-
Notifications
You must be signed in to change notification settings - Fork 21
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Contacting you #1
Comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I would love to! But you dont provide any way to contact you?
I have used a number of ways to try and get rid of noise from GPS-speed to build a speedometer that also calculates the current engine power. This is done by, at the end, differentiation and taking into account drags (air, static [speed dependend]) as well as potential energy (change in height), all calibrated by letting the car roll and thus knowing the initial kinetic energy (that has a real name in this sort of science as I found out, but I forgot it again). Since I have to work with the differences even noise <1% of the value (eg. +-0.3km/h noise at 100 km/h speed) still results in completely useless values. Since I want this to be somewhat realtime there is no simple way to get rid of that noise. A standard Kalman Filter does work, but it has hard limitations. Mainly that its inefficient unless the new sample has a low weight, at which point its "averaging" so much that normal driving operations are smoothed out too much and when you come to a stop it still takes seconds befor speed or power goes down to zero. Or you step on it and it just doesnt show up. So right now im living with this system that still shows noise and has a delay but I want to improve it. But all the more complex Kalman Filters are above me, im just a "simple engineer" and have no idea how to implement something like that (the matrix stuff etc. and then also in C++ and not just on paper).
So now here I am and found your library after hours of reading on this topic and seeing that the UKF might just do what I want. Since I already have a lot of raw data for testing It would also be good to just "feed" it to the algorithm to see how good it works. But I have no idea how I could do that. With the simple Kalman Filter I just did it in Excel, super easy. Also helps to tune the parameters since I can instantly see the effect, which is a huge problem if I had to implement it into my car over and over again and drive around to test the parameters... But this is not ur problem, its time to get your library running.
Keep up the good work!
PS: You have a bunch of other algorithms like the UKF of which I only know the EKF. Are the others better or worse for such a applications? Potentially adding data from a accelerometer etc. and combining those to get it more real-time and reduce noise would also be great, but thats another topic...
The text was updated successfully, but these errors were encountered: