Motion Generator Simulation Help?

Hello,

I"m trying to generate a specific motion profile for testing from a GPS path / google earth type GIS file.

The best documentation I've found on this is this link.

I'm essentially trying to recreate what was done in the "driver_example" - could someone please describe the workflow that generated that dataset?

Thanks,
Adam

Hi Dong,

Thanks for that - yes, I manged yesterday to simulate a drive yesterday from a google earth file by (a long) trial. In my part of the country it is very flat and altitude varies only slightly so it was close enough for me to ignore.

I wound up using this tool:

https://acscdg.com/

Plotted a map, exported CSV, and generated relative (CMD3) heading changes, also rough times from distance/velocity. At that point is learned the difficulty of turning 😉 and the strong dependence on velocity, bearing change and command duration. From there I just hand tweaked the results iterating, each change and looking on a map. Worked well enough for my purposes. It was also after this learning experience I learned what my request would entail.

Also, google earth pro can do this, with alt, just make a path, and save as KML and import, you get lat/lon/alt out and initial loc/bearing. There are a few open matlab/python libs that can support finding the bearing and arc lengths. Creating a tool that can go between that and a trajectory / motion simulation is quite another task. Certainly interesting!

Thanks for the link (I hadn't seen it) and the explanation! I know appreciate what I was asking for!

I'd be interested to know if there were any open source tool sets you have used in the past for this type of thing?

There is also explanations on some simple motion profiles in our online simulation demo. You can refer to: https://developers.aceinna.com/simulationhelp

Hi Adam,

Basically, the motion profile is generated by trial, especially for a drive simulation because you need to keep the generated trajectory aligned with a real road.
It is difficult to generate a motion file from a GPS path:

  1. Attitude information is not included, which is required to generate IMU measurements;
  2. If you want acceleration from velocity or position of a GPS path, differentiating is necessary. The GPS path points are usually collected at a relative lower sampling rate, which IMUs are usually sampled at very high rate. Besides, differentiating will increase measurement noise a lot. Maybe oversampling and smoothing the GPS path may help.

We provide a drive motion profile as motion_def-long_drive.csv. You can refer to that.

Thank you for feedback.

Dong

Log in to reply