[mems] speed prediction in the tunnel

[Tiel]

Hello,

I have question about gps speed prediction in the tunnel.

When gps loses signal in the tunnel, we can not get any speed info from gps.

In that case, can we predict speed using mems? If yes, how to solve that?

Thanks in advance for your advice.

Best regards

Tiel

[Peter BENSCH]

It is not easy, but you can estimate the speed with the appropriate effort if you know the initial velocity. An accelerometer only recognises the vectors of acceleration, so you can calculate the instantaneous speed based on the initial velocity. However, this becomes increasingly inaccurate with increasing time, so that you can only bridge a short period of time with it for justifiable accuracy.

You will certainly find some ideas on the Internet using your favourite search engine and search words such as "calculate speed accelerometer".

Hope that helps?

Good luck!
/Peter

[Federica Bossi]

Hi @Tiel ,

You could look at our Teseo, which is not a mems but it could be useful for you!

If this helps you, please mark my answer as "Best Answer" by clicking on the "Accept as Solution" button, this can be helpful for Community users to find this solution faster.

[Tiel]

To Mr.Peter Bensch,
First of all, thank you for your response and sorry for the delay in replying to your kind/fast answer.
I had a development jobs and production accident at last week, so delayed. sorry for late response.
Anyway, I also considered your way firstly, but I am wondering that it works well. Because I already had tried same way(compensating with a short-term integration using the ACC sensor) in the past when I was fresh man personally , but I didn't get satisfactory result. So now I am considering Kalman filter, but I'm not sure if it would work on no-input situation. Anyway, do you have any example case of using a Kalman filter with ST mems or acc-sensor?

[Tiel]

To Fredesica Bossi,

Thanks for your answer.

Actually, I'm not sure if you mean that I consider to use teseo module to develop my stuff or if you mean for me to find the meaningful documentation at teseo link. Anyway, looking at the brief-specs of teseo, there doesn't seem to have deep technical details. According to that brief-spec, I can think that teseo module seems to apply kalman filter using embedded 6-axis sensor to conpensate gps speed and position errors. It would be helpful if the system physics equations were shared, but is there any more details I might analyze it mode in detail?

[raptorhal2]

My 2 cents:

Calibrate the accelerometer for offset and gain while in GPS mode, then switch to integrating calibrated accelerometer data in the tunnel to get speed changes. Integration will filter accelerometer noise.

[Tiel]

Thanks for your response. 

I have question on your idea. I think simple integration might cause quite big accumulative errors in a short time. How about avoiding this accumulative errors? Do I have to add LPF prior to integration calculation?

[raptorhal2]

Any LPF will simply put a lag on the answer, it will not eliminate acceleration data errors.

You are dependent on the quality of the IMU calibration while in GPS mode.