Hi Khang,
The documentation indicates that during co-simulation, Gazebo is stepped first, followed by Simulink.
We see here that the sensor data and actuation commands are exchanged at the correct time step, but at each time step, Gazebo is one step ahead of the simulation. This sequence might lead to the observed delay in the received joint angles. The relevant section of the documentation may be accessed using the following link:
To synchronize simulation times between Gazebo and Simulink, the “Gazebo Pacer” block may be used as suggested in the following documentation:
This example showing how to set up a synchronized simulation between Simulink and Gazebomight be helpful:
I hope this helps!
