Teleoperation
Also known as: remote operation, telerobotics, remote control operation
Teleoperation is controlling a machine remotely, with a person watching live video and sensor data and sending commands over a network or radio link.
It's used for space robotics, bomb disposal, underwater vehicles, mining equipment, and as backup support for robotaxis and delivery robots. While teleoperation extends human skill to dangerous or distant places, it depends on fast, reliable connections, and delay can make precise control difficult.
Teleoperation, explained at three levels
At home
Think of a video game where your controller moves a character on screen, except the character is a real machine. In teleoperation, cameras and sensors on a robot, vehicle, or arm stream what it sees to a person far away, who sends commands back over a radio or network link. The machine carries them out, letting the operator work somewhere dangerous or distant without being there.
At work
Teleoperation is remote control of machines by a person using live video and sensor data. It lets organizations run equipment in dangerous or distant places, such as mines, disaster sites, nuclear facilities, the deep ocean, and space, from a safe control room. It also backs up autonomous systems: robotaxi and delivery-robot companies use remote staff to help when machines get stuck.
At home
A teleoperation setup has three parts: the machine with its cameras and sensors, a communication link, and an operator station with screens and controls such as joysticks. Some stations add haptic feedback, motors in the controls that push back so you feel how hard a gripper is squeezing. The biggest enemy is delay: if video arrives late, your commands are based on the past, so precise moves become slow and careful.
At work
Teleoperation sits on a spectrum. In direct control, the operator moves every joint or wheel. In supervised autonomy, the machine handles routine work and the operator approves or corrects its decisions; ground controllers often run Canadarm2 this way with pre-planned moves. Remote assistance, used by robotaxis, only offers advice such as a path to follow. The engineering constraints are latency, bandwidth for video, and link reliability, which is why mines often build private LTE or Wi-Fi networks.
At home
Radio signals take about 1.3 seconds to reach the Moon and roughly 3 to 22 minutes to reach Mars, depending on where the planets are. Driving a Mars rover by joystick is impossible, so controllers send a day's plan and the rover carries it out, avoiding hazards on its own. While teleoperation keeps humans in charge, in my view distance makes autonomy unavoidable, and the same logic applies to robots on Earth with poor connections.
At work
Key design questions are what the machine does when the link drops, usually a safe stop, how many machines each operator can supervise, and how fatigue affects judgment. Control links must be encrypted and authenticated, since a hijacked machine is a real hazard. Remote driving on public roads remains a regulatory grey area in most of Canada. Teleoperation logs are also valuable training data for robot AI models, which several humanoid companies now rely on.
Real-world examples
At home
Carlos watched a documentary showing a ground controller in Quebec operating Canadarm2 on the space station, moving cargo while the astronauts slept.
At work
Fatima is an operations manager at a Sudbury mine. Her team uses teleoperated loaders from a surface control room, keeping workers away from unstable underground areas after blasting.
How it compares
| feature | Teleoperation | Full autonomy |
|---|---|---|
| Who decides | Remote human | The machine |
| Network dependence | High | Low |
| Handling unusual situations | Strong, human judgment | Limited to training |
| Labour cost | Ongoing operators | Low per machine |
Teleoperation brings human judgment but needs a person and a good link.
For analysts
Teleoperation can remove workers from danger quickly. I'd test the network in real conditions and define what the machine does when the connection drops before deploying.
History & origin
Raymond Goertz built mechanical master-slave manipulators in the late 1940s at Argonne National Laboratory so workers could handle radioactive materials safely. NASA and the Soviet program used remote control for lunar rovers, including the Soviet Lunokhod 1 in 1970, driven from Earth. Canada's Canadarm flew on the Space Shuttle in 1981, operated by astronauts, and Canadarm2, installed on the International Space Station in 2001, is now often operated by ground controllers, including at the Canadian Space Agency in Saint-Hubert, Quebec. Military drones and underwater vehicles expanded teleoperation further. More recently, startups such as Vay began remote-driven car services, and robotaxi firms use remote assistance staff. In my view, teleoperation is the quiet backbone behind many 'autonomous' systems, and companies should be more open about how much they rely on it.