Kill task vs Cancel Task

[selvavignesh-2206]

I am wondering what is the difference between the cancel and kill task functions in rmf web?

When I test them out, they both cancel the task and send the robot to the same way point. However, the task status are different (cancelled vs killed).

Is there any reason to have two different call methods to do similar function?

[aaronchongth]

Hello @selvavignesh-2206! Please refrain from posting screenshots of terminal outputs, use gist.github.com instead

Could you please elaborate what you mean by testing, and how you are killing or cancelling the tasks?

We do not have any functions to kill a task via the web dashboard at the moment, only cancellation is supported, so I believe it was killed on the fleet adapter level. My guess is that the fleet adapter was implemented to kill the task instead of cancelling it in your scenario

The general expectations is that a task is only killed when fleet adapters decide that the task cannot proceed, and hence need to be killed, for example a box blocking the corridor and the robot is unable to proceed. While cancelling a task would actively involve a user/operator who decides to cancel it.

[aaronchongth]

Fleet adapters can be implemented to choose between kill or cancel too, https://github.com/open-rmf/rmf_ros2/blob/main/rmf_fleet_adapter/include/rmf_fleet_adapter/agv/RobotUpdateHandle.hpp#L226-L260

[selvavignesh-2206]

The general expectations is that a task is only killed when fleet adapters decide that the task cannot proceed, and hence need to be killed, for example a box blocking the corridor and the robot is unable to proceed. While cancelling a task would actively involve a user/operator who decides to cancel it.

I think I get it from this explanation. I was killing and cancelling the task through the rmf web api. My fleet adapter was implemented with cancel function instead of kill.

Thanks for the help.