Robot Vacuums Are Learning to Feel the Furniture

There is a small, unglamorous test of a home robot’s intelligence: what happens when it meets your favourite chair.

A machine can look impressive in a demonstration. Living with it is different. You notice the knocks, the awkward approaches and the moments when you have to step in.

At IFA 2026 in Berlin, held on 4–8 September, MOVA promoted a technology called TactiPercept. The company describes it as adding contact sensing to robotic perception, with the aim of making a robot’s response gentler. It is a manufacturer announcement, with the real-world benefit still requiring independent testing. MOVA’s announcement · MOVA’s IFA programme

The appeal is easy to understand. A robot that can respond carefully to contact could be more pleasant to share a room with.

Seeing the chair and feeling it are different jobs

Imagine reaching for a drinking glass. Looking helps you place your hand. Touch helps you judge what happens once your fingers arrive.

Robotics researchers have been investigating that second source of information for years. In 2017, MIT described experiments using GelSight sensors to judge the hardness of objects and help robotic grippers manipulate small tools. A soft surface deformed on contact, and an optical system measured the deformation. MIT’s research explanation

That research is background, not evidence that MOVA uses the same mechanism. Its relevance is the broader principle: physical contact can reveal information that a view of the room does not provide.

Touch sensing also has nothing to do with a machine having feelings. It means measuring a physical interaction and using the result to guide a response.

Robot vacuum touch sensors need useful feedback

Contact detection itself is already familiar. Existing robot vacuums use bumper sensors to respond to obstacles; iRobot describes how a Roomba changes direction after a bump. The worthwhile advance would be more informative feedback and better control of contact, rather than discovering for the first time that something is in the way. iRobot’s explanation of bump sensors

MOVA’s pitch is that sensing contact can help a robot react more gently. The important word is react. A sensor becomes useful when the control system turns its signal into better movement. MOVA’s description of TactiPercept

Consider a possible edge-cleaning routine. A robot approaches a wall, detects contact and changes its motion. Done well, that feedback could help it stay close while limiting pressure. Done badly, it could stop too often, push too hard or repeatedly misinterpret what is happening.

That example illustrates the engineering goal; it is not a claim that an independently tested MOVA vacuum has already mastered it.

The difference would come from the whole system: sensing, interpretation and movement working together.

Conceptual feedback loop. It describes the goal, not a measured product-performance result.

A living room is a harder test than a stage

The most useful comparison would use ordinary household obstacles: a light chair, a heavy table, a soft curtain and a low piece of furniture. It would measure contact force, count occasions when the robot needs help and check the dirt left behind.

A gentler robot still needs to clean. One that avoids every difficult edge might preserve the furniture beautifully while leaving its owner to finish the job.

Longer testing would matter as well. Any sensing surface exposed to normal household use would need to keep delivering dependable information after repeated contact and cleaning. A repairable part would be more reassuring than a delicate feature built into an expensive assembly.

These are review criteria to look for when a suitable product becomes available. The announcement alone does not answer them.

The missing details matter

The primary material reviewed for this article does not establish which Australian retail model will include TactiPercept, what it will cost or when buyers will receive it. Nor does the short company announcement provide comparative collision-force measurements or independent cleaning results.

For now, this is best treated as a development to follow. The concept is credible enough to be interesting; the purchasing case needs much more information.

Home robots will earn trust through small actions

The fascinating part of tactile sensing is how ordinary its benefits could feel. Less furniture shuffling. Fewer interruptions. A machine that handles an awkward corner without making it your problem.

Those improvements would rarely make the most dramatic demonstration video. They could matter much more after the novelty wears off.

MOVA’s announcement points towards that kind of progress. The next meaningful step will be a product that shows, repeatedly and in normal homes, that a sense of touch makes it a better house guest.