struct mola::BridgeROS2::Params

Overview

struct Params
{
    // fields

    std::string base_link_frame = "base_link";
    std::string base_footprint_frame;
    mrpt::math::TPose3D base_footprint_to_base_link_tf = {0, 0, 0, 0, 0, 0};
    std::string odom_frame = "odom";
    std::string reference_frame = "map";
    bool publish_localization_following_rep105 = true;
    bool forward_ros_tf_as_mola_odometry_observations = false;
    bool publish_odometry_msgs_from_slam = true;
    std::string publish_odometry_msgs_from_slam_source;
    bool publish_tf_from_slam = true;
    std::string publish_tf_from_slam_source;
    bool publish_geo_referenced_poses_from_slam = true;
    bool publish_tf_from_robot_pose_observations = true;
    std::string relocalize_from_topic = "/initialpose";
    bool publish_in_sim_time = false;
    double transform_tolerance = 0.1;
    double transform_publish_period = 0.05;
    double period_publish_new_map = 5.0;
    double period_publish_static_tfs = 1.0;
    double period_publish_diagnostics = 1.0;
    double period_check_new_mola_subs = 1.0;
    int wait_for_tf_timeout_milliseconds = 100;
    std::string georef_map_reference_frame = "map";
    std::string georef_map_utm_frame = "utm";
    std::string georef_map_enu_frame = "enu";
    bool odometry_as_robot_pose_observation = true;
};

Fields

std::string base_link_frame = "base_link"

tf frame name with respect to sensor poses are measured, and also used for publishing SLAM/localization results (read below).

std::string base_footprint_frame

If not empty, the node will broadcast a static /tf from base_footprint to base_link with the TF base_footprint_to_base_link_tf at start up. Normally: “base_footprint”

mrpt::math::TPose3D base_footprint_to_base_link_tf = {0, 0, 0, 0, 0, 0}

YAML format: “[x y z yaw pitch roll]” (meters & degrees)

std::string odom_frame = "odom"

Used for: (a) importing odometry to MOLA if forward_ros_tf_as_mola_odometry_observations=true (b) querying ${odom_frame} => ${base_link_frame} when publish_localization_following_rep105=true.

std::string reference_frame = "map"

tf frame used for: (a) See publish_tf_from_robot_pose_observations (b) To follow REP105 (publish_localization_following_rep105), this must match the frame used as reference in the LocalizationSource (e.g. mola_lidar_odometry)

bool publish_localization_following_rep105 = true

How to publish localization to /tf:

false (direct mode): reference_frame (“map”) -> base_link (“base_link”) Note that reference_frame in this case comes from the localization source module (e.g. mola_lidar_odometry), it is not configured here.
- true (indirect mode), following ROS REP 105 : map -> odom (such as “map -> odom -> base_link” = “map -> base_link”)

bool forward_ros_tf_as_mola_odometry_observations = false

If enabled, during spinOnce(), the tf ${odom_frame} => ${base_link_frame} will be queried and forwarded as an CObservationOdometry reading to the MOLA subsystem:

bool publish_odometry_msgs_from_slam = true

If enabled, SLAM/Localization results will be published as nav_msgs/Odometry messages.

bool publish_tf_from_slam = true

If enabled, SLAM/Localization results will be published as tf messages, for frames according to explained above for publish_localization_following_rep105.

bool publish_geo_referenced_poses_from_slam = true

If enabled, SLAM/Localization results in geo-referenced maps will be published as geographic_msgs/GeoPoseStamped messages.

bool publish_tf_from_robot_pose_observations = true

If enabled, robot pose observations (typically, ground truth from datasets), will be forwarded to ROS as /tf messages: ${reference_frame} => ${base_link}

std::string relocalize_from_topic = "/initialpose"

Default in RViz.

bool publish_in_sim_time = false

If true, the original dataset timestamps will be used to publish. Otherwise, the wallclock time will be used.

double transform_tolerance = 0.1

Future-dating offset [s] applied to the broadcast stamp of the REP-105 localization /tf (map -> odom). The stamp is set to rosNode->now() + transform_tolerance, allowing downstream consumers to lookupTransform(map, base_link, now()) without tf2 ExtrapolationException s. Ignored in direct-publish mode (map -> base_link), where the TF is stamped at the scan acquisition time so consumers can correlate the pose with the sensor data that produced it (and publish_in_sim_time is honored).

double transform_publish_period = 0.05

Period [s] at which the latest computed REP-105 localization /tf (map -> odom) is re-broadcast on a wall timer, independent of localization update rate. This keeps the TF buffer always populated for downstream consumers even when the localizer runs slower than the consumer query rate. Set to 0 to disable the rebroadcast loop (the TF is then only published when a new localization update arrives). Only effective in REP-105 mode: in direct-publish mode (map -> base_link) the rebroadcast is skipped to avoid re-stamping a stale pose as “current” when the localizer stalls in that mode there is no odom -> base_link edge carrying real motion, so the robot would silently appear frozen-but-fresh.

bool odometry_as_robot_pose_observation = true

If true (default), subscribed nav_msgs/Odometry messages are converted to mrpt::obs::CObservationRobotPose (full SE(3) pose + 6x6 covariance), suitable for multi-source fusion in the state estimation smoother. If false, they are converted to mrpt::obs::CObservationOdometry (2D pose, no covariance), which is more appropriate for 2D SLAM/mapping pipelines.