added documentation on launch manual

README.md

 # Motion Grammar Example
 
-For documentation, please see https://jastadd.pages.st.inf.tu-dresden.de/motion-grammar-example
\ No newline at end of file
+For documentation, please see https://jastadd.pages.st.inf.tu-dresden.de/motion-grammar-example
+
+# Running the Demo
+
+## Repos into directories (inkl submodules):
+  - web-interface: git@git-st.inf.tu-dresden.de:ceti/ros-internal/technical-tests/ccf_web_interface.git (branch mg-demo)
+  - motion-grammar-controller: git@git-st.inf.tu-dresden.de:jastadd/motion-grammar-example.git
+  - workspace: git@git-st.inf.tu-dresden.de:ceti/ros-internal/ccf/ccf_workspace.git
+
+## Manual
+
+- Start the Robot
+  - to start the robot disconnect it from power, reconnect after ~20s
+  - use a static IP connection to connect to the robot via ethernet using `172.31.1.13` or `172.31.1.15` or `172.31.1.17`
+    - for setting see [wiki](https://git-st.inf.tu-dresden.de/stgroup/stwiki/-/wikis/Projects/Ceti/Panda/Panda-Robot-Configuration-Data)
+  - Log in to Panda Desk in the browser at [172.31.1.13/](https://172.31.1.13/) or [172.31.1.15/](https://172.31.1.15/) or [172.31.1.17/](https://172.31.1.17/)
+    - using user `st` and password `st-cobot-1`
+    - accept security risks (if necessary)
+  - Unlock the joints
+    - if the robot is in pack position, unfold it (this is the only option anyway)
+  - Activate FCI Mode
+
+- Plug in the webcam
+  - this is only tested for the Trust webcam
+  - The webcam must be [calibrated](http://wiki.ros.org/camera_calibration). For the Trust model, do the following:
+    - create a file `~/.ros/camera_info` with the following content:
+      ```yaml
+      image_width: 2560
+      image_height: 1440
+      camera_name: head_camera
+      camera_matrix:
+        rows: 3
+        cols: 3
+        data: [1813.222796360344, 0, 1261.930645515087, 0, 1817.396430837666, 719.5808597823312, 0, 0, 1]
+      distortion_model: plumb_bob
+      distortion_coefficients:
+        rows: 1
+        cols: 5
+        data: [0.08032097481606813, -0.1159678601922493, 0.005085141441063858, -0.0007346072111796328, 0]
+      rectification_matrix:
+        rows: 3
+        cols: 3
+        data: [1, 0, 0, 0, 1, 0, 0, 0, 1]
+      projection_matrix:
+        rows: 3
+        cols: 4
+        data: [1834.692138671875, 0, 1258.756588840144, 0, 0, 1843.024536132812, 725.2023077170961, 0, 0, 0, 1, 0]
+      ```
+  - figure out the device name of the camera with `v4l2-ctl --list-devices`, if the list shows two devices for the Trust camera, it's most likely the first on, you can check with `v4l2-ctl -d /dev/video2 --all` - it should show a lot of configuration options
+
+- Make sure a MQTT server is running
+
+- start ROS (in the `workspace` directory)
+  - in every console:
+    - make sure code is compiled (`catkin build`)
+    - make sure ROS environment is loaded (`source devel/setup.bash`)
+  - load robot controller
+    `roslaunch ccf_immersive_sorting robot-cell_ceti-table-1-emptyworld-grey.launch robot_ip:=<ROBOT-IP>`
+  - load tag detection
+    `roslaunch ccf_immersive_sorting tag_detection.launch device:=/dev/video0` (or another video device, most likely `/dev/video2` if you connected the webcam after startup)
+  - load the virtual scene controller
+    `roslaunch ccf_immersive_sorting virtual_scene_provider.launch`
+  - objects.yaml to adapt scene + update yaml.file bash to generate json
+
+- start the control web interface (in the `web-interface` directory)
+  - *once* install all requirements: https://git-st.inf.tu-dresden.de/ceti/ros-internal/technical-tests/ccf_web_interface/-/blob/master/README.md?ref_type=heads
+  - run `node index.js`
+  - open the interface in the browser at http://localhost:3000/
+  - switch to *Selection Mode*
+
+- start the motion grammar (in the `motion-grammar-controller` directory)
+  - `./gradlew run`
+
+- start the motion grammar web interface (in the `motion-grammar-controller/dshboard` directory)
+  - if running for the first time, build the interface with `npm build`
+  - `npm start`
+