Getting Started

Getting Started#

The #bayrob system can be executed in multiple ways. To try out the web app locally, just follow the instructions in Use Web App in Docker to start the application in a docker container. To inspect and use the code in your own project, follow the instructions in Compile from Source. To generate and run your own docker container after updating the code (e.g. after adding more action models), see Generate Docker Container.

Compatibility#

This software should work on most Linux distributions. It has not been tested on Windows machines.

Use Web App in Docker#

Create a file docker-compose.yaml with the following contents

version: '3'

services:
  bayrobweb:
    image: "bayrob-web-img"
    container_name: bayrob-web-container
    working_dir: /bayrob-dev/src/bayrob/web
    command: "python3 server.py -p 5005 -i 0.0.0.0"
    environment:
      - "PYTHONUNBUFFERED:1"
    ports:
      - "5005:5005"

and (pull and) start the container using the command:

$ sudo docker compose up

Open the following address in a web browser:

http://127.0.0.1:5005/bayrob/

See For Users for instructions on how to use the web application.

Compile from Source#

The source code is publicly available under BSD License. Check it out with

$ git clone https://github.com/mareikep/bayrob.git

Prerequisites#

  • Python >= 3.8

  • Additional python dependencies, listed in requirements.txt

NOTE: Amongst others, the system requires pyjpt and pyrap, which are included in the 3rdparty folder as precompiled wheels. While pyjpt is essential in #bayrob since it provides the base functionality for probabilistic reasoning, pyrap is only required for the use of the web application. You can install all required python dependencies (including the two precompiled wheels) via pip:
$ pip install -r requirements.txt

If a newer version of pyjpt is required, since the framework is in constant development, update it via pip or download its source code from GitHub.

Usage#

The module :class:bayrob.core.base.BayRoB is the entry point for using the system. The two methods .search_astar() and .query_jpts() allow querying individual models or run the search algorithm to refine action plans. See sections :ref:Reasoning and :ref:Plan Refinement in For Developers for examples and the API for additional information on the module’s functions.

Generate Docker Container#

If additional models were integrated into the system (following the instructions in Learning), the Docker container needs to be rebuilt to allow querying them in the (local) web application. A Dockerfile and a docker-compose.yaml file are provided with the source code and can be found on root level of the git checkout. The Makefile contains the commands to build and compose the container. Run

$ sudo docker build --tag bayrob-web-img .

or

$ make build

in the root directory of the system and

$ sudo docker compose up

or

$ make run

to start it. As described above, the webapp can then be accessed via web browser using the address http://127.0.0.1:5005/bayrob/.

See For Users for more information and a user manual of the web app. The added models should now appear in the dropdown field of the BayRoB Query window. If they are not shown, check, if you placed the files in the correct folder. Note, that only files in subfolders of examples/demo will be loaded in the web application.

Contents