burtbick's profile - activity

I've been working with Indigo on Linux Mint 17.1 but am in the process of upgrading to Mint 18(.1) so I've installed kinetic. I've gotten the motion bits working and am working on the Speech Synthesis and Recognition parts next.

PocketSphinx has been a bit more of a hassle. There was a poacketsphinx ROS package for Indigo but not for Kinetic. So I pulled in the Indigo package and using an example from Gstreamer for Gstreamer 1.0 / Pocket Sphinx I was able to get it working, kind of. It is recognizing words / phrases from my dictionary but it is NO where near as accurate as with the older version on Indigo.

Has anyone done anything with PocketSphinx on Kinetic, and if so how did you improve the accuracy? My reworked recognizer.py file works fine, except for the accuracy. The sample code I based it off of uses a different approach from the original recognizer.py code, so that may be the issue.

Or is there a better speech recognition alternative that you are using with Kinetic and newer versions of ROS?

Thanks,

I was using actionlib.SimpleActionClient('sound_play', SoundRequestAction) under ROS Indigo and it was working fine for doing speech and sound output. Basically I had a python script that subscribed to a topic that received the output from pocketsphinx, parsed the recognized text, executed commands and spoke associated text.

This worked just fine under Indigo, but the same exact code doesn't produce ANY speech or wave file output under Kinetic.

I've determined that if I do the actionlib related processing in the main of the python script that it does work fine, but attempting to do the actionlib sound_play from the context of the callback produces no output. No errors, and sound_play reports that it has spoken or played the wav file.

Has anyone else run into this with Kinetic? There is some more information below.

I had a similar problem with OpenCV with Kinetic. It appears that callbacks may be happening on a different thread and that is breaking things. With the OpenCV issue the imshow() function would not return, no errors and nothing drawn in the window. To resolve that, at least short term, I changed the code to defer the creation of the OpenCV windows so that it happened in the context of the callback function(s). A similar approach with the actionlib processing has so far failed to resolve the issue.

I'm just getting started with ROS and ultimately want to do some experimentation with CV and physical object manipulation, but I have miles to go before I'm at that point.

I have been working with MyRobotLab but Oracle Java on my 64bit Linux machine is unstable, even after completely nuking all things Java and doing a clean reinstall. When things work it works reasonably well, but too many hiccups to be efficient.

Also issues on a different machine running Win10 and Java there. So I decided to move to ROS which has a steeper learning curve, but appears to be much more stable.

For starters I have a "head" with two cameras for eyes, servos driving the eyes in X and Y and servos driving the head in X and Y.

I have an Arduino in the head that interfaces to the servos to control them. Currently no position feedback from the servos.

In MyRobotLab the Arduino code allows specifying Acceleration and Velocity values for the servos by using the writeMicroseconds() method to drive the servo motion. I have implemented this and have it interfacing with ROS and I'm able to move the servos and set Velocity and Acceleration just fine.

The other part of the MRL implementation is also to be able to specify Min and Max values for the Servo to prevent driving it past points of physical constraint in the head. For example with the eyes the Y axis movement is quite small so the servo range of motion should be limited.

MRL does this with an actually SetMinMax() method and also with a mapping function that allows more control over the relationship of the input and output Min/Max for the servo(s).

So I wondered what people are doing with ROS to accomplish similar things when using Arduino to control your servos?

Is there some code that exists that manages the MinMax() limits for servos. Or will I need to write my own code.

Right now I'm working towards a small application to drive the head and eye motion in a pre-programmed mode as my next set of tests.

Once I get that working (I already have Python code for doing this with MyRobotLab) then it's on to looking at interfacing with OpenCV to do some initial simple object tracking as I have been able to do with MRL (on the days when Java cooperates:)

If I have missed some suitable code/tutorial/examples could someone point me in the right direction?

And I would be interested to hear if others are doing anything similar with your Arduino based servo control?

Burt

I must be blind, don't see a mark as solved option. Do I just Close the question to mark as solved? Thanks,

Just had a chance to download and test Arduino IDE 1.8.4 and that does work fine with rosserial so must have been someth

Nothing special. As stated Linux Mint 18.1 with Arduino 1.8.3 I notice that you said that it works fine with Arduino 1

I'll have to go through the code again to refresh, but I think that it was mainly fixing up recognizer.py I''l do some m

I'll have to blow the cobwebs off. I've been off working on other things since I resolved the issue. Basically I had t

rosserial_arduino with Arduino 1.8.3 I was going to build the Arduino code on my new installation of Linux Mint 18.1 and

What I did to get this working again with Kinetic was to create a new node that wraps the actionlib processing, so in my

actionlib appears to work differently with Kinetic I was using actionlib.SimpleActionClient('sound_play', SoundRequestAc

This may not be the complete answer to what is going on, but what I did to resolve the issue was to move the creation of

Last bit of data. Looking closely on Indigo I'm using Python 2.7.6 with Open CV 3.1.0 But on Kinetic it is Python 2.7.

This problem is why I thought I was not getting the raw_image messages. Everything looks OK as far as the frame image d

A bit more information. Something appears to have changed between what I have with Indigo and Kinetic. That is in the

Since this fails with the unmodified ROS by example code here's a github URL to the code. https://github.com/pirobot/rb

Code is not currently hosted anywhere external. I've done some more digging and there is something strange happening.

But with the sub to input_rgb_image the input_rgb_image topic does not appear in the topic list. I guess I'll have to m

More info. launch file is doing <remap from="input_rgb_image" to="/camera/rgb/image_raw"/> Code is doing: self.im

My code is NOT getting /camera/rgb/image_raw topic messages even though I can use rostopic echo to see that the messages

A little more information. I found a camshift python example (not integrated with ROS) and that works fine using cv2.Vi

camshift Python example not working with kinetic Hi, Had the 'ROS by example' camshift example working fine with Indigo

I got pulled in a different direction for a day or so, but I hope to compare the two this evening and will let you know.