About | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange

melodic noetic   Show EOL distros: 

Description

The sections below describe the API of this package which allows ICP-based SLAM in 2D (x,y,phi).

The SLAM algorithm is a simple incremental map building using ICP to align 2D laser scans to the map, which can be either a point map or an occupancy grid map. [For now, only point maps]

Using this technique allows building small to mid-sized maps, as long as errors do not grow excessively before closing any long loop. Building larger maps requires more advanced SLAM techniques. However, this simple icp-slam algorithm is an efficient and well-tested method which suffices for many practical situations.

Further documentation

The ROS node mrpt_icp_slam_2d is a wrapper for the C++ class mrpt::slam::CMetricMapBuilderICP, part of MRPT. Thus, check out the documentation of that class for further details.

This node has been designed to provide an interface similar to that of slam_gmapping for the convenience of users which already knew that package.

For the convention on coordinate frames see ROS REP 105: "Coordinate Frames for Mobile Platforms".

Hardware Requirements

Using mrpt_icp_slam_2d requires a horizontally-mounted, fixed, laser range-finder. Odometry is optional.

The mrpt_icp_slam_2d node will attempt to transform each incoming scan into the odom (odometry) tf frame. See the "Required tf transforms" for more on required transforms.

Usage

In order to use mrpt_icp_slam_2d package it is necessary to install the last MRPT build and themrpt_navigation(see also the tutorial) .

Example

The following video presents the example of icp SLAM both in Rviz and MRPT GUI:

The example of usage icp SLAM in online regime:

roslaunch mrpt_icp_slam_2d icp_slam.launch

The example of usage icp SLAM from .rawlog file:

roslaunch mrpt_icp_slam_2d icp_slam_rawlog.launch

Nodes

mrpt_icp_slam_2d

The mrpt_icp_slam_2d node takes in sensor_msgs/LaserScan messages and builds a point-cloud map (sensor_msgs/PointCloud) or/and gridmap (nav_msgs/OccupancyGrid) . The map can be retrieved via a ROS topic.

Subscribed Topics

tf (tf/tfMessage)

• Transforms necessary to relate frames for laser, base, and odometry (see below)

scan (sensor_msgs/LaserScan)

• Laser scans to create the map from

Published Topics

PointCloudMap (sensor_msgs/PointCloud)

• Get the point-cloud map data from this topic, which is latched, and updated periodically

map (nav_msgs/OccupancyGrid)

• Occupancy grid map topic

robot_pose (geometry_msgs/PoseStamped)

• Estimate of the robot pose

Parameters

~global_frame_id (string, default: "map")

• the frame attached to the map

~base_frame_id (string, default: "base_link")

• The frame attached to the mobile base

~odom_frame_id (string, default: "odom")

• The frame attached to the odometry system

~sensor_source (string, default: "scan")

• The name of topics with 2d laser scans(the name of topic should contaion word "scan")

~ini_filename (string, default: None)

• The path to the ini file with MRPT settings for icp SLAM (see tutorial folder)

~rawlog_filename (string, default: None)

• The path to rawlog file(if the rawlog file is not exists the SLAM starts in online regime)

~rawlog_play_delay (float, default: 0.1)

• Delay for replay SLAM from rawlog file.

Required tf Transforms

<the frame attached to incoming scans> → base_link

• The relative pose of the laser sensor w.r.t. the robot. Usually a fixed value, broadcast periodically by a robot_state_publisher, or a tf static_transform_publisher.

base_link → odom

• The robot odometry, provided by the odometry system (e.g., the driver for the mobile base)

Provided tf Transforms

map → odom

• The current estimate of the robot's pose within the map frame