sym-cps

Symbiotic Design for Cyber Physical Systems

Project description available here

Documentation available here

Update November 22

You can evaluate random designs generated by the grammar via command line

pdm run generate_random --n=500 --n_wings_max=0

Generates 500 new designs with no wings and evaluates them.

Updates October 22

Make sure that in your eval_pipeline there is this branch. If you have problems with the submodule, just clone it inside eval_pipeline.

After pulling the latest main branch from this repo and challenge_data repo, you can launch these commands to have a clean installation:

make uninstall

sudo rm -r eval_pipeline

git clone -b features/oct-22-fdm-update --single-branch https://github.com/LOGiCS-Project/swri-simple-uam-pipeline.git ./eval_pipeline

pdm install

Make sure that the challenge_data is updated to the latest version and then run again:

sudo pdm run suam-config install --no-symlink --input=../challenge_data/data/broker.conf.yaml

If the file exists already on your system, delete it and run the command again.

Local Installation

System Requirements

1. pdm
2. Access to this repo.

Quick start

1. Clone the two repos and their submodules in the same folder

   git clone --recurse-submodules https://github.com/uc-berkeley-data-discovery-2022/berkeley-cps-symbiotic-design.git

   git clone https://github.com/uc-berkeley-data-discovery-2022/challenge_data.git

2. Navigate to berkeley-cps-symbiotic-design folder and install dependencies

   cd berkeley-cps-symbiotic-design

   You don't have to create a virual environment with pdm, you can launch the command:

   pdm config python.use_venv false

   and then

   pdm install

Configure AWS

1. Download the aws-cvpn-config.ovpn configuration file from the challenge_data repo repo and use it to connect to the AWS VPN. Here are the instructions for Linux , MacOS and Windows.

2. Create the folder ../challenge_data/aws if it does not exist already.

   1. (Optional) notify your IDE to exclude the aws folder from being indexed. For example in PyCharm do Right Click on aws - Mark directory as - Excluded

3. Mount the shared folder in ../challenge_data/aws. Instructions here. Make sure you are connected via the VPN before mounting the shared drive.

   1. Suggestion: from the root folder of the repo, you can try one of the following commands, according to your OS and preferences
      1. sudo mount -t nfs 10.0.137.113:/fsx/ ../challenge_data/aws
      2. sudo mount_nfs -o resvport 10.0.137.113:/fsx/ ../challenge_data/aws
      3. sudo mount -o resvport -t nfs 10.0.137.113:/fsx/ ../challenge_data/aws

4. After have successfully installed the dependencies and activated the conda environment. Launch the following command from the root of the repo

   sudo pdm run suam-config install --no-symlink --input=../challenge_data/data/broker.conf.yaml

Docker Installation

System Requirements

1. Docker
2. Access to this repo.

Quick start

1. Clone the two repos and their submodules in the same folder

   git clone --recurse-submodules https://github.com/uc-berkeley-data-discovery-2022/berkeley-cps-symbiotic-design.git

   git clone https://github.com/uc-berkeley-data-discovery-2022/challenge_data.git

2. Navigate to berkeley-cps-symbiotic-design folder and launch docker script.

   cd berkeley-cps-symbiotic-design & \
   ./docker_run_old.sh bash

You can now launch the scripts and the python files. For example:

pdm run init

To initialize the library of components and seed design

pdm run custom-design "working/test_quad_abstraction_3"

To generate a custom design from json file (this script does not send it to evaluation)

pdm run evaluate "test_quad_cargo"

To generate a custom design from json file, send it to evaluation and retrieve the results. The results appear in the challenge_data/output/<DESIGN_NAME> automatically (a json file with the evaluation results and an STL file with the 3D model).

Any change you make to the files in berkeley-cps-symbiotic-design will be immediately available inside the docker container.

Remote deployment from IDE

The docker image contains an ssh server that you can connect from your IDE and run your code/debug directly inside docker without the need of installing any dependencies.

Launch:

  ./docker_run_old.sh

to launch the docker image as a new container running in background.

Then connect to the python interpreter via SSH on port 9922 of localhost as explained below:

PyCharm

• Go to: Preference > Build, Execution… > Deployment
• Add new SFTP.
• Give a name, e.g. : docker-sym-cps.
• Add a new SSH configuration:
  • Host: localhost
  • Port: 9922
  • Username: root
  • Password: password
  • Root path: /root
• Mappings:
  • Local path ‘<LOCAL_PATH>/berkeley-cps-symbiotic-design’
  • Deployment path ‘/ide’

Got to: Preferences > Project… > Python Interpreter > Add Interpreter > On SSH and select Existing Connection the SSH connection previously created, then select the “System Interpreter” as Python runtime configuration, i.e. the interpreter at /usr/bin/python3’. “Sync folders” insert →/root/ide`

On Python Interpreter, click on Show all and select the Remote Python interpreter that you have just created. On the top right of the window clock on Show Interpreter Paths, and add a new path /root/host/__pypackages__/3.10/lib.

Run

Launch Python scripts from command line

When running a python script from command line, insert pdm run before. For example:

pdm run python src/sym_cps/examples/library.py

Run

Launch Python scripts from command line

When running a python script from command line, insert pdm run before. For example:

pdm run pdm run python src/sym_cps/examples/library.py

Working with IDEs

With PEP 582, dependencies will be installed into pypackages directory under the project root. With PEP 582 enabled globally, you can also use the project interpreter to run scripts directly. Check pdm documentation on PEP 582.

PYCHARM Add __pypackages__/3.10/lib and src folders to your PYTHONPATH. With PyCharm you can simple right click on the folders and select Mark Directory as - Source folder.

VSCODE

To configure VSCode to support PEP 582, open .vscode/settings.json (create one if it does not exist) and add the following entries:

{
  "python.autoComplete.extraPaths": ["__pypackages__/3.10/lib"],
  "python.analysis.extraPaths": ["__pypackages__/3.10/lib"]
}

Examples

Check the example folder and look at the code of the python files to get familiar with the API. Here's some example you can launch.

Populate the library of components from the data folder

pdm run python src/sym_cps/examples/library.py

Extracts the seed designs

pdm run python src/sym_cps/examples/designs.py

Create a new design from scratch. First choosing and topology and then concretize it.

pdm run python src/sym_cps/examples/topology.py

Evaluate designs from command line specifying the path of the design_swri,json in <design-file>:

sudo pdm run suam-client direct2cad.process-design --design=<design-file> --results=../challenge_data/aws/results

For example:

pdm run suam-client direct2cad.process-design --design=../challenge_data/output/designs/TestQuad/design_swri_orog.json --results=../challenge_data/aws/results

Evaluate designs from python

pdm run python src/sym_cps/examples/evaluation.py

Web interface

Run the back-end by executing backend/app.py

Install react dependencies via npm by running:

npm install --legacy-peer-deps

inside the frontend folder.

Then run the front-end:

npm run start

Troubleshooting

Dot not found on your PATH

Mac users:

brew install gprof2dot

Installation process

If you have problems with the environment you can clean up the files and folders created by pdm and conda by running

make uninstall

Then try installation process again.

pdm install

Submodules

Make sure the submodules are correctly imported and updated

git submodule init
git submodule update --

Make sure that eval_pipeline folder is not empty. If the submodule was not pulled correctly you can simply clone this repo and copy its content inside eval_pipeline folder.

Apple Silicon

For Mac M1 users, if you have architecture problems (i.e. some packages are not available for arm), you can try simulating x86_64 architecture and install the dependencies again.

env /usr/bin/arch -x86_64 /bin/zsh

However, all the dependencies right now should support arm and you should not change architecture.

To run a x86 terminal by default in VSCode, add the following to your settings.json

"terminal.integrated.profiles.osx": {
   "x86 bash": {
       "path": "/usr/bin/arch",
       "args": ["-arch", "x86_64", "/bin/bash"]
   }},
"terminal.integrated.defaultProfile.osx": "x86 bash"

To run a x86 terminal by default in PyCharm, go to Preferences/Tools/Terminal and set the shell path to be:

env /usr/bin/arch -x86_64 /bin/zsh --login

Features and Credits

Fully typed with annotations and checked with mypy, PEP561 compatible