DECISION

DECISION Ground on the Loop Autonomy Planning Tool

Conda Environment

DECISION uses Conda to maintain dependancy versions. Follow the instruction here to set up the DECISION conda environment.

Submodules

Run the follow commands to recursively collect and update submodules. bash cd DECISION git submodule init git submodule update --recursive If encountering an error, download the submodule directly from the repository (https://github.com/JPLMLIA/OWLS-Autonomy.git) and put it in decision/OSIA/OWLS-Autonomy/

Building the ACME OSIA Docker container

DECISION runs the ACME OSIA algorithm inside a docker container. To build this container run the following commands. bash cd DECISION/decision/OSIA/OWLS-Autonomy/ docker build . -f Dockerfile_ACME --tag owls-autonomy-acme:v1

Increasing Docker Memory

Default RAM allocated to Docker containers can be too low if you're attempting to optimize many ACME observations at once. RAM allocation can be increased from the Docker desktop application under gear icon->Resources->Memory. We recommend running the ACME Docker container with 10-12GB of RAM if your system allows.

DECISION Demo Data

Demo data to use with DECISION can be found at https://ml.jpl.nasa.gov/projects/decision/ACMEDemoData.zip. Unzip this repository and place it in DECISION/decision/data/.

Installing Dakota

DECISION utilizes Dakota for parameter optimization. Follow instructions at https://dakota.sandia.gov/installation/ to install Dakota.

Testing DECISION

Verify installation by running DECISION unit tests. This is accomplished with the following commands. Test coverage reports will be generated in docs/coveragereport/. View docs/coveragereport/index.html for a summary of testing results. bash cd DECISION/decision/ pytest -rP --ignore=OSIA --disable-warnings --cov=decision --cov-report html:docs/coverage_report test/ -v

Running DECISION

Start DECISION using the following commands. bash cd DECISION/decision/ python decision.py

From a web browser visit http://127.0.0.1:8050/ to see the DECISION application. We recommend Google Chrome for optimal compatibility.

Copyright

Copyright 2023, by the California Institute of Technology. ALL RIGHTS RESERVED. United States Government Sponsorship acknowledged. Any commercial use must be negotiated with the Office of Technology Transfer at the California Institute of Technology.

This software may be subject to U.S. export control laws. By accepting this software, the user agrees to comply with all applicable U.S. export laws and regulations. User has the responsibility to obtain export licenses, or other export authority as may be required before exporting such information to foreign countries or providing access to foreign persons.