Wrong heading angle reading by M8U when tight turn maneuvers and also when left running overnight

GPS/GNSS Development Boards
1

Been using NEO-M8U to read the Heading angle.

After calibrating the fusion mode (Dead Reckoning) and reading right values when driving, I tried to park my car.

2 issues:

1- when doing very slow maneuvers and tight turns (left and right) to park the car, I noticed there is a 10 deg discrepancy. If you go directly to the parking spot it reads the heading right, but if you go back and forth and do very tight turns at the end of the parking, the heading drifts by max 10 deg.

2- I let the car sit reading 280 deg heading and left the car over night while the sensor was ON the whole time. The sensor reading in the morning was 260. So 20 deg drift over night.

I am using the following command to read the heading:

myGNSS.getHNRheading();

To me, 20 deg drift overnight make these sensors not usable for heading. Is this normal? Is there anyway to make these sensors more reliable and have a better reading?

Thanks

2

Hi,

To be completely honest, I think you are expecting too much of the M8U’s IMU. Based on what you have described, I would say it is performing very well. All IMUs (especially the Gyros) drift over time and with temperature.

If the parking spot is covered, that further complicates the measurement as it can not use the GNSS signal to help correct the heading.

I am sure the heading will correct itself as soon as you move the vehicle under clear sky and it can use the GNSS signal to reestablish the heading?

You may find the ZED-F9R performs better, especially if you can connect up the wheel tick signals.

I hope this helps,

Paul

3

Setting the “Dynamic Model” to DYN_MODEL_AUTOMOTIVE may also help. Please see:

https://github.com/sparkfun/SparkFun_u- … cModel.ino

4

Thanks for the response.

I did not know 20 deg drift is normal when stationary.

my parking spot was not covered. so I am thinking the GNSS signal was still there and good to be used by the algorithm.

Does the position of Antenna with respect to the sensor matter? to explain the tight turns and slow maneuvers drift?

5

Hi,

The M8 receiver description contains a lot of information - see section 29:

https://www.u-blox.com/sites/default/fi … 003221.pdf

u-blox also have some webinars and “white papers” which describe “dead reckoning” in more detail. They contain links to other papers too:

https://register.gotowebinar.com/record … 0663945730

https://www.u-blox.com/sites/default/fi … adr%29.pdf

https://www.u-blox.com/sites/default/fi … 187%29.pdf

I think for your application, adding “wheel ticks” (sensing the rotation of the car wheels) will make the biggest improvement.

I hope this helps,

Paul