Tune the Dispatching Node to use all available robots instead of queuing task on the same robot (#528)

Open-RMF Open-RMF General
1

Posted by @Odin-byte:

I am finishing up my RMF implementation on a small fleet of our custom robots, but I am not satisfied with the way tasks are currently dispatched to the fleet.

If there are multiple robots available (all having the same power level) the first task is dispatched to the robot with the shortest way to the starting point of the task, which is what I am expecting. But as soon as new task come in, it seems like that the first 2 or 3 task all get queued for the same robot instead of utilizing the remaining robots waiting idle.

Around the 4th dispatched tasks an additional robot is awarded a task.

Is there a way to tune the dispatching to utilize the remaining robots earlier? Or is there any setting within my fleet adapter that is causing this behavior?
If you need more specific information to point out my problem, I am happy to report more.

Thanks for your time!

2

Posted by @Odin-byte:

Could this happen because the static times for pickups and dropoffs are estimated to short? They do take longer then the pickups in the demos. If that is the case, how could I modify the cost calculations?

3

Posted by @Odin-byte:

The issue seemed to be, that the speed and acceleration values within the fleet config were to high