Navigation using the move_base and amcl packages

Overview

The move_base package provides an implementation of an action that, given a goal in the world, will attempt to reach it with a mobile base. The move_base node links together a global and local planner to accomplish its global navigation task. The move_base node also maintains two costmaps, one for the global planner, and one for a local planner (see the costmap_2d package) that are used to accomplish navigation tasks.

You can find more information on the move_base ROS Wiki page.

Structure

Usage

Note

In the differential, Mecanum mode, and tracked modes, the launch file for navigation is the same.

Note

Use Ctrl + C command to end all the processes before running the below commands.

1. Start the LIMO. Open a new terminal, and enter the command:

   $ roslaunch limo_bringup limo_start.launch pub_odom_tf:=false
   

2. Launch the navigation function. Open a new terminal, and enter the command:

   $ roslaunch limo_bringup limo_navigation_diff.launch
   

   Note

   If it is Ackermann motion mode, please run

   $ roslaunch limo_bringup limo_navigation_ackerman.launch
   

   After launching successfully, RViz will be opened and display something like the image below.

   

   Note

   To select the map to be loaded, open the limo_navigation_diff.launch file and change the args for the map_server node to the filepath of the map you wish to load. In the image below, this can be changed on line 19.

   <node pkg="map_server" type="map_server" name="map_server" args="$(find limo_bringup)/maps/map02.yaml" output="screen">
   <!--                                                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^               -->
   

   

3. After launching the navigation, you will see that the laser scan display does not quite overlap with the map, meaning that the robot does not have a good estimate of where it is. To correct this, use the 2D Pose Estimate tool in RViz to publish an approximate position of the robot. When the laser scan overlaps the map, the correction is complete.

   

4. Now that we have an accurate estimate of the location of the robot, we can command navigation goals to it. To do this, use the 2D Nav Goal tool in RViz to select a point on the map you want the LIMO to navigate to.

   

   A purple path will be displayed in the map indicating the planned path, and the robot will automatically navigate to the goal.

   

Configuration

AMCL

The parameter configuration files of the amcl package are: amcl_param_diff.yaml (the file is the amcl parameter file used in the four-wheel differential, omnidirectional wheel, and track motion modes), and amcl_param.yaml (the file is the amcl parameter file used in the Ackermann motion mode) .

Parameter	Type	Default	Description
min_particles	int	100	The minimum number of particles allowed.
max_particles	int	5000	The maximum number of particles allowed.
kld_err	double	0.01	The maximum error between the true distribution and the estimated distribution.
kld_z	double	0.99	The upper normal quantile of (1-p), where p is the probability that the error on the estimated detuning will be less than kld_err.
update_min_d	double	0.2m	A translation movement needs to be performed before performing the filter update.
update_min_a	double	π/ 6.0 radians	A rotation movement needs to be performed before performing the filter update.
resample_interval	int	2	The number of filter updates required before resampling.
transform_tolerance	double	0	The time at which the published transformation will be post-processed to indicate that the transformation will be effective in the future.
recovery_alpha_slow	double	0	The exponential decay rate of the slow average weight filter is used to decide when to recover by adding random poses. A good value may be 0.001.
recovery_alpha_fast	double	0.0m	The exponential decay rate of the fast average weight filter is used to decide when to recover by adding random poses. A good value may be 0.1.
initial_pose_x	double	0.0m	The initial pose average (x), used to initialize the filter with Gaussian distribution.
initial_pose_y	double	0.0rad	The initial pose average (y), used to initialize the filter with Gaussian distribution.
initial_pose_a	double	0.5 * 0.5m	The initial pose average (yaw), used to initialize the filter with Gaussian distribution.
initial_cov_xx	double	0.5 * 0.5m	The initial pose covariance (x * x), used to initialize the filter with Gaussian distribution.
initial_cov_yy	double	-1.0 Hz	The initial pose covariance (y * y), used to initialize the filter with Gaussian distribution.
initial_cov_aa	double	0.5 Hz	The initial pose covariance (yaw * yaw), used to initialize the filter with Gaussian distribution.
gui_publish_rate	double	FALSE	The maximum rate (Hz) of publishing visual scans and paths. -1.0 is disabled.
save_pose_rate	double	FALSE	Store the maximum rate (Hz) of the last estimated pose and covariance of the parameter server in the variables ~initial_pose_ and ~initial_cov_. This saved pose will be used in subsequent runs to initialize the filter. -1.0 is disabled.
use_map_topic	bool	-1	When set to be true, AMCL will subscribe to the map topic instead of making a service call to receive its map.
first_map_only	bool	-1	When set to be true, AMCL will only use the first mapping it subscribes to instead of updating each time a new mapping is received.

DWA

The configuration parameter file of DWA is ~/agilex_ws/src/limo_ros/limo_bringup/param/diff/planner.yaml.

Parameter	Type	Default	Description
acc_lim_x	double	2.5	Robot’s x acceleration limit (m/s2)
acc_lim_y	double	2.5	Robot’s y acceleration limit (m/s2)
acc_lim_th	double	3.2	Robot’s rotational acceleration limit (m/s2)
max_vel_trans	double	0.55	The absolute value of the maximum translational velocity of the robot (m/s).
min_vel_trans	double	0.1	The absolute value of the minimum translational velocity of the robot (m/s).
max_vel_x	double	0.55	Robot’s maximum x velocity (m/s)
min_vel_x	double	0.0	Robot’s minimum x velocity (m/s), negative when moving in reverse
max_vel_y	double	0.1	Robot’s maximum y velocity (m/s)
min_vel_y	double	-0.1	Robot’s minimum y velocity (m/s)
max_rot_vel	double	1.0	The absolute value of the maximum rotation velocity of the robot (rad/s)
min_rot_vel	double	0.4	The absolute value of the minimum rotation velocity of the robot (rad/s)
yaw_goal_tolerance	double	0.05	The radian tolerance of the yaw/rotation when the controller achieves its goal
xy_goal_tolerance	double	0.10	The tolerance of the controller in the distance between x and y when achieving the goal (m/s)
latch_xy_goal_tolerance	bool	false	If the goal tolerance is locked, when the robot reaches the goal xy position, it will simply rotate into position, even if it eventually exceeds the goal tolerance while doing so.
sim_time	double	1.7	Time to simulate the trajectory forward in seconds
sim_granularity	double	0.025	Step taken between points on a given trajectory (m/s)
vx_samples	int	3	The number of samples used when exploring the x velocity space
vy_samples	int	10	The number of samples used when exploring the y velocity space
vth_samples	int	20	The number of samples used when exploring the theta velocity space
controller_frequency	double	20.0	Call the controller’s frequency. If it is not set in the controller’s namespace, use searchParam to read the parameters from the parent namespace. Use together with move_base, which means you only need to set its “controller_frequency” parameter and you can safely not set this parameter.
path_distance_bias	double	32.0	The weight that how close the controller should be to the given path
goal_distance_bias	double	24.0	The weight that the controller should try to reach its local goal and it should also control the velocity
occdist_scale	double	0.01	The weight that the controller should try to avoid obstacles
forward_point_distance	double	0.325	The distance from the center of the robot to the additional scoring point, in meters
stop_time_buffer	double	0.2	The amount of time the robot must stop before colliding for the trajectory to be valid, in seconds
scaling_speed	double	0.25	The absolute value of the speed at which the robot’s footprint is scaled (m/s)
max_scaling_factor	double	0.2	The biggest factor in scaling a robot’s footprint
publish_cost_grid	bool	false	Whether the cost grid that the planner will use when planning will be published? When it’s true, sensor_msgs/PointCloud2 will be available on the ~/cost_cloud topic. Each point cloud represents a cost grid and has a field for each individual scoring function component and the total cost of each cell, taking the scoring parameters into account.
oscillation_reset_dist	double	0.05	How far the robot must move in meters before resetting the oscillation tag
prune_plan	bool	true	Define whether the robot will eat the plan when moving along the path. If it’s set to be true, the points will fall from the end of the plan as soon as the robots move more than 1 meter.

TEB

The parameter configuration file of TEB is: ~/agilex_ws/src/limo_ros/limo_bringup/param/carlike2/teb_local_planner_params.yaml.

Parameter	Type	Default	Description
acc_lim_x	double	0.5	Robot’s maximum translational acceleration (m/s ^2)
acc_lim_theta	double	0.5	Robot’s maximum angular acceleration (radian/s ^2)
max_vel_x	double	0.4	Robot’s maximum translational velocity (m/s)
max_vel_x_backwards	double	0.2	The maximum absolute translational velocity (in m/s) when the robot is traveling backwards.
max_vel_theta	double	0.3	Robot’s maximum angular velocity (radian/s)
min_turning_radius	double	0.0	Automotive robot’s minimum turning radius (set to be zero for differential drive robots).
wheelbase	double	1.0	The distance between the rear axle and the front axle. For rear-wheel robots, this value may be negative (only required when cmd_angle_instead_rotvel is set to be true).
cmd_angle_instead_rotvel	bool	false	Replace the rotation velocity in the command velocity message with the corresponding steering angle [-pi/2, pi/2]. Note that it is not advisable to change the semantics of the yaw rate according to the application. Here, it is only the input required by the stage simulator.The data type in ackermann_msgs is more appropriate, but move_base does not support it. The local planner itself does not intend to send commands.
max_vel_y	double	0.0	Robot’s maximum sweep velocity (it should be zero for incomplete robots!)
acc_lim_y	double	0.5	Robot’s maximum sweep acceleration
footprint_model/type	double	point	Specify the type of robot footprint model used for optimization. The different types are “point”, “circle”, “line”, “two_circles” and “polygon”. The type of model can significantly affect the required calculation time.
footprint_model/radius	double	0.2	This parameter is only related to the “circle” type. It contains the radius of the circle. The center of the circle is on the rotation axis of the robot.
footprint_model/line_start	double	[-0.3, 0.0]	This parameter is only related to the “line” type. It contains the starting coordinates of the line segment.
footprint_model/line_end	double	[0.3, 0.0]	This parameter is only related to the “line” type. It contains the ending coordinates of the line segment.
footprint_model/front_offset	double	0.2	This parameter is only related to the “two_circles” type. It describes how much the center of the front circle has moved along the x-axis of the robot. Assume that the rotation axis of the robot is located at [0,0].
footprint_model/front_radius	double	0.2	This parameter is only related to the “two_circles” type. It contains the radius of the front circle.
footprint_model/rear_offset	double	0.2	This parameter is only related to the “two_circles” type. It describes how much the center of the back circle has moved along the negative x-axis of the robot. Assume that the rotation axis of the robot is located at [0,0].
footprint_model/rear_radius	double	0.2	This parameter is only related to the “two_circles” type. It contains the radius of the back circle.
footprint_model/vertices	double	[0.25,-0.05]	This parameter is only related to the “polygon” type. It contains a list of polygon vertices (each is a two-dimensional coordinate). Polygons are always closed: do not repeat the first vertex at the end.
is_footprint_dynamic	bool	false	If it’s true, the footprint is updated before checking the trajectory’s feasibility.
xy_goal_tolerance	double	0.2	Allowable final Euclidean distance to the goal position (in meters).
yaw_goal_tolerance	double	0.2	Allowable final direction error (in radians).
free_goal_vel	bool	false	Remove the goal velocity constraint, so that the robot can reach the goal at the maximum velocity.
dt_ref	double	0.3	The required time resolution of the trajectory (the trajectory is not fixed to dt_ref, because the time resolution is part of the optimization, but if dt_ref +-dt_hysteresis is violated, the trajectory size will be adjusted between iterations.
dt_hysteresis	double	0.1	The lag that is automatically resized according to the current time resolution, usually about 10% of the recommended dt_ref.
min_samples	int	3	Minimum number of samples (should always be greater than 2).
global_plan_overwrite_orientation	bool	true	Override the direction of the local sub-goals provided by the global planner (because they usually only provide a two-dimensional path).
global_plan_viapoint_sep	double	-0.1 (disabled))	If it is positive, the via points are extracted from the global plan (path following mode). This value determines the resolution of the reference path (the minimum period between every two consecutive via points along the global plane, if it is negative.
max_global_plan_lookahead_dist	double	3.0	Specify the maximum length (cumulative Euclidean distance) of the subset of the global plan considered for optimization. The actual length is determined by the logical combination of the size of the local cost map and this maximum limit. Set to be zero or a negative number to deactivate this limit.
force_reinit_new_goal_dist	double	1.0	If the previous goal update period is greater than the specified value (in meters), re-initialize the trajectory (skip hot start).
feasibility_check_no_poses	bool	4	Specify that the feasibility of the pose on the prediction plan should be checked during each sampling period.
publish_feedback	bool	false	Publish planner feedback with complete trajectory and active obstacle list (it should only be enabled for evaluation or debugging). See the publisher list above.
shrink_horizon_backup	bool	true	Allow the planner to temporarily shrink the scope (50%) in the event that a problem (such as infeasibility) is automatically detected. See also parameter shrink_horizon_min_duration.
allow_init_with_backwards_motion	bool	false	If it’s true, the base trajectory may be initialized with backward motion in case the goal is behind the starting point in the local cost map (this is only recommended if the robot is equipped with a rear sensor).
exact_arc_length	double	false	If it’s true, the planner uses the precise arc length (-> increased CPU time) in the calculation of velocity, acceleration and turn rate, otherwise it uses the Euclidean approximation.
shrink_horizon_min_duration	double	10.0	If an infeasible trajectory is detected, please specify the shortest duration for shrinking the horizon (see the parameter shrink_horizon_backup to activate the shrink horizon mode).
min_obstacle_dist	double	0.5	The minimum expected distance to the obstacle (in meters).
include_costmap_obstacles	double	true	Specify whether or not the obstacles of the local cost map should be considered. Each cell marked as an obstacle is treated as a point obstacle. Therefore, do not choose a very small cost map resolution, because it will increase the calculation time. In future versions, this situation will be resolved and additional api will be provided for dynamic obstacles.
costmap_obstacles_behind_robot_dist	bool	1.0	Limit the occupied local cost map obstacles that are taken into account when planning behind the robot (specify the distance in meters) .
obstacle_poses_affected	double	30	Each obstacle position is attached to the nearest pose on the trajectory to maintain the distance. You can also consider additional neighbors. Please note that this parameter may be removed in a future version because the obstacle association strategy has been modified in kinetic+. Refer to the parameter description of legacy_obstacle_association.
inflation_dist	double	pre kinetic	Buffer around obstacles with non-zero penalty cost (should be greater than min_obstacle_dist to take effect). Refer weight_inflation.
include_dynamic_obstacles	string	false	If this parameter is set to be true, the motion of obstacles with non-zero velocity will be predicted and considered through the constant velocity model during the optimization process (provided through user-provided obstacles or obtained from the costmap_converter). New
legacy_obstacle_association	bool	false	The strategy for connecting trajectory poses and optimizing obstacles has been revised (see change log). You can switch to the old/previous strategy by setting this parameter to be true. Old strategy.
obstacle_association_force_inclusion_factor	double	1.5	The non-legacy obstacle association strategy tries to connect only the relevant obstacles with the discretized trajectory in the optimization process. But all obstacles within the specified distance are forcibly included (as a multiple of min_obstacle_dist). For example, choose 2.0 to force the consideration of obstacles within a radius of 2.0*min_obstacle_dist. [This parameter is used only when the parameter legacy_obstacle_association is false]
obstacle_association_cutoff_factor	int	5	See obstacle_association_force_inclusion_factor, but all obstacles that exceed the multiple of [value] * min_obstacle_dist are ignored in the optimization process. The parameter obstacle _association_force_inclusion_factor is processed first. [This parameter is used only when the parameter legacy_obstacle_association is false]
costmap_converter_plugin	int	“”	Define the plug-in name to convert cost map cells to points/lines/polygons. Set an empty string to disable conversion so that all cells are treated as point obstacles.
costmap_converter_spin_thread	double	true	If set to be true, the cost map converter calls its callback queue in a different thread.
costmap_converter_rate	double	5.0	Rate defines the frequency that the costmap_converter plugin processes the current cost map(This value should not be higher than the cost map update rate) [in Hertz].
no_inner_iterations	double	5	The actual number of solver iterations called in each outer loop iteration. See parameter no_outer_iterations.
no_outer_iterations	double	4	Each outer loop iteration will automatically adjust the trajectory size and call the internal optimizer (execute no_inner_iterations) according to the required time resolution dt_ref. Therefore, the total number of solver iterations in each planning cycle is the product of the two values.
penalty_epsilon	double	0.1	Add a small safety margin to the penalty function of the hard constraint approximation.
weight_max_vel_x	double	2.0	Optimized weight to meet the maximum allowable translational velocity.
weight_max_vel_theta	double	1.0	Optimized weight to meet the maximum allowable angular velocity.
weight_acc_lim_x	double	1.0	Optimized weight to meet the maximum allowable translational acceleration.
weight_acc_lim_theta	double	1.0	Optimized weight to meet the maximum allowable angular acceleration.
weight_kinematics_nh	double	1000.0	The optimized weight used to meet non-holonomic kinematics (this parameter must be very high, because the kinematics equation constitutes an equality constraint, and the “original” cost value is small compared with other costs, so even a value of 1000 does not mean the matrix condition is not good).
weight_kinematics_forward_drive	double	1.0	The optimized weight to force the robot to select only the forward direction (positive translational velocity). The small weight (eg. 1.0) still allows driving backwards. A value around 1000 can almost prevent backward driving (but it cannot be guaranteed).
weight_kinematics_turning_radius	double	1.0	The optimized weight to force the minimum turning radius (only for automobile robots).
weight_optimaltime	double	1.0	Optimized weight to shorten trajectory wrt conversion/execution time.
weight_obstacle	bool	50.0	Optimized weight to keep the minimum distance from the obstacle.
weight_viapoint	bool	1.0	The optimized weight (corresponding reference path) used to minimize the distance to the passing point. 0.4 new version.
weight_inflation	int	0.1	The optimized weight of the inflation penalty (should be small).
weight_adapt_factor	double	2.0	In each outer TEB iteration (weight_new = weight_old*factor), some special weights (currently weight_obstacle) are repeatedly scaled by this factor. Iteratively increasing the weights instead of setting a huge prior value will lead to better numerical conditions for the basic optimization problem.
enable_homotopy_class_planning	double	true	Activate parallel planning in different topologies (requires more CPU resources because multiple trajectories are optimized at once).
enable_multithreading	double	true	Activate multiple threads to plan each trajectory in different threads.
max_number_classes	bool	4	Specify the maximum number of different trajectories to be considered (limits the computational effort).
selection_cost_hysteresis	int	1.0	Specify how much trajectory cost the new candidate must have compared to the previously selected trajectory to be selected (select if new_cost <old_cost*factor).
selection_obst_cost_scale	double	100.0	Additional scaling of the obstacle cost term used only to select the “best” candidate.
selection_viapoint_cost_scale	double	1.0	The additional scaling used only to select the “best” candidate through the point cost clause. 0.4 new version.
selection_alternative_time_cost	double	false	If it’s true, the time cost (sum of squares of the time difference) will be replaced with the total conversion time (sum of the time difference).
roadmap_graph_no_samples	double	15	Specify the number of samples generated to create the roadmap.
roadmap_graph_area_width	bool	6	Sample the random keypoints/waypoints in the rectangular area between the starting point and the goal. Specify the width of the area in meters.
h_signature_prescaler	bool	1.0	The internal parameter (H-signature) of the ratio used to distinguish homotopy classes. Warn
h_signature_threshold	double	0.1	If the differences between the real part and the complex part are lower than the specified threshold, it is assumed that the two H signatures are equal.
obstacle_heading_threshold	string	1.0	Specify the value of the scalar product between the obstacle course and the goal course so that they (obstacles) are taken into account when exploring.
visualize_hc_graph	string	false	Visualize the graph created to explore the unique trajectory (check the tag message in rviz).
viapoints_all_candidates	bool	true	If it’s true, all trajectories of different topologies are attached to a set of via points, otherwise only trajectories that share the same topology with the initial/global plan are connected to them (no effect on test_optim_node). 0.4 new version
switching_blocking_period	double	0.0	Specify the duration (in seconds) that needs to expire before being allowed to switch to the new equivalence class.
odom_topic	double	odom	The subject name of the odometer message, provided by the robot driver or simulator.
map_frame	bool	odom	Global planning framework (if it is a static map, this parameter usually must be changed to “/map”.