The coming era of robotics is not simply about increasing capability, it is about fostering a new kind of integration between machines and their environments, as well as, perhaps more importantly, society as a whole. Robots are beginning to move beyond fenced-off production lines into homes, hospitals, warehouses, and fields — spaces that are inherently more dynamic, unstructured, and often shared with people.

This transition demands not only advances in artificial intelligence (AI) but also a reconsideration of the way robots perceive the world. In these environments, physical interaction is both unavoidable and often unpredictable, and this demands a true imitation of human senses, with touch being critical. Robots must not only decide what to do but understand how they are doing it: whether a grip is secure, whether contact has occurred, whether resistance has changed.

Realizing this vision requires more than isolated technological improvements. The convergence of feedback mechanisms, like tactile sensors, and smart and adaptive control systems, is an interdisciplinary challenge requiring extensive collaboration among electronic hardware, software, embedded systems, mechanical robotics, and mechatronics experts. This article explores how the work of Melexis and Brubotics, the Brussels Human Robotics Research Center of the Vrije Universiteit Brussel (VUB), offers one such example of how collaboration can help to unlock new frontiers in robotic interaction.

The Drive for Intelligent Robotic Perception

Modern robotic systems have achieved impressive precision and adaptability, particularly when equipped with high-accuracy vision systems and operating in repetitive applications. However, tactile perception, the ability to sense and interpret physical contact, remains comparatively limited. Many robots today are envisioned to utilize visual feedback and advanced force sensors, yet the practical integration of such sensors is often limited due to factors such as the prohibitive cost and the challenge of achieving the necessary precision.

While effective in highly structured environments, this basic approach can fall short when robots are required to manipulate fragile or irregular objects, adjust grip dynamically, or interact safely with humans. Unlike the human sense of touch, which is a rich combination of pressure, direction, friction, and vibration, most robotic platforms operate with a comparatively narrow understanding of contact.

For example, a lack of reliable shear force detection can limit a system’s ability to detect when grip is slipping. Without it, robots must rely on predefined grip strategies or overcompensate with force, increasing the risk of either dropping objects or damaging them. This also limits their ability to learn from interaction and adapt to changing conditions in real time, as the control systems’ modeling and understanding of gripping tasks is essentially limited.

To compensate for the lack of multimodal touch feedback inherent in humans, robots require advanced tactile hardware that can deliver comprehensive 3D information, enabling more accurate perception and interaction with their environment. This parallelism of mirroring human operation also carries into the software side. Complex embedded software, such as edge AI, is crucial for robotic systems to interpret tactile data, especially in understanding and adapting to the complex interactions between robots and objects during gripping tasks that are not predefined. But AI development of complex physical interactions can be inherently time-consuming, especially if carried out in the real world.

These challenges reflect a broader issue in robotics: the historical separation between sensing hardware and AI development. To advance meaningful robotic touch, hardware and software must not only improve but also evolve together, informed by shared use cases, common datasets, and coordinated design. It is this gap between sensor design and the practical applications in robotics that has driven the collaboration between Melexis and Brubotics.

A Collaboration in Action: Melexis and Brubotics

With the support of the Flemish Agency for Innovation and Entrepreneurship (VLAIO) project SKINAXIS, Brubotics, a consortium of eight Vrije Universiteit Brussels research groups, is undertaking the task of equipping robots with a real sense of touch using Melexis technology. The initiative is designed not only to address the technical problem of measuring contact forces accurately but also the equally complex challenge of using that data to enhance real-world robot performance.

Central to the project’s hardware is Melexis’ latest innovation, designed from the ground up for the robotics market – Tactaxis®. Unlike conventional tactile sensors that typically detect only vertical (normal) pressure, the novel 3D magnetic tactile sensor can detect both normal and lateral (shear) forces with a high degree of accuracy and resolution. This is achieved through an elegant mechanical–magnetic design – a magnet embedded in a soft elastomer structure is displaced by external forces, and this displacement is measured by a Triaxis® 3D magnetic sensing IC below. With this configuration, every compact 6 x 6 x 4.4 mm³ tactile sensor transmits real-time vector data on contact force at up to 1000 samples per second, sensitive to 30 mN, with a normal force range of 5 N, and an overload resistance of 15 N, ensuring the smallest deviations are captured.

Within the project, Brubotics’ role is to transform raw sensory data into practical robotic applications, showcasing innovative solutions for enhanced safety and precision in robotic interactions. This means developing AI algorithms, robotic control strategies and experimental environments required to allow robots to interpret sensor feedback, predict slippage, and regulate gripper force to maintain a secure grip of numerous objects without deformation.

Rather than collecting training data from scratch — a slow and laborious process — Brubotics implemented a simulation-based training approach using NVIDIA’s Isaac Sim platform.

By integrating physics-based models of the Tactaxis® sensor into virtual robotic systems, the Brubotics teams can expose their AI models to thousands of manipulation tasks in simulated environments. These digital twins included common variables such as object properties, surface friction, and dynamic interactions, allowing the models to learn rich contact behavior without the bottleneck of real-world data collection.

The AI models are being continuously refined and validated through real-world tests with the support of Melexis, ensuring alignment between virtual and physical performance. A highly accurate mathematical model of the sensor is also being developed and undergoing experimental validation to characterize exact limitations. This rigorous process is allowing the teams to quantify the sensor’s performance per application use case, laying the groundwork for broader deployment. This work is inherently supported by the sensors’ small, light, and inexpensive design, compared to other alternatives, which makes them suitable for a wide range of applications.

Potential Use Cases

While the research is ongoing and the AI model is being refined to further enhance system performance, Brubotics has successfully prototyped Melexis technology in a number of applications.

First is in rehabilitation and assistive robotics, where the team of researchers is exploring how robots equipped with Tactaxis® can be used to detect patient-initiated movement. Rather than acting as passive devices or rigid force applicators, the systems can instead more accurately infer movement intent and deliver proportionate, dynamic support — a crucial feature in devices like exoskeletons or therapeutic manipulators.

For generalized robotic manipulation, the collaboration between hardware and software entities shows how adaptive control based on accurate, 3D tactile data can improve the performance of grippers and robots. The project’s continued success highlights how engineers can design systems that detect changes in object stability and adapt gripping strategies – a crucial feature for humanoids, collaborative robots (Cobots), and autonomous mobile systems. This partnership uniquely combines public funding, academic research, industry know-how, and real-world testing, reflecting the complex process of robotic development to show how meaningful robotics applications can be accelerated from lab to deployment.

Conclusion: Moving Towards a More Responsive Robotics Landscape

While the Melexis-Brubotics collaboration focused on specific pilot applications, the implications extend much further. Tactile sensing of this kind, accurate, directional, compact, and cost-effective, has the potential to serve as a foundational capability for the next generation of robotic systems.

For cobots, the ability to detect and interpret contact events, not just at the actuator, but at the surface, allows for safer, more flexible behavior. Rather than stopping entirely when encountering resistance, robots can assess whether a nudge is intentional, accidental, or a signal to adapt. Tactile sensing enhances the ability of robotic grippers to respond to variations in object weight, shape, or compliance — traits that are difficult to assess visually. It also supports the use of softer, more compliant materials, since control is no longer solely dependent on fixed grip force or predefined trajectories.

Importantly, the rapid development of humanoid and service robots necessitates a true sense of touch. Here, this sense is vital for effective object handling and navigating environments that may be unpredictable, crowded, or specifically designed for human interaction. Tactile feedback enables these robots to adapt to minor contact, guide interaction through compliance, and behave in ways that are more intuitive for the people around them.

For Melexis, this collaboration reflects a broader commitment to innovation. By engaging with university researchers, supporting student involvement, and contributing its engineering capabilities to shared challenges, the company aims to translate foundational technologies into systems with tangible societal impact. As robotics continues to expand beyond the factory and into everyday life, collaboration of all types will be keen to develop safe, reliable, and meaningful robotic interactions. For more information on how Melexis’ advanced sensor technologies can help engineers to drive the next generation of intelligent robotic systems, please follow this link – www.melexis.com/robotics-e-guide