pr2_read_text
This package provides code for PR2 reading text using its prosilica camera.Subscribed Topics
prosilica/image_rect_color (sensor_msgs/Image)- Rectified color image from prosilica camera
- PR2 state message, indicate moving or stable
Published Topics
prosilica/text_detect (sensor_msgs/Image)- Output detection result image
Tutorials
The following is the tutorial of making PR2 literate:
First switch prosilica camera into streaming mode. For more information on Prosilica camera: prosilica_camera
rosrun dynamic_reconfigure dynparam set_from_parameters prosilica_driver _trigger_mode:=streaming
Also use dynamic_reconfigure to set frame rate to 1. With lower frame rate, the camera generate less data publishing through the topic. See the dynamic_reconfigure package for details on dynamically reconfigurable parameters. You can also do the previous step together in dynamic_reconfigure GUI.
Then kill sound_play node before running pr2_read_text.
rosnode kill sound_play
We had some problems with sound_play on our PR2, it keep generating very low and deep voice for text-to-speech. So, instead, we use festival directly as a system call, but we have to kill the sound_play otherwise the sound device will be occupied by sound_play node . It's only a temporary solution right now.
Now you can finally making PR2 literate:
roslaunch pr2_read_text read_text_from_pr2.launch
Drive around the PR2 using teleop, and facing it towards text, it will start reading! You can also see the detection result with image_view (default turned on in the launch file)
You can also record the required topics and reproduce the action later using the bag file
roslaunch pr2_read_text read_text_from_bag.launch <bag_file>
