Currently, the `dt` of the simulation is constant and equal to 1 second. It would be great to make it variable, handling agent movements.