teleoperation (interaction mode)

Seen as a way of interacting with a robot, teleoperation is the mode where a human steers the machine in real time and the robot acts as the person's distant hands and eyes. The human watches a live feed from the robot's cameras and sensors and sends commands — moving a joystick, swinging their own arm, squeezing a grip — while the robot copies them somewhere else. Among the ways people and robots can work together, this one keeps the human firmly in the driver's seat moment by moment, rather than handing the robot a goal and stepping back.

What makes this mode tricky is the loop the human is closing: see, decide, command, watch the result, correct. Two things can break that loop. The first is latency — the lag between sending a command and seeing what it did. Even a third of a second turns smooth control into a frustrating overcorrect-and-wait dance, and over long distances, such as to a robot on Mars, the lag is far worse. The second is feedback: a remote operator cannot feel how hard the robot is gripping or pushing unless the system deliberately sends that force back to them, often through a controller that resists the hand. Good force feedback can be the difference between cracking an egg and crushing it.

Because pure moment-to-moment control is so demanding, real systems usually blend it with autonomy in what is called shared control. The human supplies intent — 'go to that valve,' 'grasp that handle' — and the robot fills in the fast, fiddly details, stabilizing itself and avoiding obvious mistakes between human commands. This division of labor lets one operator manage a robot, or even several, far more comfortably than raw teleoperation would allow, and it gracefully covers for delay by letting the robot keep itself safe while it waits for the next instruction.