New Feature: Extended Kalman Filter State Estimator for SimpleFlight

[subedisuman]

Fixes: #

About

In the following branch, I have implemented an Extended Kalman Filter (EKF) based state estimator. It was motivated by a feature request comming from AirSim to have an EKF based state estimator in the SimpleFlight firmware. The contribution in this pull request uses the following sensor measurements: IMU, GPS, Barometer, and Magnetometer to estimate the following states: local (NED) x-y-z positions, x-y-z velocities, attitudes, and IMU and barometer sensor biases.

How Has This Been Tested?

• Verified that in absence of sensor noises and following the same integration scheme as used in FastPhysicsEngine, the ground truth states can be replicated by the EKF estimated states.
• Ran multiple flight maneuvers and plotted the estimated states against the true states with sufficient accuracy.

Screenshots (if appropriate):

Please follow this issue for more details.
The updates can be found in the issue. There is also a demo to try things out and see results in plots :). Please contact me for questions, assistance, or feature requests either in the issue or here. Thank you.

[xxEoD2242]

Hey @subedisuman, AirSim is being archived and no new features are being added.

I have built a fork and we are actively integrating new features into the system here:Colosseum. Feel free to make a PR there if you like or I can pull this one through your branch directly.

[VctorAHernndez]

Hi @subedisuman, not sure if you're familiar with this fork of AirSim called Project AirSim (https://github.com/iamaisim/ProjectAirSim) which essentially ports/extends some of the functionality from the original repository. It's also a bit more active than other ports/forks of AirSim.

I think adding the possibility of extract estimated states through EKF is of great value, particularly with the same level of granularity you've implemented here (i.e. with sensor modularity kept in-mind). I'm currently working on a fork of Project AirSim and am looking to integrate some of your code into it. I think the open source community would benefit from your contributions if you were to port it to Project AirSim. Otherwise, I'd like to continue your work there!

Thanks for your help.