High level question

[ValentinFunk]

Hi there, I stumbled upon this repo through github search. I'm trying to build a small module to help with a small annoyance that I have: when driving a small boat under a bridge the GPS gets lost for a short time and during that time the position doesn't update on screen. Not a huge issue but I thought it'd be cool to fix. Plan is to use accelerometer data to estimate the position. Could this be achieved with the code here?

[sergehog]

Hi Valentin.
I afraid my code is not 100% finalized, and not ready for real use case.
Also, in this project I was aiming at 6 degrees of freedom (DoF) dead reckoning, which is probably too much for your need.
For surface vehicles you usually need to estimate only 3 DoFs. And for the marine applications, like yours, you might also need to add one-or-two DoF in order to take waves into account.
Additionally, If you only have an accelerometer but no gyroscope, that could be a problem too. Uncertainty will grow too fast.

So, if you need mathematically rigorous solution, then you would need to create your own Kalman Filter with all the knowledge of your boat and its motion, and also making use of all available sensors.

If you need some easy solution, I would recommend to use something simple, like LMS/NLMS adaptive filter https://en.wikipedia.org/wiki/Least_mean_squares_filter