A robot task completion check observes the environment after an action and compares it with the intended state. It might confirm that an object is inside the correct bin, a container is closed or a handoff was received.
The check prevents a motion command from being mistaken for success merely because it finished running. Reliable verification accounts for sensor uncertainty and can trigger a retry, replanning or human assistance when evidence is inconclusive.
ELI5
A robot task completion check confirms that a physical action achieved the requested result. Finishing a movement command is not enough because the object may have slipped, landed in the wrong place, or never been grasped.
For example, after putting a bottle in a bin, the robot can use a camera or weight sensor to verify that the bottle is really inside. If the evidence is unclear, it can retry, make a new plan, or ask a person for help.
How can a robot verify that a task is complete?
It can use cameras, force sensors, object tracking, device state or another robot's confirmation to compare the observed result with the required condition.
Why is action completion not the same as task completion?
A robot can finish a commanded movement without grasping, moving or placing the intended object correctly, so the result itself must be checked.
