AI-industrie.png

Smart Manufacturing & Materials-iStock-1135159658-BEWERKT.jpg

<p><strong>Humanoid robots that are learning to walk, lift objects, tighten screws, or even work in a restaurant. It sounds futuristic, but development is moving at lightning speed. Behind many of these robots lies technology from Enschede. Xsens, which began as a spin-off from the University of Twente, provides motion-tracking technology that converts human movements into data. This information is then used to train humanoid robots using AI.</strong></p><p><img src="https://cms.aihub-oost.nl/uploads/Foto_Xsens_6ef563cb4c.png?token=vj9EF8FrlNYsjPVyVkNqB-motM3mBEEYkZKNCQiYCE8&amp;expires=1785162370" alt="Foto Xsens.png" srcset="https://cms.aihub-oost.nl/uploads/thumbnail_Foto_Xsens_6ef563cb4c.png?token=Y2qV-fP52onwU_zUDhGJIIC12vX8d869pX_r0jspf6M&amp;expires=1785162370 215w,https://cms.aihub-oost.nl/uploads/small_Foto_Xsens_6ef563cb4c.png?token=FdEPXYqi7pD5yh1HNF47f1NFov8lQn2tyBE4dkfqA1o&amp;expires=1785162370 500w,https://cms.aihub-oost.nl/uploads/medium_Foto_Xsens_6ef563cb4c.png?token=cxtCdMbL2du1ukYVTX8DnN-4GsQtP3A1b9F1ffgQB_k&amp;expires=1785162370 750w,https://cms.aihub-oost.nl/uploads/large_Foto_Xsens_6ef563cb4c.png?token=n_tKRpHbPn3fyx2vESd5_b62fs9qQqF7C66fjp_nAro&amp;expires=1785162370 1000w," sizes="100vw" width="1127" height="817"></p><p>“The ultimate goal is for robots to be able to act completely autonomously in human environments,” says Niek Hollak, Senior Director of Growth &amp; Business Development at Xsens. “And to achieve that, you first need to be able to digitize human movement very accurately.”</p><p>Xsens has been developing motion capture technology for many years. This involves recording human movements using sensors. The company gained international recognition in the entertainment industry. There’s a good chance that the movements of characters in games like FIFA or in Marvel movies were captured using Xsens technology. Now, the focus is increasingly shifting toward humanoid robotics.</p><h2>The Shift Toward Humanoid Robots</h2><p>The shift toward humanoid robots actually happened by chance. A major international automaker purchased systems from Xsens a few years ago. At first, the team thought the technology would be used for ergonomic research in the workplace. It wasn’t until later that it became clear the data was being used to train humanoid robots with the goal of eventually having them replace human tasks.</p><blockquote><p>That’s when we realized just how big this market could become</p></blockquote><p>“This gave us access to this development very early on,” says Hollak. Since then, Xsens has grown into a major player in humanoid robotics. According to Hollak, a large portion of the international humanoid industry now uses the motion tracking systems from Enschede.</p><h2>Motion Data as Fuel for Embodied AI</h2><p>Xsens’ technology uses small sensors that are attached to various parts of the body. A full suit contains seventeen sensors, distributed across the arms, legs, shoulders, chest, and head, among other areas. Together, they continuously record how a person moves. “Essentially, we digitize every part of the human body that moves,” Hollak explains. “We then link that data to biomechanical models we’ve developed in collaboration with universities.”</p><p>The major advantage over camera-based systems is flexibility. Because the system works without cameras, it can be used anywhere: in a factory, outdoors on the street, or during complex tasks. In addition, Xsens has developed technology that corrects for interference caused by metal and magnetic fields—a key advantage in industrial environments.</p><p>The recorded movements are used to train AI models from robot manufacturers. This allows robots to learn, for example, how to hold tools, pick up objects, or move safely in an environment designed for humans.</p><h2>Why Traditional Robot Programming Is No Longer Enough</h2><p>Whereas robots used to operate primarily via fixed programming code, humanoid robotics is increasingly centered on learning from data. And that, according to Xsens, is precisely where the challenge lies. “A human environment is full of variation,” says Hollak. “A small hill, a different angle of light, or an object positioned just slightly differently. That alone makes all the difference.”</p><p>Traditional programming falls short in this regard. Robots must learn to handle thousands of small variations in movement and environment. That’s why motion capture data is often combined these days with AI simulation systems from companies like NVIDIA. In these systems, recorded movements are digitally scaled up to thousands of variations.</p><blockquote><p>“We’re actually at the forefront of that training model,” says Hollak. “We provide the realistic human movements that those AI systems use to learn.”</p></blockquote><p>Specialized tasks are particularly valuable. Think of specific movements on production lines or complex physical tasks that are difficult to program manually.</p><h2>Robots for Physically Demanding and Repetitive Work</h2><p>According to Hollak, humanoid robots will primarily be used for physical, repetitive, and strenuous tasks in the coming years. “You’ll likely see the first applications in environments where many cyclical actions take place,” he says. “Work that people would rather not do, or that is physically demanding.”</p><p>Societal developments also play a role, such as labor shortages and an aging population. In the future, humanoid robots could perform support tasks in logistics, industry, or healthcare. At the same time, Xsens also sees applications in entertainment and hospitality, particularly in Asia. Think of robots that welcome guests, dance, or function as digital assistants.</p><h2>The biggest challenge isn’t the hardware</h2><p>Although humanoid robots are becoming increasingly advanced, according to Hollak, the biggest challenge lies not with the robot itself, but with training its “brain.” “The hardware is evolving incredibly fast,” he says.&nbsp;</p><blockquote><p>But the amount of data needed to make robots truly autonomous remains a huge challenge.</p></blockquote><p>That’s why Xsens is increasingly collaborating with other technology companies, universities, and research institutions. Motion tracking alone isn’t enough. Robots must also learn to sense, recognize objects, and understand why someone makes a certain movement. “Training robots is a multidisciplinary endeavor,” says Hollak. “It’s about movement, but also about context, intention, and interaction.”</p><p>Would you like to learn more about Xsens’ technology? Visit <a target="_blank" rel="noopener noreferrer" href="https://www.xsens.com/humanoid-robots">Humanoid Robots | Xsens.com</a> or contact Niek Hollak at niek.hollak@xsens.com.</p>

Xsens provides the motion data for humanoid robots

These entrepreneurs have paved the way for you

Meet our frontrunners

Foto Xsens.png

<p>Humanoid robots that are learning to walk, lift objects, tighten screws, or even work in a restaurant. It sounds futuristic, but development is moving at lightning speed. Behind many of these robots lies technology from Enschede. Xsens, which began as a spin-off from the University of Twente, provides motion-tracking technology that converts human movements into data. This information is then used to train humanoid robots using AI.</p>