Robots are getting better at many tasks, but touch remains a difficult part of human-robot interaction. A system led by the German Aerospace Center and published in Science Robotics points to a different route: instead of wrapping a machine in artificial skin, it reads contact through forces sensed inside the robot itself.
The result is a robotic arm that can detect where it is being touched, analyze how pressure is applied and interpret simple signals drawn on its surface. The goal is not just to give robots a new sense, but to make communication with them less dependent on tablets, keyboards or specialist interfaces.
Why robot touch is hard
Human touch works through more than one pathway. When a person presses lightly on a fingertip, the skin can register the contact through the many receptors in the hand and fingers. That is the model roboticists often try to imitate with artificial skins: cover the robot in sensors so contact can be detected directly at the surface.
That approach has clear appeal, but the source article describes two major drawbacks. Artificial skins can be expensive, and they may struggle with impacts or harsh environments. For robots that need to work around people, especially larger machines, a surface full of delicate sensing hardware can become a practical limitation.
The new system leans on a second way people sense touch. If pressure is stronger, the body can feel the effect through joints, not just skin. In robotics language, that means sensing torque. The researchers recreated that idea in a robotic arm, using the robot’s internal response to contact as the basis for touch detection.
How the system reads contact
The robotic arm contains six sensors. Each one can register very small amounts of pressure applied anywhere along the device. Once the system measures the force and the angle of that force, algorithms map where the touch happened and interpret what the person appears to be communicating.
That turns the surface of the robot into an input area. A person could draw letters or numbers on any part of the robotic arm with a finger, and the robot could interpret those movements as directions. Any part of the robot could also function as a virtual button.
Maged Iskandar, researcher at the German Aerospace Center and lead author of the study, frames the value in terms of interaction. As he puts it,
“Human-robot interaction, where a human can closely interact with and command a robot, is still not optimal, because the human needs an input device.”
He adds that if the robot itself can serve as the device, interactions can become more fluid. In practical terms, the interface is no longer a separate screen or controller. It is the machine’s own body.
Why this could matter for larger robots
The article describes the system as a potentially cheaper and simpler way to provide both touch sensing and communication. That matters because larger robots, including humanoids, continue to attract billions in venture capital investment. If those robots are expected to operate near people, they need ways to receive quick, natural input from humans around them.
A touch-based interface could support that by making commands more immediate. Instead of requiring a person to use a tablet or understand a technical control system, the robot could respond to contact on its own surface. The key idea is straightforward: every square inch of the robot can behave like a touch screen, without needing the wiring, fragility and cost of a conventional touch surface.
Calogero Maria Oddo, a roboticist who leads the Neuro-Robotic Touch Laboratory at the BioRobotics Institute and was not involved in the work, calls the development significant. He points to the combination of sensors, mathematical mapping and new AI methods as the reason the system stands out.
Oddo also suggests that commercial adoption could be fairly quick because the added investment is more about software than hardware. That distinction matters: software changes are generally less expensive than adding complex physical systems across a robot’s body. The sensors required are commercially available, though the article notes they can cost tens of thousands of dollars.
The limits are still important
The system is not a complete answer to robot touch. One major limitation is that the new model cannot handle more than two points of contact at once. In controlled environments such as a factory floor, that may be manageable. In settings where human-robot interaction is less predictable, it could restrict what the robot can reliably understand.
There is also the cost of the sensing hardware itself. Even without artificial skin, the required sensors are not necessarily cheap. The article describes them as commercially available, but also notes that they can cost tens of thousands of dollars.
Those caveats do not erase the broader implication. The research shows that robot touch does not have to come only from a skin-like covering. A robot can also infer contact by understanding how force moves through its own structure.
What comes next
Oddo expects the future of robot touch to combine both approaches. Humans and other animals use skin-based sensing and joint-based sensing together, and he expects robots working in the real world to do the same.
That points to a more complete model of human-robot interaction. Artificial skin may still be useful for direct surface sensing, while joint-based systems could add durability, coverage and a built-in communication layer. Together, they could help robots interact more safely and smoothly with people and their surroundings.
For now, the key advance is conceptual as much as technical. The robot does not need to be covered in artificial skin to understand touch. With six sensors, careful force measurement and algorithms that interpret contact, the machine’s body can become both a sensing surface and a control interface.