handling task failures in a fleet adapter for robots using an API

[osama-z-salah]

I am currently facing an issue where my robot sometimes fails during a specific phase of a task, such as docking at the charging station. My implementation utilizes a fleet adapter that is almost identical to the one provided in the demo. However, I am experiencing difficulties with the is_command_completed function provided by my API. In the demo, the function behaves consistently—returning false when the command is still ongoing and true when it is finished. In my case, the API provides a function like:

command_status = self.api.is_command_completed()

if command_status == "completed":
    self.execution.finished()
    self.execution = None
elif command_status == "abort":
   // no idea 
else:
    activity_identifier = self.execution.identifier

I need the fleet adapter to restart the failed phase whenever an abort occurs. However, I am not entirely sure what exactly should go in the "abort" condition to ensure that the fleet adapter restarts the failed phase reliably or even report that the full task has failed and return the robot into the parking if that is normal go to place task.
Any suggestions on how to make this more robust or ways to handle the abort scenario so that the phase can be retried effectively would be greatly appreciated.