TODO:

• Add CI
• Add Tests

Mission Chicken Soup

• Mission Chicken Soup
• Getting Started
• Dowload Repo
• Create Project
• Update Controller
• How to Compile
• How to Run
• What Results to Expect

Mission Chicken Soup is a multi agent system in which one agent converts the soup ingredients into soup particles. The second agent takes the soup particles and checks for inconsistent ingredients that need to be filtered before final soup delivery.

The first agent is a publisher whereas the second agent is a subscriber.

ZMQ TCP protocol example usage and options are in how to run. First, lets go over how to get started.

Getting Started:

• Getting started can be daunting given the time it takes to build of GAMS, MADARA and ACE. We assume that you have GAMS and MADARA environment variables set appropriately. In my context, this is how the .bashrc file looks like

  export MPC_ROOT=/home/bikram/MPC
  export EIGEN_ROOT=/home/bikram/eigen
  export CAPNP_ROOT=/home/bikram/capnproto
  export MADARA_ROOT=/home/bikram/madara
  export GAMS_ROOT=/home/bikram/gams
  export VREP_ROOT=
  export ZMQ_ROOT=/usr/local
  export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$MADARA_ROOT/lib:$GAMS_ROOT/lib:$VREP_ROOT:$CAPNP_ROOT/c++/.libs
  export PATH=$PATH:$MPC_ROOT:$VREP_ROOT:$CAPNP_ROOT/c++
  
  export CORES=4
  export MY_INSTALL_DIR=/home/bikram/ace
  export ACE_ROOT=$MY_INSTALL_DIR/ACE_wrappers
  export TAO_ROOT=$ACE_ROOT/TAO
  export LD_LIBRARY_PATH=$ACE_ROOT/lib:$TAO_ROOT/lib:$LD_LIBRARY_PATH
  export PATH=$ACE_ROOT/bin:$TAO_ROOT/bin:$PATH
  
  

• Disclaimer: In the beginning I found it confusing regarding installation and pre-requisites between MADARA, GAMS and ACE. Also, ended up looking into deprecated documentation and video tutorials that are largely not relevant any more. I suggest anyone who attempts to get started with using MADARA, GAMS refer to repository wiki page explicitly through GAMS wiki. MADARA wiki documentation is not clear for installation process. I later learnt that ACE is not required as long as MADARA is installed as part of GAMS and we use action.sh script to build our projects. In short, install GAMS, use gpc.pl to create projects and use action.sh to build and run, the rest of the magic follows.

In the following text, we documented the process for creating and editing files to get the results in this repository.

Dowload Repo

You may choose to download the repository and build and run it instead of following through story of how this was done in the first place.

mkdir -p $HOME/gams_ws
git clone https://github.com/bikramA/mission-chicken-soup.git
cd mission-chicken-soup
./action.sh compile

Next, skip how to run to test.

Create Project

We assume that you have installed GAMS and MADARA and their corresponding pre requisites already.

mkdir -p $HOME/gams_ws
cd $HOME/gams_ws
$GAMS_ROOT/scripts/projects/gpc.pl --path mission-chicken-soup  --container MissionChickenSoup mission-chicken-soup/action.sh compile

Update Controller

Major code implementation of this mission is in src/controller.cpp. The following key steps were taken to create this controller.

• Reviewed MADARA documentation to find a sample code for ZMQ TCP type transport.
• An example was found in /madara/tests/transports/network_profiler.cpp
• This code was used as reference to create the mission-chicken-soup controller.cpp

The controller generates random numbers pertaining to the size of soup ingredients from 0 to 1000. The publisher agent takes the soup ingredients and converts a fraction of them into soup particles of size less than 100. The subscriber agent takes all soup particles and filters fine soup particles from large granules. The subscriber agent aggregates the soup particles and publishes soup conversion ratio in fraction.

How to Compile

• It was initially difficult to comprehend errors when using mwc.pl with gnuace option as suggested in the default documentation. GAMS documentation to create and build project came in handy.

cd $HOME/gams_ws/mission-chicken-soup
./action.sh compile

• And that's it. Assuming, you didn't run into compile time errors.

How to Run

Before running the agents, here are the description of options that we use

• [-i|--id id] the id of the agent ( 0 is publisher, rest is subscriber)
• [--zmq|--0mq proto://ip:port] a ZeroMQ endpoint to connect to. examples include tcp://127.0.0.1:30000 or any of the other endpoint types like pgm://. For tcp, remember that the first endpoint defined must be your own, the one you are binding to, and all other agent endpoints must also be defined or no messages will ever be sent to them. Similarly, all agents will have to have this endpoint added to their list or this karl agent will not see them.

For more options and descriptions type

$HOME/gams_ws/mission-chicken-soup/bin/custom_controller help

• Run the publisher in one terminal

cd $HOME/gams_ws/mission-chicken-soup/bin
./custom_controller -i 0 --zmq tcp://127.0.0.1:30000 --zmq tcp://127.0.0.1:30001

• Run the subscriber in another terminal.

cd $HOME/gams_ws/mission-chicken-soup/bin
./custom_controller -i 2 --zmq tcp://127.0.0.1:30001 --zmq tcp://127.0.0.1:30000

What Results to Expect

• After sometime expect to see the following results on the publisher terminal

Publishing soup packets in 100 variables @5 hz for 30 s on 0MQ transport
Publisher is done. Check results on subscriber.

Knowledge in Knowledge Base:
ingredient_size0=90
ingredient_size1=939
ingredient_size10=90
ingredient_size11=90
ingredient_size12=788
ingredient_size13=90
...
...
ingredient_size99=90
num_ingredients=100

• The knowledge base is also saved as a context file in bin/recent_context.kkb.

• After sometime expect to see the following results on the subscriber terminal

Receiving for 30 s on 0MQ transport
Test: SUCCESS
Settings:
  Transport type: 0MQ
  Hosts: 
    tcp://127.0.0.1:30001
    tcp://127.0.0.1:30000
  Soup Contents recieved: 121 
  Soup Conversion Ratio: 0.752066 

Factory How to Instruction that I could not get to work:

COMPILE ON LINUX:

 mwc.pl -type gnuace workspace.mwc
 make vrep=1

COMPILE ON WINDOWS:

 mwc.pl -type vc12 workspace.mwc 

<Open Visual Studio and compile project>. Note that you should compile the
solution in the same settings you used for ACE, MADARA, and GAMS. For most
users, this is probably Release mode for x64. However, whatever you use
make sure that you have compiled ACE, MADARA, and GAMS in those settings.

RUN THE SIMULATION:

open VREP simulator
perl sim/run.pl