Foxy to Galactic¶
Moving from ROS 2 Foxy to Galactic, a number of stability improvements were added that we will not specifically address here.
BackUp BT-node Interface Changes¶
backup_speed input ports should both be positive values indicating the distance to go backward respectively the speed with which the robot drives backward.
BackUp Recovery Interface Changes¶
speed in a backup recovery goal should be positive indicating the speed with which to drive backward.
target.x in a backup recovery goal should be positive indicating the distance to drive backward.
In both cases negative values are silently inverted.
nav2_waypoint_follower has an action server that takes in a list of waypoints to follow and follow them in order. In some cases we might want robot to
perform some tasks/behaviours at arrivals of these waypoints. In order to perform such tasks, a generic plugin interface WaypointTaskExecutor has been added to
Users can inherit from this interface to implement their own plugin to perform more specific tasks at waypoint arrivals for their needs.
Several example implementations are included in
PhotoAtWaypoint plusings are included in
nav2_waypoint_follower as run-time loadable plugins.
WaitAtWaypoint simply lets robot to pause for a specified amount of time in milliseconds, at waypoint arrivals.
PhotoAtWaypointtakes photos at waypoint arrivals and saves the taken photos to specified directory, the format for taken photos also can be configured through parameters. All major image formats such as
jpgetc. are supported, the default format is
Loading a plugin of this type is done through
nav2_bringup/params/nav2_param.yaml, by specifying plugin’s name, type and it’s used parameters.
For instance; .. code-block:: yaml
loop_rate: 20 stop_on_failure: false waypoint_task_executor_plugin: “wait_at_waypoint”
- plugin: “nav2_waypoint_follower::WaitAtWaypoint” enabled: True waypoint_pause_duration: 0
Original GitHub tickets:
A new concept interacting with spatial-dependent objects called “Costmap Filters” appeared in Galactic (more information about this concept could be found at Navigation Concepts page). Costmap filters are acting as a costmap plugins. In order to make a filtered costmap and change robot’s behavior in annotated areas, filter plugin reads the data came from filter mask. Then this data is being linearly transformed into feature map in a filter space. It could be passability of an area, maximum speed limit in m/s, robot desired direction in degrees or anything else. Transformed feature map along with the map/costmap, sensors data and current robot position is used in plugin’s algorithms to make required updates in the resulting costmap and robot’s behavor.
Architecturally, costmap filters consists from
CostmapFilter class which is a basic class incorporating much common of its inherited filter plugins:
KeepoutFilter: keep-out/safety zones filter plugin.
SpeedFilter: slow/speed-restricted areas filter.
- Preferred lanes in industries. This plugin is covered by
KeepoutFilter(see discussion in corresponding PR for more details).
Each costmap filter subscribes to filter info topic (publishing by Costmap Filter Info Publisher Server) having all necessary information for loaded costmap filter and filter mask topic.
High-level design of this concept could be found here. The functionality of costmap filters is being disscussed in the ticket #1263 and carried out by PR #1882. The following tutorial: Navigating with Keepout Zones will help to easily get involved with
A new package,
SmacPlanner was added containing 4 or 8 connected 2D A*, and Dubin and Reed-shepp model hybrid-A* with smoothing, multi-resolution query, and more.
smac_planner package contains an optimized templated A* search algorithm used to create multiple A*-based planners for multiple types of robot platforms. We support differential-drive and omni-directional drive robots using the
SmacPlanner2D planner which implements a cost-aware A* planner. We support cars, car-like, and ackermann vehicles using the
SmacPlanner plugin which implements a Hybrid-A* planner. This plugin is also useful for curvature constrained planning, like when planning robot at high speeds to make sure they don’t flip over or otherwise skid out of control.
The SmacPlanner fully-implements the Hybrid-A* planner as proposed in Practical Search Techniques in Path Planning for Autonomous Driving, including hybrid searching, CG smoothing, analytic expansions and hueristic functions.