ein_demo
Navigation
In this project, we learning the ein system first. Ein is software for pick-and-place for Baxter. It can map one side of an object with Baxter's wrist camera in about thirty seconds, then use that model to detect the object, localize it to within 2mm and pick it up. Ein use the back language to control the Baxter.
We set up the system and calibrate it with the Ein, program the specific targets for Baxter with the Back Language. Find and pick the object, finish the movement we want.
####step 1: get start Before you start this program, you should install the Ein (see the following link for installation: http://h2r.github.io/ein/install/).
Then use GUN screen to run the project. Go to the root of your catkin workspace and run:
$ BAXTER=your_baxter screen -c src/ein/ein_baxter.screenrc
“your_baxter” is the ID number of your baxter.
if you don't have the screen, you should install it:
$ sudo apt-get install screen
This command will start a screen session preloaded with useful windows. Our program for the right arm is primed in window 0. typing ``1` switches to window 1.
If at any time you need to quit, you can type``:quit` in the screen session.
Next we should calibrate the camera(see the following link for calibration: http://h2r.github.io/ein/calibration/ ).
####step2: picking objects Now you need to create a background map.
goHome tableUpdateBg
then make a model for your object
tableQuickScan tableLock3dGrasp clearClass3dGrasps
you can incrememtally move the arm to the gripping position. for example:
zDown zDown ... xUp yUp zDown
save it
add3dGrasp writeFocusedClass
Now you can note the output folder(catkin_ws/src/ein/default/). You can rename that folder but don't move it. Once renamed to, say, “bottle”, you can reload the object with
endArgs”bottle”setClassLabels
####step3: movement
load the workspace
story import
movement
haveADrink
####step1:
- installation
- get start
- calibration
####step2:
- create background map
- make a model
- picking objects
####step3:
- load a workspace
- finish a movement
<iframe height=498 width=510 src="http://player.youku.com/embed/XMTY3MzIxMjY4NA==" frameborder=0 allowfullscreen></iframe>
###Why can not receiving wrist camera images? you should open the camera with the Baxter command manually.
This command lists the cameras: rosrun baxter_tools camera_control.py -l
Run camera_control.py with no arguments to see usage. Open and close the cameras until only the left and right hand cameras are open. When it is set up correctly you should see this:
$ rosrun baxter_tools camera_control.py -l
head_camera
left_hand_camera - (open)
right_hand_camera - (open)
In the birl_demos/ein_demo of birl_baxter.