Laser Odometry

This is the patched fork project of rf2o_laser_odometry by MAPIRlabs

Estimation of 2D odometry based on planar laser scans. Useful for mobile robots with innacurate base odometry. http://mapir.isa.uma.es/work/rf2o For full description of the algorithm, please refer to: Planar Odometry from a Radial Laser Scanner. A Range Flow-based Approach. ICRA 2016

The rf2o_laser_odometry module is ROS node which estimates path and displacements of the mobile robot using planar LiDAR scans. The module designed as standard ROS package with following dependencies:

catkin
roscpp
sensor_msgs
std_msgs
tf
cmake_modules
eigen
nav_msgs

To build the module within the existing workspace, standard catkin_make script could be used.

Runnning

The module runs under control of roslaunch server with initialization script in corresponding .launch-file containing XML like following:

<launch>
    <node   pkg="rf2o_laser_odometry" 
            type="rf2o_laser_odometry_node" 
            name="rf2o_laser_odometry" 
            output="screen">
        <rosparam file="$(find rf2o_laser_odometry)/config/default.yml"/>
    </node>
</launch>

Configuration

Using <param> XML tag the configuration parameters could be placed into the .launch-file directly but it is strongly recommended to create special dedicated YAML configuration file for <rosparam> tag (the example configuration file with default values for all parameters is shipped with the module, see default.yml). All parameters in this file could be omitted to remain with default values.

General settings

Name	Default value	Type	Description
`verbose`	`false`	`bool`	Toggles debug messages on/off
`counter_clockwise`	`true`	`bool`	Set this flag to `true`, if your LiDAR sensor or mirror block rotates counter clockwise. For some robots, e.a. Turtlebot 3 the rotation direction of the mirror is clockwise, so this flag should be set `false` to avoid negative values of robot coordinate accretions
`laser_scan_topic`	`/laser_scan`	`std::string`	Topic name to subscribe for laser scans
`laser_frame_id`	`/laser`	`std::string`	Frame ID for laser scan message header. CAUTION! The coordinate transform should exist from `base_frame_id` to `laser_frame_id` to successfully run the module
`odom_topic`	`/odom_rf2o`	`std::string`	Topic name to publish output odometry messages
`publish_tf`	`true`	`bool`	Toggles publishing of coordinate transformation between `base_frame_id` and `odom_frame_id`
`base_frame_id`	`/base_link`	`std::string`	Frame ID for coordinate system/transformation snapped to the robot itself
`odom_frame_id`	`/odom`	`std::string`	Frame ID for output odometry message header
`init_pose_from_topic`	`/base_pose_ground_truth`	`std::string`	Topic name to subscribe for initial pose
`freq`	`10.0`	`double`	Odometry update frequency (Hz). This setting should be approximately equal to LiDAR sensor polling rate

Covariation settings

Name	Default value	Type	Description
`pose_covariance_matrix`	Zeros 6х6	`std::vector<double>`	Static pose covariation matrix
`twist_covariance_matrix`	Zeros 6х6	`std::vector<double>`	Static covariation matrix for kinematical parameters

Covariation auto-adjustment settings

This patched version of the module can automatically boost the covariation when odometry gap (for example, because of long straight corridor) is detected. The corresponding settings are placed into dynamic_covariance_boost group in YAML file.

Name	Default value	Type	Description
`enable`	`false`	`bool`	Toggles automatic covariation boost on/off
`initial_multiplier`	`1.0`	`double`	Base multiplpier for covariation matrix
`progressive`	`false`	`bool`	Toggles additional boosting progression on/off when the odometry gap longs more that one polling period
`progression_factor`	`0`	`double`	Progression booster value

Trusted ranges settings

These settings are dedicated to detection of odometry gaps. They allows user to set thresholds and fallback sources to replace the odometry if the obtained values become impossible or untrusted for the current robot. These settings are placed into the group pose_fallback in YAML file.

Trusted values settings

Name	Default value	Type	Description
`continuous_fallback_topic`	Empty string	`std::string`	Fallback topic to replace the pose data for the detected gap
`velocity_fallback_topic`	`continuous_fallback_topic` value	`std::string`	Fallback topic to replace the motion data for the detected gap
`enable_thresholds`	`false`	`bool`	Toggles trust thresholds on/off
`linear_velocity_threshold_x`	$10^6$	`double`	Lenghtwise velocity value trust threshold
`linear_velocity_threshold_y`	`linear_velocity_threshold_x` value	`double`	Transversal velocity value trust threshold
`angular_velocity_threshold`	$10^6$	`double`	Angular velocity value trust threshold. The angular velocity is considered to be define in `base_frame_id` coordinate system

twdragon/rf2o_laser_odometry