Do service robots use artificial intelligence?

Parts of the stack often do. Perception — interpreting camera or lidar data to identify obstacles and people — is a common place for learned models. Route planning, task scheduling and safety stops are typically deterministic algorithms, because predictability matters more than cleverness there.

Does Hybot use machine learning?

Hybot's task assignment is a deterministic weighting of priority, distance and battery, re-evaluated every two seconds, not a learned policy. Where learned perception exists it belongs to the robot hardware vendor rather than to our platform. We will not claim capabilities that sit in somebody else's product.

Why be so careful about the word AI?

Because a buyer comparing two systems needs to know which behaviours are predictable and which are probabilistic. Labelling a scheduling rule as AI does not make it better; it makes the buyer unable to ask the right question about how it will behave under load.

What should a buyer actually ask?

Ask which component makes each decision, whether it is deterministic, and what it does when it is uncertain. A system that explains its own decision boundaries is more trustworthy than one that describes everything with the same three-letter word.

Cluster guide

AI in robotics: what is real and what is marketing

Artificial intelligence appears in service robotics mostly through perception and navigation rather than through anything resembling reasoning. Learned models help a robot interpret sensor data and predict how people move. Scheduling, safety limits and fleet coordination are usually deterministic engineering, and vendors who blur that line make evaluation harder for buyers.

Written by Hybot technical lead, Technical lead, Hyrcan-Tech · 2 February 2026 · 8 min read

Why this cluster exists

Because "AI-powered" is now applied to almost every robot on the market, and a buyer cannot evaluate a claim that vague. These articles describe the techniques at industry level — what they are, where they genuinely help in service robotics, and what they cost — so that you can ask a supplier a question they have to answer specifically.

Where Hybot stands

We will state this plainly here and repeat it in every article in the cluster, because it is the part that is usually left ambiguous.

Task assignment in Hybot is deterministic. Priority, distance to pickup and remaining battery are weighed together and the queue is re-scored every two seconds. It is not a learned policy, and we would not want it to be: an operator needs to be able to predict which robot takes a job.
Perception belongs to the robot hardware. Where learned models interpret sensor data, that capability is part of the robot vendor's product. Hybot integrates through a vendor-agnostic abstraction layer; we do not claim somebody else's perception stack as our own.
We do not ship a language model. There is no conversational assistant in the product, and the articles here about conversational AI are educational rather than descriptions of a Hybot feature.
We do not do predictive maintenance. The platform records battery, state and health by heartbeat. Recording telemetry is not the same as predicting a failure, and we will not describe it as if it were.

What is in this cluster

Machine learning in robot navigation — where learning genuinely improves route and motion decisions, and where classical planning still wins.
Computer vision in robotics — what cameras and depth sensors are used for, and the privacy questions a venue should ask before any camera is installed.
Conversational AI and robots — the case for and against a talking robot in a noisy venue.
Edge AI in robotics — why inference on the robot matters when the building's connection is unreliable.

The question that cuts through

For each behaviour, ask: is this decision deterministic, and what does it do when it is uncertain?

A system that can answer that per component is one you can operate. A system described entirely as "AI-powered" is one where nobody has separated the parts that must be predictable from the parts that are allowed to guess.

Frequently asked questions

Do service robots use artificial intelligence?: Parts of the stack often do. Perception — interpreting camera or lidar data to identify obstacles and people — is a common place for learned models. Route planning, task scheduling and safety stops are typically deterministic algorithms, because predictability matters more than cleverness there.
Does Hybot use machine learning?: Hybot's task assignment is a deterministic weighting of priority, distance and battery, re-evaluated every two seconds, not a learned policy. Where learned perception exists it belongs to the robot hardware vendor rather than to our platform. We will not claim capabilities that sit in somebody else's product.
Why be so careful about the word AI?: Because a buyer comparing two systems needs to know which behaviours are predictable and which are probabilistic. Labelling a scheduling rule as AI does not make it better; it makes the buyer unable to ask the right question about how it will behave under load.
What should a buyer actually ask?: Ask which component makes each decision, whether it is deterministic, and what it does when it is uncertain. A system that explains its own decision boundaries is more trustworthy than one that describes everything with the same three-letter word.

Everything in this cluster

Each of these goes deeper on one part of the topic above.

Computer vision in robotics
What cameras and depth sensors are actually used for on a service robot, what they are bad at, and the privacy questions a venue should ask before installation.
Conversational AI and service robots
The case for and against a talking robot in a noisy venue — what language models are good at, what fails on a restaurant floor, and what Hybot ships instead.
Edge AI in robotics
Why inference on the robot matters when a building's connection is unreliable, what belongs in the cloud instead, and how Hybot's architecture splits the two.
Machine learning in robot navigation
Where learned models genuinely improve how a robot moves through a crowded venue, where classical planning still wins, and what Hybot actually uses.

Take it further

If a question here applies to a venue you actually run, the specifics matter more than the general case.

See how task assignment works