A research team has introduced a robotic fingertip that can perceive color, marking a notable step forward in the effort to give machines a fuller sense of touch. The device, called FlexiRay, is described in a paper published in *Nature Communications* in 2026. The study, titled "Flexible robotic hand harnesses large deformations for full-coverage human-like multimodal haptic perception," appears in volume 17 as article number 458.
The core innovation is not just that the finger can see color—though that alone is significant—but that it does so while maintaining a high degree of physical flexibility. According to the paper's abstract and accompanying data, FlexiRay's sensing area measures 1,560 square millimeters. That figure is 5.9 times larger than the sensing area of GelSight Mini, a well-known tactile sensor used in robotics research, which comes in at 265.98 square millimeters.
That larger surface area is not merely a matter of size. It allows FlexiRay to capture what the researchers describe as "non-planar features" at greater distances than GelSight Mini can manage. In practical terms, this means the finger can detect shapes and textures that are not flat, and it can do so even when the object is not pressed directly against the sensor. The compliance of the material—its ability to deform and conform to what it touches—enables single-contact conformation to curved surfaces. In other words, one touch is enough for the finger to wrap around a curved object and gather data from that interaction.
The paper also reports a specific capability that sets FlexiRay apart: it can detect dual protrusions simultaneously. In the experiments described, the researchers used magenta, purple, and cyan balls as test objects. FlexiRay achieved detection rates of 66.7 percent for magenta, 66.7 percent for purple, and 73.3 percent for cyan. More importantly, the finger could identify two protrusions on the magenta ball at the same time—something that is impractical with GelSight Mini. This leads to what the authors call "richer touch cues," meaning the sensor provides more detailed information per contact event than existing alternatives.
The publication metrics are also worth noting. As of the time the article was indexed, it had been accessed 14,000 times, cited 13 times, and attracted 21 Altmetric mentions. These numbers suggest the work is drawing attention from the research community, though the article itself does not specify who funded the work or whether any commercial partners were involved.
What the source material does not disclose is equally important. The paper does not state the exact materials used to construct FlexiRay, nor does it provide details on manufacturing processes, cost, or durability under repeated use. It also does not specify the robot platform on which the finger was tested, beyond the general description of a "flexible robotic hand." The article does not mention any plans for commercialization, licensing, or open-source release of the design. Those details remain outside the scope of what has been published so far.
Why it matters for European robot service
For the European robotics industry, particularly the service robotics sector, the development of FlexiRay touches on a long-standing bottleneck: tactile perception. Most robots deployed in warehouses, hospitals, and homes rely heavily on vision. Cameras and LiDAR systems tell a robot where objects are, but they do not tell it how those objects feel. A robot that can only see a cup does not know whether the cup is made of glass, ceramic, or paper. It does not know if the surface is slippery, rough, or warm. It cannot tell if it is about to crush a fragile item or drop a heavy one.
Tactile sensors have existed for decades, but they have historically been limited in two ways. First, they tend to cover small areas, which means a robot must make many contacts to build a picture of an object. Second, they are often rigid, which limits their ability to conform to curved or irregular shapes. FlexiRay addresses both of these limitations in a single device. Its sensing area is nearly six times larger than a common commercial alternative, and its compliance allows it to deform and wrap around objects.
This matters for European service robots because so many of the tasks these machines are asked to perform involve manipulation of objects that are not flat. Consider a robot in a hospital that must hand a patient a glass of water. The glass is cylindrical, possibly curved, and may have a handle. A rigid sensor would only make contact at a few points. FlexiRay, by contrast, can conform to the curve of the glass and gather data from a much larger contact area. That means the robot can better judge how much force to apply, whether the glass is slipping, and whether the grip is secure.
Similarly, in logistics and warehousing—a sector where Europe has been rapidly deploying robots—the ability to handle varied items is critical. Parcels come in all shapes and sizes. Some are rigid boxes; others are soft pouches or irregularly shaped goods. A robot gripper equipped with FlexiRay could, in principle, detect the shape and texture of an item before deciding how to grasp it. The dual-protrusion detection capability is particularly relevant here. Many packaged goods have multiple features—handles, caps, spouts, or seals—that a robot must account for when picking them up. The ability to detect two such features simultaneously, without needing multiple contacts, could reduce cycle times and improve reliability.
The European service robotics market has been growing steadily, with applications in healthcare, agriculture, logistics, and domestic assistance. The European Commission has funded numerous projects aimed at improving robot perception and manipulation. FlexiRay, if it can be integrated into commercial systems, would align with those goals. However, it is important to note that the paper does not describe any integration with existing European robot platforms. The research appears to be at the prototype stage, and the path from laboratory to product is not detailed in the source material.
There is also a broader strategic consideration. Europe has been seeking to reduce its dependence on non-European robotics components, particularly in areas like sensors and actuators. A new tactile sensor developed by researchers—wherever they are based—could become part of the European supply chain if it is licensed or manufactured locally. But the paper does not state the affiliation of the researchers beyond their names: Wang, Y., Guo, H., Wu, H., and colleagues. The source material does not specify whether the work was conducted at a European institution, an Asian institution, or a multinational collaboration. That information is not disclosed in the available text.
For European robot service providers, the practical implication is this: the technology is promising, but it is not yet a product. The paper demonstrates a proof of concept with specific performance metrics, but it does not address questions of cost, reliability over thousands of cycles, or ease of integration with existing robot control systems. Those are the questions that will determine whether FlexiRay becomes a component in European robots or remains a research curiosity.
What buyers and operators should know
For those in the market for tactile sensing solutions, the FlexiRay paper offers several data points worth considering, but it also leaves many practical questions unanswered. It is essential to distinguish between what the research demonstrates and what it does not.
First, the demonstrated capabilities. FlexiRay has a sensing area of 1,560 square millimeters, which is 5.9 times larger than the GelSight Mini's 265.98 square millimeters. This is a concrete, verifiable number. It means that, per contact event, FlexiRay can gather significantly more tactile data than a common benchmark sensor. For applications where a robot must identify an object by touch alone, this larger area could translate into faster recognition and fewer required contacts.
Second, the compliance of the sensor. The paper states that FlexiRay's compliance enables single-contact conformation to curved surfaces. This is a qualitative claim, but it is backed by the experimental results showing detection of non-planar features at greater distances. For operators dealing with curved or irregularly shaped objects—bottles, tools, produce, medical devices—this capability could be valuable. A single touch that conforms to the object's shape provides more information than a rigid sensor that only contacts at a few points.
Third, the dual-protrusion detection. The experiments with magenta, purple, and cyan balls show detection rates of 66.7 percent, 66.7 percent, and 73.3 percent, respectively. These numbers are not perfect—they indicate that the sensor does not always succeed—but they demonstrate a capability that GelSight Mini lacks. For tasks where objects have multiple features that must be detected simultaneously, this could be a differentiator.
Now, what buyers and operators should not assume. The paper does not provide any information on the following:
- **Durability**: How many contact cycles can FlexiRay withstand before performance degrades? The paper does not say.
- **Environmental tolerance**: Can the sensor operate in extreme temperatures, high humidity, or dusty environments? Not disclosed.
- **Integration complexity**: Does FlexiRay require custom electronics, or does it interface with standard robot controllers? Not specified.
- **Cost**: There is no pricing information in the source material.
- **Availability**: The paper does not indicate whether FlexiRay is available for purchase, licensing, or evaluation.
- **Lead times**: No information on manufacturing or delivery timelines is provided.
- **Support and service**: The paper does not mention any warranty, technical support, or maintenance arrangements.
These are not minor omissions. For a robotics integrator or an end-user operator, the difference between a promising sensor and a deployable product lies precisely in these details. A sensor that works beautifully in a laboratory but fails after a few hundred contacts in a dusty warehouse is not useful. A sensor that requires custom electronics and bespoke software might cost more to integrate than the sensor itself. A sensor with no clear supply chain or support structure is a risk, not an asset.
The publication metrics—14,000 accesses, 13 citations, 21 Altmetric mentions—suggest that the research community is paying attention. But attention from researchers does not equal readiness for deployment. The paper is a research publication, not a product announcement. It does not state whether the authors are affiliated with any company, whether a patent has been filed, or whether a commercial version is in development.
Operators who are considering tactile sensors for their robot fleets should therefore treat FlexiRay as a signal of what is coming, not as a solution that is available today. The underlying technology—large-area, compliant, color-sensing tactile perception—is clearly advancing. The gap between GelSight Mini and FlexiRay in sensing area is substantial, and the dual-protrusion capability is a genuine advance. But the path from this paper to a commercial product involves engineering work that the paper does not describe.
It is also worth noting that the paper does not compare FlexiRay to any sensors other than GelSight Mini. There are many tactile sensors on the market and in research, and the source material provides no data on how FlexiRay compares to those. The 5.9× figure is relative to one specific benchmark. Without comparative data against other sensors, it is difficult to assess where FlexiRay stands in the broader landscape.
Finally, the color-sensing capability itself deserves scrutiny. The paper states that FlexiRay senses color, and the experimental results involve colored balls. But the source material does not explain how color information is used in manipulation. Does it help the robot identify the object? Does it aid in grasping? Does it improve force control? The paper does not say. For a buyer, the practical value of color sensing is unclear from the available information.
In summary, the FlexiRay paper is a legitimate and interesting contribution to tactile sensing research. The numbers are specific, the capabilities are demonstrated, and the comparison to GelSight Mini is concrete. But the paper is not a buying guide. It does not address the operational, commercial, or logistical questions that a buyer or operator would need answered before making a procurement decision. Those questions remain open, and the source material does not provide answers.
As with any emerging technology, the prudent approach is to monitor the development, engage with the researchers if possible, and wait for further publications or product announcements that address the gaps identified above. The technology is promising; the product is not yet here.
Sources
https://spectrum.ieee.org/robot-finger
Published by Vigla Media OÜ (Estonia).