Is there a possibility to calculate the velocity of a car if I just have the GPS data and my acceleration

I already tried the integration of acceleration, but the values I got where not realistic.

for that I used https://de.mathworks.com/help/matlab/ref/cumtrapz.html

[David K.]

What is the time step of your acceleration data? cumtrapz assumes it is 1. If it is smaller than that then your values could be way greater.

Changing your function calls to this may help fix the results to what you are expecting.

vx.data = cumtrapz(ax.time, ax.data); 

[Megan]

My time step is: 0.0204

[David K.]

Ok, so I do expect that that is causing at least some of your issues. If you try the change I proposed does the output look closer to what you are expecting?

Also, What is the plot you are showing? Just one of the velocities or a combination of all three?

Answer 1

[Sid Singh]

Hi, it is possible with just the GPS data alone or just the Accelerometer data. You have to differentiate the GPS data and integrate the Accelerometer data to get raw velocity and ideally they should be the same.

You can compare the two velocities and look for a similarity.

Usually the data has noise so you may want to use a filtering technique like Kalman filtering which works best if your noise is gaussian or Particle filtering if your noise model can't be estimated.