FlightGear + Jupyter + Bokeh

Streaming live data from FlightGear to plots in a Jupyter notebook

Andreas Klostermann

Demo Video: https://www.youtube.com/watch?v=aXNy1r5UDEE

• @BayesianHorse
• www.github.com/akloster/jupyter-flightgear

This is a demonstration of how to use the flexibility of Jupyter and Bokeh to create rich and impressive data science environments.

FlightGear is a very mature and sophisticated open source flight simulation for multiple platforms. It offers several ways to monitor the variables of the simulation, one of which is a websocket interface, which can be easily integrated into the IPython kernel framework.

In order to run this notebook you need a recent Anaconda install for Python 3.5 with Bokeh and Jupyter. You also need a resonably recent FlightGear installation (3.6 for example).

In the demonstration video I used this command line to start FlightGear:

./run_fgfs.sh --fg-aircraft=/media/internal_2/aircraft --disable-ai-models  --prop:/sim/ai-traffic/enabled=false --prop:/sim/traffic-manager/enabled=false  --disable-enhanced-lighting --disable-clouds3d --disable-clouds3d --geometry=500x500 --httpd=9095

A couple of the options have to do with optimizing for screen capture. The only important option is "--httpd=9095" which enables the internal webserver.

This notebook file can be viewed either just-as-is, with all the notes and details, or as a presentation with live-reveal.