Hello,
Sorry if this is a double post, i can seem to find anyone with the same issue.
I have been working with move_base for a while now using different platforms. But I have never experienced this issue, maybe some of you did.
The local costmap seems to only be showing some of the obstacles but not all of them, particularly dynamic ones.
The picture below shows an exmaple, where the local cost map picks up objects by the side of the robot, but not the table that is just infront of him.
Similar things happen with dynamic ones (people) the costmap shows them at random points around the robot, but not all the time.
The robot is setup with a Hokuyo lidar. here are the config files:
common_costmap.yaml
```
global_frame: map
robot_base_frame: base_link
footprint: [[0.42, 0.26], [-0.42, 0.26], [-0.42, -0.26], [0.42, -0.26]]
footprint_padding: 0.03
inflater_layer:
inflation_radius: 0.05
cost_scaling_factor: 0.5
plugins:
- {name: obstacles_layer, type: "costmap_2d::ObstacleLayer"}
- {name: inflater_layer, type: "costmap_2d::InflationLayer"}
```
global_costmap.yaml:
```
global_costmap:
update_frequency: 5.0
static_map: true
rolling_window: false
resolution: 0.05
static_layer:
map_topic: /map
subscribe_to_updates: true
lethal_cost_threshold: 60
inflater_layer:
inflation_radius: 0.25
cost_scaling_factor: 0.5
plugins:
- {name: static_layer, type: "costmap_2d::StaticLayer"}
- {name: inflater_layer, type: "costmap_2d::InflationLayer"}
```
local_costmap.yaml:
```
local_costmap:
update_frequency: 5.0
publish_frequency: 5.0
static_map: false
rolling_window: true
width: 8.0
height: 8.0
resolution: 0.05
obstacles_layer:
observation_sources: laser_scan_sensor
laser_scan_sensor: {sensor_frame: horizontal_laser_link, data_type: LaserScan, topic: scan, clearing: true, marking: true}
max_obstacle_height: 3
obstacle_range: 10
raytrace_range: 11
```
If anyone have any Idea why this is happening, you will help me sleep at night.
Best,
Pepe
↧
Local_costmap not showing every obstacle
↧
Local_costmap not showing all obstacles
Hello,
Sorry if this is a double post, i can seem to find anyone with the same issue.
I have been working with move_base for a while now using different platforms. But I have never experienced this issue, maybe some of you did.
The local costmap seems to only be showing some of the obstacles but not all of them, particularly dynamic ones.
The picture below shows an exmaple, where the local cost map picks up objects by the side of the robot, but not the table that is just infront of him.
Similar things happen with dynamic ones (people) the costmap shows them at random points around the robot, but not all the time.
The robot is setup with a Hokuyo lidar. here are the config files:
common_costmap.yaml
global_frame: map
robot_base_frame: base_link
footprint: [[0.42, 0.26], [-0.42, 0.26], [-0.42, -0.26], [0.42, -0.26]]
footprint_padding: 0.03
inflater_layer:
inflation_radius: 0.05
cost_scaling_factor: 0.5
plugins:
- {name: obstacles_layer, type: "costmap_2d::ObstacleLayer"}
- {name: inflater_layer, type: "costmap_2d::InflationLayer"}
global_costmap.yaml:
global_costmap:
update_frequency: 5.0
static_map: true
rolling_window: false
resolution: 0.05
static_layer:
map_topic: /map
subscribe_to_updates: true
lethal_cost_threshold: 60
inflater_layer:
inflation_radius: 0.25
cost_scaling_factor: 0.5
plugins:
- {name: static_layer, type: "costmap_2d::StaticLayer"}
- {name: inflater_layer, type: "costmap_2d::InflationLayer"}
local_costmap.yaml:
local_costmap:
update_frequency: 5.0
publish_frequency: 5.0
static_map: false
rolling_window: true
width: 8.0
height: 8.0
resolution: 0.05
obstacles_layer:
observation_sources: laser_scan_sensor
laser_scan_sensor: {sensor_frame: horizontal_laser_link, data_type: LaserScan, topic: scan, clearing: true, marking: true}
max_obstacle_height: 3
obstacle_range: 10
raytrace_range: 11
If anyone have any Idea why this is happening, you will help me sleep at night.
Best,
Pepe
↧
↧
move_base vs geometry_msgs/PoseStamped md5 error, Electric, deb-install
I've just been trying to run (for the first time, so maybe I just set something wrong) `move_base` together with `gmapping` on ROS Electric (on Ubuntu 10.04). As soon as `move_base` goes up, an error appears:
[ERROR] [1323959697.184014051]: Client [/move_base] wants topic /move_base/goal
to have datatype/md5sum [move_base_msgs/MoveBaseActionGoal/660d6895a1b9a16dce51fbdd9a64a56b],
but our version has [geometry_msgs/PoseStamped/d3812c3cbc69362b77dc0b19b345f8f5].
Dropping connection.
I've installed all ROS packages via apt-get. There has been some update today, so maybe some of the packages weren't rebuilt properly. Has anyone else noticed this problem?
↧
Dynamic reconfigure of costmaps
Hello everyone!
I have a mobile robot in Gazebo with defined dimensions, whose task is to pick-up tables and to transport them to other locations. I am using move_base for path planning and navigation. The robot is navigating properly when there is no load on him, but fails to avoid obstacles when there is a table on it. Obviously, it is because dimensions of tables are not taken into account of costmaps during navigation. The question is following: is there any possibility to dynamically reconfigure or recompute my costmaps during robot motion on move_base?
My first thought was to two parallel running move_base nodes each taking control after some command, but I am not sure if this is even possible, going to try it soon, but any other/better solutions are very appreciated!
↧
Custom costmap values of non-trinary character (move_base with global_planner)
Hello together,
I do have the problem that I would like to create a custom costmap2D layer as described in [http://wiki.ros.org/costmap_2d/Tutorials/Creating%20a%20New%20Layer] that just holds values in a certain range, e.g. [0,100] that when aggregating all layers (static- from SLAM, obstacle- and inflation-layer + custom-costmap) the robot does not just move along the shortest path possible within the former free space. Instead, it would be nice if the robot could explore more of the room based on a user-specific metric, e.g. entropy (how often the robot visited a particular location in the recent timeframe). The overall goal is then to encourage the robot to not take the shortest path from a Euclidean-standpoint but from a mixture of Euclidean distance and maximum room exploration.
Our idea was to compute a potential field externally that reflects which positions are desirable and which are not and feeding this potential to a costmap2d-instance. However, it seems like as if the global_planner in the move_base just discriminates between free, obstacle, and inflation area and does not take values in between into account.
Is there another way to achieve this like modifying the global planner?
Thanks in advance!
↧
↧
Exclude obstacles from Navigation Stack
Hello everyone!
Is there any way to ignore almost static obstacles read from laser scan and sent to navigation stack? My robot's task is to lift objects and navigate with them to other location. Before lift they are defined as obstacles and added to costmap, which is absolutely fine, but after lift they prevent robot from proper navigation. Is there any way to exclude some obstacles from the costmaps?
Their distance is varying in some defined range.
↧
move_base request failed
When running move base with the navigation stack I get the following warning:
[ WARN] [1562249513.118359067]: Request for map failed; trying again...
After that nothing happens anymore. The Logfiles show that all the nodes (amcl, map_server) started correctly and that the map_server was able to find the map. My launch file looks as following:
↧