation Stack Overview

In one sentence, the function of this stack is to drive the robot to complete navigation based on the information of external sensors. As for how it is implemented, I will introduce and analyze each package in the future (Welcome to follow), let's take a look at the introduction of each function package in this framework.

Introduction of each function package

• amcl: amcl's role is to use particle filtering to correct the robot's position within a known map based on the robot's own mileage and map characteristics
• base_local_planner: local path planner
• dwa_local_planner: also a local path planner, using the dynamic window method
• carrot_planner: Very simple global path planner, the generated path is the point on the line connecting the target point to the current point of the robot
• clear_costmap_recovery: the recovery algorithm of the path cannot be planned
• costmap_2d: cost map implementation
• fake_localization: mainly used for positioning simulation
• global_planner: global path planning algorithm package
• map_server: Provide cost map management service
• move_base: Robot mobile navigation framework (the main logical framework for navigation)
• move_slow_and_clear: also a recovery strategy
• nav_core: provides an interface that can implement the main package of the plug-in replacement algorithm
• nav_fn: global path planning algorithm
• robot_pose_ekf: integrated odometer, GPS, imu data, location estimation through extended Kalman filtering
• rotate_recovery: rotation recovery strategy implementation package
• voxel_grid: 3D cost map
  
  Figure 1. Navigation Stack source package structure
  As you can see from the picture above, the entire navigation stack can be divided into three parts. In the navigation, move_base is used to implement the logical framework, and nav_core provides a global path planner / local path planner. Interface, through which the algorithm can be replaced according to the form of plug-ins, and two (two-dimensional and three-dimensional) grid maps are also provided;
  The map server mainly manages maps, reads and writes maps, and publishes map messages for other feature packages to subscribe.
  Fake_localization provides an implementation for positioning. robot_pose_ekf is mainly used to modify the robot's mileage based on sensor information, and amcl is mainly used to provide a combination of the robot's odometer and map to estimate the robot Location on the map.

Related connections

Official introduction:http://wiki.ros.org/navigation
Source address:https://github.com/ros-planning/navigation.git