Home        Store        Docs        Blog

Ideal measurements for SITL testing

(barain) #1

The SITL works fine; thanks to @patrickelectric tutorials and answers to different questions on the forum, which helps to run the simulations seamlessly. One of the objective is to use the ideal measurements from sensors (or configure the noise in sensor measurements) to test the overall system. Furthermore, we have noticed that the EKF switched to different modes. Please have a look at the attached screen shots:

The parameters used in the simulations are from @patrickelectric bluerov2_gazebo_sitl.param

The questions are as follows:

  1. For high level testing in simulations, we assume that the low-level controller and sensors are calibrated and works fine. How can we achieve this in SITL testing? Is it possible to force this capability?
  2. One of the main objective for many users of SITL to vary the noise in the sensors. How can we change the sensor noise parameters?
  3. We want to have as many sensors as possible for SITL testing. It can be seen in one of the screen shot that Compass is not install and Gyroscope is not present. Furthermore, it will be greatly appreciated if anyone can guide about getting the vision/camera data in QGC or Gazebo SITL.

Any advice or guidance would be greatly appreciated.

(Patrick José Pereira) #2


I’m glad that the forum posts was useful for you.

Some points about your simulation:

  1. bluerov2_gazebo_sitl.param should only be used if you are doing a simulation with Gazebo. This can explain why your mission is a bit weird. You can check the following picture that I did with a fresh clone of ArduSub.

  2. EKF will change the state only in the first seconds, you are probably having this messages while doing the simulation because of the wrong parameters.

About your questions:

Do you want to do a calibration in SITL ? What are you trying to accomplish with it ? SITL will run all sensors calibrated.

It’s possible, check SIM_ parameters.

This can be done by improving the Gazebo model and including such signals/connection with SITL via socket. For that you’ll need to study and modify with the source code. Any pull request is much appreciate.
I just check it here, and I can see the sensors in QGC, reset your parameters and undo any modification.


(barain) #3

Thanks @patrickelectric for the comprehensive answer.

  1. The simulations using QGroundControl works fine using bluerov_sitl_fast.param.

The sensors works fine as you have described. As you have mentioned, the problem was the parameters used for the simulation.
2. For other users who might be interested in this topic, the calibrations are not required in SITL and it is good to know that the simulations will run all sensors calibrated, provided we use the correct parameters.
3. Sensor noise characteristics and new sensors can be added, as mentioned in the answer above. Much appreciated for guidance where to look in in the system to add this capability.

Many thanks for help.