The On-board Artificial Intelligence Research (OnAIR) Platform

The On-board Artificial Intelligence Research (OnAIR) Platform is a framework that enables AI algorithms written in Python to interact with NASA's cFS. It is intended to explore research concepts in autonomous operations in a simulated environment.

Generating environment

Create a conda environment with the necessary packages

conda env create -f environment.yml

Redis example

Using the redis_adapter.py as the DataSource, telemetry can be received through multiple Redis channels and inserted into the full data frame.

OnAIR config file (.ini)

The redis_example.ini uses a very basic setup:

• meta : redis_example_CONFIG.json
• parser : onair/data_handling/redis_adapter.py
• plugins : one of each type, all of them 'generic' that do nothing

Telemetry config file (.json)

The telemetry file defines the subscribed channels, data frame and subsystems.

• subscriptions : defines channel names where telemetry will be received
• order : designation of where each 'channel.telemetry_item' is to be put in full data frame (the data header)
• subsystems : data for each specific telemetry item (descriptions only for redis example)

Receipt of telemetry

The Redis adapter expects any published telemetry on a channel to include:

• time
• every telemetry_item as described under "order" as 'channel.telemetry_item'

All messages sent must be json format (key to value) and will warn when it is not then discard the message (outputting what was received first). Keys should match the required telemetry_item names with the addition of "time." Values should be floats.

Running the example

If not already running, start a Redis server on 'localhost', port:6379 (typical defaults)

redis-server

Start up OnAIR with the redis_example.ini file:

python driver.py onair/config/redis_example.ini

You should see:

Redis Adapter ignoring file

---- Redis adapter connecting to server...

---- ... connected!

---- Subscribing to channel: state_0

---- Subscribing to channel: state_1

---- Subscribing to channel: state_2

---- Redis adapter: channel 'state_0' received message type: subscribe.

---- Redis adapter: channel 'state_1' received message type: subscribe.

---- Redis adapter: channel 'state_2' received message type: subscribe.

***************************************************
************    SIMULATION STARTED     ************
***************************************************

In another process run the experimental publisher:

python redis-experiment-publisher.py

This will send telemetry every 2 seconds, one channel at random until all 3 channels have recieved data then repeat for a total of 9 times (all of which can be changed in the file). Its output should be similar to this:

Published data to state_0, [0, 0.1, 0.2]
Published data to state_1, [1, 1.1, 1.2]
Published data to state_2, [2, 2.1, 2.2]
Completed 1 loops
Published data to state_2, [3, 3.1, 3.2]
Published data to state_1, [4, 4.1, 4.2]

And OnAir should begin receiving data similarly to this:

--------------------- STEP 1 ---------------------

CURRENT DATA: [0, 0.1, 0.2, '-', '-', '-', '-']
INTERPRETED SYSTEM STATUS: ---

--------------------- STEP 2 ---------------------

CURRENT DATA: [1, 0.1, 0.2, 1.1, 1.2, '-', '-']
INTERPRETED SYSTEM STATUS: ---

--------------------- STEP 3 ---------------------

CURRENT DATA: [2, 0.1, 0.2, 1.1, 1.2, 2.1, 2.2]
INTERPRETED SYSTEM STATUS: ---

--------------------- STEP 4 ---------------------

CURRENT DATA: [3, 0.1, 0.2, 1.1, 1.2, 3.1, 3.2]
INTERPRETED SYSTEM STATUS: ---

--------------------- STEP 5 ---------------------

CURRENT DATA: [4, 0.1, 0.2, 4.1, 4.2, 3.1, 3.2]
INTERPRETED SYSTEM STATUS: ---

Running unit tests

Instructions on how to run unit tests for OnAIR

Required python installs:

pytest, pytest-mock, coverage

Optional python install:

pytest-randomly

Running the unit tests from the driver.py file

From the parent directory of your local repository:

python driver.py -t

A few optional settings for the driver.py file

Options that may be added to the driver.py test run. Use these at your own discretion.

--conftest-seed=### - set the random values seed for this run --randomly-seed=### - set the random order seed for this run --verbose or -v - set verbosity level, also -vv, -vvv, etc. -k KEYWORD - only run tests that match the KEYWORD (see pytest --help)

NOTE: Running tests will output results using provided seeds, but each seed is random when not set directly. Example start of test output:

Using --conftest-seed=1691289424
===== test session starts =======
platform linux -- Python 3.11.2, pytest-7.2.0, pluggy-1.3.0
Using --randomly-seed=1956010105

Copy and paste previously output seeds (or type them out) as the arguments to repeat results.

Running pytest directly from command line

For the equivalent of the driver.py run:

python -m coverage run --branch --source=onair,plugins -m pytest ./test/

Command breakdown:

python -m - invokes the python runtime on the library following the -m coverage run - runs coverage data collection during testing, wrapping itself on the test runner used --branch - includes code branching information in the coverage report --source=onair,plugins - tells coverage where the code under test exists for reporting line hits -m pytest - tells coverage what test runner (framework) to wrap ./test - run all tests found in this directory and subdirectories

A few optional settings for the command line

Options that may be added to the command line test run. Use these at your own discretion.

--disable-warnings - removes the warning reports, but displays count (i.e., 124 passed, 1 warning in 0.65s) -p no:randomly - ONLY required to stop random order testing IFF pytest-randomly installed --conftest-seed=### - set the random values seed for this run --randomly-seed=### - set the random order seed for this run --verbose or -v - set verbosity level, also -vv, -vvv, etc. -k KEYWORD - only run tests that match the KEYWORD (see pytest --help)

NOTE: see note about seeds in driver.py section above

To view testing line coverage after run:

NOTE: you may or may not need the python -m to run coverage report or html

coverage report - prints basic results in terminal or coverage html - creates htmlcov/index.html, automatic when using driver.py for testing

then <browser_here> htmlcov/index.html - browsable coverage (i.e., firefox htmlcov/index.html)

License and Copyright

Please refer to NOSA GSC-19165-1 OnAIR.pdf and COPYRIGHT.

Contributions

Please open an issue if you find any problems. We are a small team, but will try to respond in a timely fashion.

If you would like to contribute to the repository, GREAT! First you will need to complete the Individual Contributor License Agreement (pdf). Then, email it to [email protected] with [email protected] CCed.

Next, please create an issue for the fix or feature and note that you intend to work on it. Fork the repository and create a branch with a name that starts with the issue number. Once done, submit your pull request and we'll take a look. You may want to make draft pull requests to solicit feedback on larger changes.