Simulate IMU data

[weisongwen]

Hi @superjax , Thanks for sharing such a good package. Can this package simulate the raw imu data?

Best,
Weisong,

[superjax]

Yes!

It simulates raw Accel and gyro measurements at each simulation time step with a slowly-wandering offset plus random noise. The parameters for both the random walk and the additive noise are given in the yaml file.

This package also supports custom imu extrinsics (transform from body frame to imu frame) that are also specified in the yaml file as q_b_u and p_b_u (quaternion rotation and position vector from body to imu frame)

More advanced noise models could be added pretty easily, but I haven't seen the benefit yet. Cross-axis alignment or temperature effects would probably the next best things to add.

[weisongwen]

Hi @superjax Thanks for your reply. This is really a good package! BTW, may I know how you simulate the pseudorange and carrier-phase measurements?

Best,
Weisong,

[superjax]

Sure!

I cheated, a little bit.

I collected some raw ephemeris data in an open courtyard of my university with a ublox m8T. This data was post processed a little bit and stored in the samples/ephemeris.dat file. I load up this file and I stimulate the satellites as being exactly where the ephemeris says they are. I believe the data collect is from sometime in 2019, the simulation will always be during this collect.

As for noise models -- I use a klobuchar and saastamoinen model, additive gaussian noise and a simple logic for introducing cycle slips and multipath. Nothing too fancy, but it works for getting the plumbing set up before providing real data.

[weisongwen]

Hi @superjax Thanks for the explanation. Very clear. I will look into this wonderful work first!

Best,
Weisong,

[superjax]

The logic for the multipath is written in my dissertation.

http://hdl.lib.byu.edu/1877/etd11453 page 140