Anzubo
Your country

Your choices

VLAs and teleoperation, explained

The short answer: a VLA is a kind of software that lets a robot turn what it sees and what it is told into movements. Teleoperation means a human controlling the robot remotely. Home robot makers use both, often together, and a buyer should know which is doing the work in their home, because it changes what you are buying: a robot that acts on its own, or a robot that is sometimes driven by a person working for the vendor.

What a VLA is

VLA stands for vision-language-action model. It is a type of AI model that takes in images from the robot’s cameras (vision) and an instruction in words (language), and outputs the robot’s movements (action). Google DeepMind used the term in 2023 for its RT-2 model, which learned from both web data and robot data and turned that knowledge into instructions for controlling a robot.1

The appeal is generality. A robot running a VLA is not following a script for each chore; in principle it can be asked to do something new in plain words and work out how. In practice, what any particular VLA can do depends on what it was trained on, and few of the vendors on this site have published that.

Several robots on this site are described by their makers with VLA or similar terms: SwitchBot names its system “Omni Sense VLA” for onero H1, Flourish Robots says Flourish 1 uses a VLA to grasp objects, and Futuring Robot describes an “AVLA” design for Futuring 2. A vendor calling something a VLA tells you the general approach, not how well it works.

What teleoperation is

Teleoperation is a person controlling a robot from somewhere else, usually through cameras and a control rig. It shows up in home robotics in three different ways, and the difference matters.

  1. To train the robot. Researchers and companies record people guiding a robot through a task by remote control, then train the robot’s software to imitate those recordings.2 1X describes its model for NEO as trained this way. This happens before you own the robot.
  2. To help the robot in your home. A person at the vendor controls the robot live, in your home. NEO and Isaac 1 both offer this: 1X lets owners schedule a remote “1X Expert” to guide NEO through chores it does not yet know, and Weave says Isaac 1 is autonomous by default, “with teleoperation assistance when needed to guarantee we complete tasks”. UniX AI also advertises a remote-operation system for Wanda, but no source says whether any household task is done that way.
  3. By you, to teach it. Whoever buys a Flourish 1 teaches it each chore by driving the robot from a phone; the robot then repeats the chore by itself. Here the remote control happens once, in your hands, rather than at the vendor.

Why the distinction matters to a buyer

If a robot relies on live teleoperation in your home, then some of the work you see is being done by a person, not by the robot. That is not necessarily bad (it can be how a new robot gets through jobs it cannot yet manage alone), but it changes the product. It raises practical questions about when an operator can connect, what they can see, and what happens when no operator is available.

The same question applies to demonstrations. In a press hands-on test at NEO’s October 2025 launch, the chores shown were carried out by a remote operator. Where the sources say how a demonstration was done, each robot’s page records it next to the claim.

This site labels robots that use live remote help, and does not count remote operation as the robot acting on its own. A robot that is autonomous for some tasks and hands others to a human operator is listed, with the label. A robot that is only ever remote-controlled is not.

Questions worth asking a vendor

  • Which of the tasks you advertise does the robot do without any human help?
  • When does a remote operator take over, and who decides?
  • Can the operator see into my home, and when?
  • Were your demonstration videos autonomous, remote-controlled, or a mix?

For the other questions worth asking before you order, see What to ask before buying a home robot.

Notes

  1. Google DeepMind, “RT-2: New model translates vision and language into action”, 2023-07-28. https://deepmind.google/discover/blog/rt-2-new-model-translates-vision-and-language-into-action/ ↩

  2. Zhao, Kumar, Levine and Finn, “Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware”, arXiv, 2023-04-23. https://arxiv.org/abs/2304.13705 ↩

Last updated 12 November 2026.