java-player-users

[Java-player-users] Problem with resource allocation while using Planner Interface

[Salvatore F. <ei...@gm...>]

Hello everyone,

i am using in 2 different threads, a Fiducial Detector, and a Planner Interface.

The fiducial detector, gives me the id of the detected fiducial, as well as the position of the robot that detected it (position returned from the fiducial itself was less accurate, so i decided to use the one of the robot).

This position, is then used by another robot as a goal in a PlannerInterface (robot has to move to the position the fiducial was detected).

The second robot, has a fiducial detector running at the same time (All robots move in the environment, while detecting fiducials).

The problem is, that while the fiducial is detected, and data is fed to the second robot's planner, the robot cannot get to the goal point.

Although things seem to work properly, the robot at some point fails to adjust its speed, and carries out a course, usually crashing on to an obstacle.

I have noticed that when this happens, my cpu usage reaches 100%, and in the terminal window where the planner configuration is run, i get an output like the following :



warning : requested cell (1684,21) is offmap
warning : requested cell (1685,21) is offmap
warning : requested cell (1686,21) is offmap
warning : requested cell (1687,21) is offmap
warning : requested cell (1688,21) is offmap
warning : requested cell (1689,21) is offmap
warning : requested cell (1690,21) is offmap
warning : requested cell (1691,21) is offmap
warning : requested cell (1692,21) is offmap
warning : requested cell (1693,21) is offmap
warning : requested cell (1694,21) is offmap
warning : requested cell (1695,21) is offmap
warning : requested cell (1696,21) is offmap
warning : requested cell (1697,21) is offmap
warning : requested cell (1698,21) is offmap
warning : requested cell (1699,21) is offmap
warning : requested cell (1700,21) is offmap
warning : requested cell (1701,21) is offmap
warning : requested cell (1702,21) is offmap
warning : requested cell (1703,21) is offmap
warning : requested cell (1704,21) is offmap
warning : requested cell (1705,21) is offmap



In the java console, my output seems normal :

Sep 28, 2011 9:01:40 AM javaclient3.MapInterface handleResponse
WARNING: Data count doesn't match cells number: 2431 != 250000
Sep 28, 2011 9:01:40 AM javaclient3.MapInterface handleResponse
INFO: Map decompress: 250000 bytes
Robot Navigation1@testPlatform New Goal : (0.6018865726369702,-6.377219609508479,-0.7258935634496316)
Robot Navigation1@testPlatform map.isDataReady() true
Robot Navigation1@testPlatform MapInfo = javaclient3.structures.map.PlayerMapInfo@9485d
Robot Navigation1@testPlatform Map data ready!
Robot Navigation1@testPlatform pmd.toString()javaclient3.structures.map.PlayerMapData@16dd2b6
Robot Navigation1@testPlatform PlayerPlannerData = javaclient3.structures.planner.PlayerPlannerData@1e22a7e
Robot Navigation1@testPlatform Still on path


The map, amcl, wavefront and vfh configurations i use are the following :


driver
(
  name "mapfile"
  provides ["7000:map:0"]

  filename "bitmaps/cave.png"
  resolution 0.032 # meters per pixel
  negate 0
  origin [-16 -16 ] # real-world location of the bottom-left-hand corner of the map
)


driver
(
  name "amcl"
  provides ["7000:localize:0" "7000:position2d:2"]
  requires ["odometry::6665:position2d:0" "7000:laser:0" "laser::7000:map:0"]

 odom_init 0

  init_pose [-6.432 -5.895 0]
)


driver
(
  name "vfh"
  provides ["7000:position2d:1"]
  requires ["6665:position2d:0" "7000:laser:0"]

  max_turnrate_0ms 90
  max_turnrate_1ms 35
)

driver
(
  name "wavefront"
  provides ["7000:planner:0"]
  requires ["output::7000:position2d:1" "input::7000:position2d:2" "7000:map:0"]

  safety_dist 0.4 
  distance_epsilon 1.0
  angle_epsilon 10
  alwayson 1
)


If anyone has an idea as to what causes the problem, you are welcome to give some insight.
Thanks in advance

Re: [Java-player-users] Problem with resource allocation while using Planner Interface

[Jorge S. S. <jsa...@gm...>]

Hello Salvatore,

I think the problem is with player, not javaclient. Have you posted
the question on pla...@li... list?

I´m quite lost with your problem, but maybe it´s related with the
mapfile parameter:

  resolution 0.032 # meters per pixel

It must be exactly the physical size of the world divided by the
number of pixels.

Sorry for not being more helpful



2011/9/28 Salvatore Famoso <ei...@gm...>:
> Hello everyone,
>
> i am using in 2 different threads, a Fiducial Detector, and a Planner Interface.
>
> The fiducial detector, gives me the id of the detected fiducial, as well as the position of the robot that detected it (position returned from the fiducial itself was less accurate, so i decided to use the one of the robot).
>
> This position, is then used by another robot as a goal in a PlannerInterface (robot has to move to the position the fiducial was detected).
>
> The second robot, has a fiducial detector running at the same time (All robots move in the environment, while detecting fiducials).
>
> The problem is, that while the fiducial is detected, and data is fed to the second robot's planner, the robot cannot get to the goal point.
>
> Although things seem to work properly, the robot at some point fails to adjust its speed, and carries out a course, usually crashing on to an obstacle.
>
> I have noticed that when this happens, my cpu usage reaches 100%, and in the terminal window where the planner configuration is run, i get an output like the following :
>
>
>
> warning : requested cell (1684,21) is offmap
> warning : requested cell (1685,21) is offmap
> warning : requested cell (1686,21) is offmap
> warning : requested cell (1687,21) is offmap
> warning : requested cell (1688,21) is offmap
> warning : requested cell (1689,21) is offmap
> warning : requested cell (1690,21) is offmap
> warning : requested cell (1691,21) is offmap
> warning : requested cell (1692,21) is offmap
> warning : requested cell (1693,21) is offmap
> warning : requested cell (1694,21) is offmap
> warning : requested cell (1695,21) is offmap
> warning : requested cell (1696,21) is offmap
> warning : requested cell (1697,21) is offmap
> warning : requested cell (1698,21) is offmap
> warning : requested cell (1699,21) is offmap
> warning : requested cell (1700,21) is offmap
> warning : requested cell (1701,21) is offmap
> warning : requested cell (1702,21) is offmap
> warning : requested cell (1703,21) is offmap
> warning : requested cell (1704,21) is offmap
> warning : requested cell (1705,21) is offmap
>
>
>
> In the java console, my output seems normal :
>
> Sep 28, 2011 9:01:40 AM javaclient3.MapInterface handleResponse
> WARNING: Data count doesn't match cells number: 2431 != 250000
> Sep 28, 2011 9:01:40 AM javaclient3.MapInterface handleResponse
> INFO: Map decompress: 250000 bytes
> Robot Navigation1@testPlatform New Goal : (0.6018865726369702,-6.377219609508479,-0.7258935634496316)
> Robot Navigation1@testPlatform map.isDataReady() true
> Robot Navigation1@testPlatform MapInfo = javaclient3.structures.map.PlayerMapInfo@9485d
> Robot Navigation1@testPlatform Map data ready!
> Robot Navigation1@testPlatform pmd.toString()javaclient3.structures.map.PlayerMapData@16dd2b6
> Robot Navigation1@testPlatform PlayerPlannerData = javaclient3.structures.planner.PlayerPlannerData@1e22a7e
> Robot Navigation1@testPlatform Still on path
>
>
> The map, amcl, wavefront and vfh configurations i use are the following :
>
>
> driver
> (
>  name "mapfile"
>  provides ["7000:map:0"]
>
>  filename "bitmaps/cave.png"
>  resolution 0.032 # meters per pixel
>  negate 0
>  origin [-16 -16 ] # real-world location of the bottom-left-hand corner of the map
> )
>
>
> driver
> (
>  name "amcl"
>  provides ["7000:localize:0" "7000:position2d:2"]
>  requires ["odometry::6665:position2d:0" "7000:laser:0" "laser::7000:map:0"]
>
>  odom_init 0
>
>  init_pose [-6.432 -5.895 0]
> )
>
>
> driver
> (
>  name "vfh"
>  provides ["7000:position2d:1"]
>  requires ["6665:position2d:0" "7000:laser:0"]
>
>  max_turnrate_0ms 90
>  max_turnrate_1ms 35
> )
>
> driver
> (
>  name "wavefront"
>  provides ["7000:planner:0"]
>  requires ["output::7000:position2d:1" "input::7000:position2d:2" "7000:map:0"]
>
>  safety_dist 0.4
>  distance_epsilon 1.0
>  angle_epsilon 10
>  alwayson 1
> )
>
>
> If anyone has an idea as to what causes the problem, you are welcome to give some insight.
> Thanks in advance
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2dcopy1
> _______________________________________________
> Java-player-users mailing list
> Jav...@li...
> https://lists.sourceforge.net/lists/listinfo/java-player-users
>

Re: [Java-player-users] Problem with resource allocation while using Planner Interface

[Salvatore F. <ei...@gm...>]

Hi Jorge,

I saw that in a forum earlier, and set the pixels to 800 and resolution to 0,04 in order to get a floorplan of 16x16.
Unfortunately the problem was still there.
I sent a slightly altered copy of my message to this list to playerstage-users.
Will wait to see if anyone has any idea, while working on other parts of the code, hoping something will come up.
I think it might have to do something with the fiducial detector and the planner running simultaneously.
At some point i used a blobfinder to detect differently colored robots as landmarks, and this wasn't happening.
However, i couldn't set a blob id, so i gave a shot to the fiducial, which was not working for me some time ago (is now).
I was kind of forced to change to fiducial, so as to improve scalability (many more fiducial ids available than colors -> many more distinct landmarks i can use).
Anyway, i hope that either by experimenting, or using someones advice, i will solve this.

By the way, regardless of whether you are able to help on a specific matter or not, the fact that you try to is a big thing by itself, and i really appreciate it.

Thanks


On Sep 28, 2011, at 6:15 PM, Jorge Santos Simón wrote:

> Hello Salvatore,
> 
> I think the problem is with player, not javaclient. Have you posted
> the question on pla...@li... list?
> 
> I´m quite lost with your problem, but maybe it´s related with the
> mapfile parameter:
> 
>  resolution 0.032 # meters per pixel
> 
> It must be exactly the physical size of the world divided by the
> number of pixels.
> 
> Sorry for not being more helpful
>