tgibbons's profile - activity

Ok, I think I forgot a step when moving maps. I think the map.yaml file hardcodes the location of the pgm file in it. Try editing the .yaml file with gedit or some other text editor. The first line should be the location of the .pgm file. You may have to change this line to match the new location

Make sure you are saving the map when you create it, that is step 4 of the tutorial you linked to above.

rosrun map_server map_saver -f /tmp/my_map

This should create two files associated with the map you create, my_map.yaml and my_map.pgm. If you list the files in /tmp you should see these two files.

The next thing to check is which pc you are running amcl_demo.launch on and which pc you saved the maps. If I remember the tutorials correctly, the maps are saved on the turtlebot's laptop, which amcl_demo.launch is run on the remote control laptop. We generally have to manually transfer the map files from one machine to the other for this to work. That is if you are running roslaunch turtlebot_navigation amcl_demo.launch on the remote control laptop and not ssh'ing into the turtlebot, then make sure the map files are on the remote control laptop, called the PC in the tutorials.

If you move the two map files to a new location, you will need to edit the .yaml file since it hardcodes the full path of the .pgm file. You can use gedit or any other text editor on the .yaml map file. The first line of this file should be the location of the .pgm file which you need to update.

I have a launch file which starts up a node for the pocketsphinx voice recognizer:

  <node name="recognizer" pkg="pocketsphinx" type="recognizer.py" output="screen" >
    <param name="lm" value="$(find rbx1_speech)/config/turtle_commands.lm"/>
    <param name="dict" value="$(find rbx1_speech)/config/turtle_commands.dic"/>   
 </node>

This works fine when I run it locally on my turtlebot, but when I ssh into the turtlebot and run it, I get the following error messages:

(recognizer.py:3629): GConf-WARNING **: Client failed to connect to the D-BUS daemon:
Unable to autolaunch a dbus-daemon without a $DISPLAY for X11

** (recognizer.py:3629): WARNING **: gst_gconf_get_string: error: No D-BUS daemon running

So the node doesn't stay up. I didn't think pocketsphinx was doing anything graphical, so I don't understand why it might need an X windows display. I'm not familiar with the D-BUS daemon or what it does, so I'm a little lost.

Anyone know if I should be able to run the recognizer node from pocketsphinx through ssh? Anyone familiar with this type of error?

I am running ROS indigo on Ubuntu 14.04. Been search for solutions or explanations and read through the recognizer.py without gaining any insight.

I noticed an issue with the voice control example, voice_nav.py, in pocketsphinx. I'm new to ROS and don't know who to tell to suggest a fix. Could someone tell me how to go about notifying someone about this issue?

Here is the issue as I understand it. In pocketsphinx, voice_nav.py publishes on the /cmd_vel topic to move the robot. I think this works fine for the original turtlebot, but the turtlebot2 with a Kobuki base needs this published on /cmd_vel_mux/input/navi topic.

To get the pocketsphinx samples running on my turltebot2, I had to change line 44 of voice_nav.py from

self.cmd_vel_pub = rospy.Publisher('cmd_vel', Twist, queue_size=5)

to

self.cmd_vel_pub = rospy.Publisher('cmd_vel_mux/input/navi', Twist, queue_size=5)

But I think this fixes it for the turtlebot2, but breaks it for the original turtlebot.

So, how do I go about notifying someone of this issue?

I ran into this same problem. As mentioned above, there seems to be a bug in say.py which is part of sound_play. If you want to temporarily get around this bug while they fix it, you can edit say.py replacing these two lines

    print 'Volume: %s' % volume
    soundhandle.say(s, voice, volume)

with

    #print 'Volume: %s' % volume
    soundhandle.say(s, voice)

This simply comments out the printing of the volume and does not pass it to soundhandle.say() which only expects the two parameters in the current version.

It is generally not a good idea to modify code in /opt/ros/indigo/lib/sound_play directly, but I did anyway to avoid moving all of sound_play to a work space. I had to chmod the file permissions which restrict permissions in this folder, but thought it was OK since I want later updates to overwrite this fix.

I am trying to use the current version of rospeex on a Tutrtlebot 2 running Ubuntu 14.04 Trusty and ROS Indigo. I have tried the apt package install and also the complied from the current source.

In all cases I am having problems with the audio monitor when I run rosrun rospeex_audiomonitor audio_monitor_epd

The audio monitor window starts up and displays the waveform from the mic, but soon turns gray and become inactive, that is non of the UI works. The console dislays:

   state= 0
   state=2
   state=0
   =====================EPDON

I have been going through the related launch files and source files and cannot see anything wrong. I have also tried different microphone sources.

This has worked vary briefly once, but that is not repeatable. This makes me worried it is something with the rospeex server since this is a cloud based solution.

Is anyone else using rospeex currently and can verify that the service is up?

I have considered switching to the Google speed API listed in the source code, but this seems to have been discontinued since I don't see it listed in my available Google APIs.

Any suggestions on what else to try or other speech recognition systems for ROS?

I am working on a configuration for my fall AI/Robotics undergraduate course. I would like to find the best version of Ubuntu and ROS for use on both NXT and Turtlebot2. Here is my problem:

Currently I am testing Ubunto 12.04 (Precise) with ROS Hydro. I think this will work fine with Turtlebot, both on the bot and the workstation. But I also want to use Lego Mindstorm NXT to get the students started. It looks like ROSNXT needs ROS Electric, but ROS Electric will not run on Ubunto 12.04.

So, it looks like I might have to go back to Ubunto 10.04 (Lucid) and ROS Electric to get support for ROS NXT, but Turtlebot2 is not supported in ROS Electric (you need Groovy)

So, I don't see a single configuration that supports ROS NXT and Turtlebot 2... Does that seem correct?

   -- I could upgrade to Lego Mindstorm EV3 for the fall, if ROS Hydro will support it by September...?  
   -- Should I try building ROS NXT from source under Hydro, but this looks like it has challenges
   -- Is it possible to run ROS Hydro under Ubuntu 10.04? It does not look like it

Any suggestions?