Best practices for testing, simulation before hardware test?

asked 2020-08-31 22:56:03 -0600

mugetsu gravatar image

updated 2022-09-11 11:26:59 -0600

lucasw gravatar image

Because of Covid, a lot of my work has been remote and doing many tests on gazebo robot instead of real hardware.

My robot uses rtabmap, robot_localization, all of which can be tested in simulation with gazebo sensors. However, there's a lot of work to make parameters work for hardware.

My thought for the next step is to gather rosbags of sensor data live from the hardware, and then play it in simulation, to have more fidelity with simulation.

This would give me a more realistic expectation of what to expect on hardware and to troubleshoot issues without having to run hardware.

Is there a term for this? Are there any other similar techniques that would: -minimize troubleshooting on real hardware -narrow differences between simulation and hardware

edit retag flag offensive close merge delete