imclab / polaris

The Extended Kalman Filter (EKF) for aircraft

Polaris - inertial navigation system

Polaris is open-source inertial nagivation software (INS) for fixed-wing aircraft. It provides realtime estimates to allow for higher-level navigation software to be developed.

For installation instructions (you will need to follow these if you want to get it working), more details and screenshots go to http://dronedynamics.com/polaris/.

Basics

Polaris is a software INS that runs against X-Plane out of the box but can also be used on real hardware.

Polaris is designed to take the following inputs:

• 3-axis accelerometer
• 3-axis gyro
• 3-axis magnetometer (including magnetic declination lookup table)
• Static air sensor (altitude)
• Differential air sensor (indicated airspeed)
• GPS
• Various temperatures (both on the sensors and ambient)

And turn those inputs into a realtime probabilistic model with the following estimates:

• Orientation (pitch, roll, and yaw)
• Position (longitude, latitude, and altitude)
• Speeds (indicated airspeed, true airspeed, speed over ground)
• Wind (x/y/z axis windspeed estimates which help make a more accurate dead-reckoning estimate)

Features

Polaris is able to build and maintain its realtime probabilistic model primarily using an (extended Kalman filter](http://en.wikipedia.org/wiki/Extended_Kalman_filter) (EKF). There are two major steps to the EKF: predicting your current state and updating your model based on what actually happened.

Right now Polaris uses Euler angles but in the future it may use quaternions.

More to come in this section…

Todo

1. Local magnetic declination lookup table
2. Get course over ground working (emulated or hpath)
3. Get cpress working (true airspeed / indicated airspeed ratio)
4. Code up alternate algorithms such as DCM and benchmark it against the EKF
5. Quaternion-based EKF
6. More advanced display/logging
7. Clean up website, installation instructions
8. Make a screencast for setting it up and explaining it
9. Port to other platforms
10. Email Austin Meyer (X-Plane) and see if he's willing to include raw magnetometer values
11. Once all the algorithms are verified (including a write-up on the testing perhaps), introduce sensor noise
12. Your idea here ✈ [email protected]

Issues

Have a bug or feature request? Please create an issue here on GitHub!

https://github.com/dronedynamics/polaris/issues

Authors

Ryan Beall

Tim Trueman

• http://twitter.com/timtrueman
• http://timtrueman.com/
• http://github.com/timtrueman

Add your name above this line if you send a pull request…

License

Copyright 2011 Tim Trueman and Ryan Beall

Licensed under the Apache License, Version 2.0: http://www.apache.org/licenses/LICENSE-2.0