This is the repo for new code for the iCub crawling. The code is under development :)

CrawlManager:

• it is the main manager module, used for high-level interface and control of the limb controllers (generators)
• input port /check_port/ gets the speed and the status from the generators
• output port /parts_port sends the parameters to each part (amplitude, stance, swing etc.)
• reads parameters from a configuration file crawling_managerConfig.ini which is currently in icub_crawling/app/contexts/crawling/
• when launched, it is expecting the generators (x6, one for each limb) to be already running and connected to the robot. if they are not running, it gives a warning but the module starts

CrawlGenerator:

• lower level module generating the motion patterns, using parameters given by the CrawlManager module
• this class is instanciated once for each robot part :
  • the two arms
  • the two legs
  • the head
  • the torso each instance needs a parameter file describing the limb to be controlled
• through connexions to other instances of this class, it couples the motion of the body parts_port
• the actual velocity control of the actuators is delegated to the velocityControl module to which it sends the desired velocities
• one instance of CrawlGenerator needs a corresponding one of velocityControl

For the three modules:

cd iCub_crawling/modules/crawlManager
mkdir build
cd build
ccmake ..
make

Where "crawlManager" has to be replaced by the name of the module to be compiled.

To make sure the configuration file crawling_managerConfig.ini is read correctly by the module, add these lines to .bashrc (if you are on Linux) or .bash_profile (if you are on OSX).

export crawling_root=PATH_WHERE_YOU_PUT_THE_CRAWLING_PROJECT/iCub_crawling
export YARP_DATA_DIRS=$YARP_DATA_DIRS:$crawling_root/app/
export PATH=$PATH:$crawling_root/app/contexts/crawling

To allow the programs to be executed without the need to explicitly type their path, one has also to update the PATH environment variable as follows in .bashrc (if you are on Linux) or .bash_profile (if you are on OSX).

export PATH=$PATH:$crawling_root/modules/crawlGenerator/build:$crawling_root/modules/crawlManager/build

It is possible not to do this, but then yarpmanager won't be able to find the crawlGenerator modules.

Finally, in order to create a crawling iCub model in Gazebo, do the following.

go to icub_gazebo folder

cp -r PATH_WHERE_YOU_PUT_THE_CRAWLING_PROJECT/gazebo_addin/icub_crawling/ PATH_WHERE_YOU_PUT_THE_GAZEBO_PLUGIN/icub-gazebo/

For now, the ports the crawlGenerator connects to are hard-coded in the source. Thus, it would only work with iCub_SIM out of the box but one could change the port names in generatorThread.cpp to use the Gazebo simulator or the actual robot. Also the application description file crawling.xml should be adapted.

How to use it on iCub:

1. start the yarp server with yarpserver
2. start the gazebo simulator and insert the iCub into the world
3. move to the app folder
4. launch the yarp manager with the application description for the crawlGenerator module

   yarpmanager --application yarpmanager/crawling-gazebo.xml
   

   One only needs to open the Crawling applicaion (couble-click on it) and launch it by pressing the green button with a play-like shape.
5. launch crawlManager (from modules/crawlManager/build)

The terminal is used to choose the appropriate action to do:

1. init pos
2. crawl straight
3. faster
4. slower
5. turn right
6. turn left
7. stop (also closes the manager module)