A natural question is whether we can replace some of this real-robot work with a simulator. Classical physics-based simulators are powerful, but building one for a new task means manually modeling geometries, contacts, friction, and deformation, and the resulting simulator often still does not match reality closely enough for policies trained inside it to transfer.

In our work, we take a different route. We build an Interactive World Simulator: a learned, action-conditioned video prediction model that, given the current image and a sequence of robot actions, predicts the next frames purely in pixel space, with no physics engine inside. You can plug in a teleoperation device and control the robot through this learned world model for more than 10 minutes at 15 FPS on a single RTX 4090, and the predicted video stays stable and physically plausible.

The key idea is that, if the simulator is faithful enough, we could unlock two long-standing bottlenecks in robot learning:

– Data generation for training becomes cheap, because we can collect demonstrations inside the simulator.

What the world simulator can do

A few examples of what the model captures:

Rope routing: it correctly distinguishes between the rope actually being inserted into the clip and the rope swinging past it without making contact. Crucially, it does not bias toward either outcome — it follows what the actions imply.

Mug grasping: it captures fine-grained effects such as the mug slipping out of the gripper, or the handle being nudged and rotated.

Pile sweeping: it can generate video for multiple viewpoints consistently

You can try this directly on our project page: just open a browser and play with it using your keyboard!

How we built the interactive world simulator

At a high level, Interactive World Simulator is trained in two stages. First, we trained an autoencoder that compresses RGB images into compact 2D latent representations and reconstructs them back into images. This lets the model reason in a lower-dimensional space while still producing high-fidelity pixel-level outputs.

Second, we froze this autoencoder and trained an action-conditioned dynamics model in the latent space. Given past visual latents and robot actions, the model predicts the next latent state, which is then decoded back into an image. At inference time, this process is repeated autoregressively: predicted frames become part of the context for predicting future frames. Because prediction happens in latent space and uses consistency models, the simulator can run interactively while remaining stable over long horizons.

How is it different from prior works?

This differs from several existing approaches to robot simulation and world modeling. Classical physics simulators can be powerful, but they often require manually specifying object geometry, contact dynamics, friction, and deformation, and the resulting simulation may still have a large sim-to-real gap. Recent video-generation and world-model methods offer a more data-driven alternative, but many are either not explicitly conditioned on robot actions, too slow for real-time interaction, closed-source, or unstable under long-horizon rollouts.

In contrast, our Interactive World Simulator is action-conditioned, produces physically accurate pixel-level predictions, and supports stable interactions for more than 10 minutes at 15 FPS on a single consumer RTX 4090 GPU. This makes it practical not only as a visual prediction model, but as an interactive engine for collecting policy-training data and evaluating robot policies reproducibly. It has multiple applications.

Application 1: scalable data generation

If our simulator is good enough, can demonstrations collected using this world model actually replace real demonstrations for training imitation learning policies?

To answer this, we collected demonstrations entirely inside our world simulator and then trained imitation learning policies on 0% real-world data and 100% generated data. We deployed the trained policies directly on the real robot. The deployed policies do not just complete the tasks. They remain robust under continuous human perturbations. This demonstrates that the generated data quality is comparable to real data.

Application 2: faithful policy evaluation

Evaluating a robot policy in the real world is hard: you need to reset the scene, rerun the policy many times, and compare across checkpoints under matched conditions. In practice, this is not scalable and reproducible.

In contrast, we could roll out policies in our world model for reproducible evaluation. We evaluated the same four policies (DP, ACT, π0, π0.5) inside the simulator and on the real robot, with the same initial configurations. Each point in our evaluation is a policy checkpoint, scored in both real world and world model. We observe a strong correlation between the two scores, across tasks and across policies. Qualitatively, good policies that succeed in the simulator also succeed on the real robot, and bad policies that fail in the simulator also fail on the real robot.

Looking forward

Action-conditioned video prediction has long shown great potential for robotics, but existing approaches have been either too slow for interactive use or too brittle under long-horizon inference. By addressing both of these — stability and computational efficiency — the Interactive World Simulator becomes a practical engine for two things at once: training imitation policies on simulator-generated data that perform comparably to those trained on real-world demonstrations, and evaluating policies in a way that closely tracks real-world performance.

Going forward, we will extend the framework to more diverse environments and increasingly complex manipulation tasks, to further unlock the potential of large-scale robotic data. An important direction for future work is to study how the performance of world models scales with increasing amounts of interaction data and computational resources. Understanding these scaling behaviors could guide the design of larger and more capable world models for robotics.

The second day’s play at RoboCup 2026 has drawn to a close with another bumper set of matches. Teams have come from far and wide to take part in the humanoid soccer competition this year, with 17 different countries represented. China is the most represented country, boasting 15 teams across the three divisions. Other countries taking part are geographically widespread, ranging from Colombia to Malaysia, from Germany to Australia.

In advance of the competition, all applying teams provided a video, team description paper, and information about the robots and software that they use. You can see the complete set of these here. As a taster, here is the qualification video from team CAU Mountain&Sea (from China Agricultural University) who are currently leading the small division competition, and are in fourth place in the large division.

Yesterday’s play saw the first issuing of a red card. The robot received two yellow cards for unsafe challenges and was removed from play on safety grounds. You can see a clip of the second foul in question here.

In terms of the vital statistics of the robots, the heaviest and tallest robot is HERoEHS’s ALICE 4 robot, weighing in at 48kg and measuring 160cm tall. There is a big jump down in weight to the second heaviest bot, which is team BigHeroX’s Z4 bipedal humanoid, at 37kg. A number of teams are using the 35kg Unitree G1. At the other end of the spectrum, the lightest and shortest robot is ITAndroids’ Chape, at just 3.8kg and 53cm tall.

With the seeding rounds almost complete (the knockout rounds start tomorrow) the competitions are hotting up nicely. After four competed rounds in the small division, CAU Mountain&Sea is the only team to win all four matches, and sits atop the table with 12 points, and impressively only conceding one goal. Hamburg Bit-Bots and GeoHBots are in second and third place respectively, having both won three games and lost one.

The middle division has fast established itself as one of the most compelling competitions. B-Human has taken a commanding lead, winning all four matches so far with a +35 goal difference. Behind them, there are four teams on nine points: HTWK Robots, Rhoban, whIRLwind Amsterdam, and THMOS.

Over in the large division, teams have completed three rounds of seeding, with just one further round to come tomorrow. At this stage Tsinghua Hephaestus is the only perfect team, sitting on nine points. Behind them, two teams have each won two matches and drawn the third: PCMS-HRG and Robo-Erectus.

Once the seeding rounds have been completed, 12 teams from each division will make it forward to the knockout stages, with the quarter finals taking place in the evening.

Many thanks to JT Genter for providing the photos, videos, and facts for this article.

From the 1st-5th June, the robots descended on Vienna. The 2026 IEEE International Conference on Robotics & Automation (ICRA) brought together the top minds in robotics for one short week to showcase the latest technologies, form new collaborations, and exchange ideas. Held at the Messe Wien, a stone’s throw from the bank of the Danube, ICRA proved to be equal parts technological marvel and thought-provoking discussion. 

Workshop on robot ethics

My week at ICRA began with the 2nd ICRA 2026 Workshop on Robot Ethics: Ethical, Legal and User Perspectives in Robotics & Automation (WOROBET). WOROBET provided a space for researchers to share ideas, thoughts, and concerns on the future of robot-human interaction, and how to create ethical frameworks to navigate this rapidly changing technology. 

Yasuhisa Hirata, Professor at Tohoku University, began by presenting his vision of a world with physically assistive robots, such as detachable exoskeletons or cycling wheelchairs. These tools can affect people’s sense of self-efficacy and motivation, and there are a host of ethical implications that come with this – how do you help people just enough to build their confidence, without slipping into deception? 

We then heard from Prof. Minoru Asada from Osaka University, who discussed his aim to implement pain signals into robots, so they can experience the world as we do. This was a highly interesting and niche proposition that brought up more ethical questions than we have answers for at the moment. Perhaps most pertinent to a technical conference: is a sense of embodied morality necessary for true intelligence? 

Alan Winfield, Professor of Robot Ethics at UWE Bristol, presented a vision of robotics that acted as a counterweight to Prof. Asada’s: robots as tools, not potential beings. This also somewhat reflects differing attitudes in Eastern vs Western cultures. Thinking about the practical issues we are likely to face in the near term, he outlined a framework for social robot accident investigation. In his view, robot ethics is not just an engineering problem, but requires appropriate social and governance frameworks, just as we do for aviation. His talk also emphasised the risk that programming ethics into robots runs the risk of removing moral responsibility from the roboticist, and can always give rise to unethical robots via malicious hacking. The theme that we must focus on human morality, as opposed to machine morality, was repeated throughout the day. 

After a morning of differing ideas and visions of what robot-human interaction could be, we were invited to ground this into a real robot social care scenario by Praminda Caleb-Solly, Professor of Embodied Intelligence at the University of Nottingham. In this red-teaming exercise, we examined the safety risks and possible mitigations of an assistive robot for a schoolteacher recovering from a stroke at home. Our group discussion circled around human agency: how ethical is it to make design choices for people that take away some of their autonomy, in the name of their best interests? As robots in social care will become a more urgent need in the years to come, these questions may become more salient. 

I left WOROBET with plenty to think about, and a renewed sense of appreciation for the ethicists who are already grappling with the problems that are to come. I hope that progress in robot ethics keeps pace with progress in robotics, so that we are well prepared as robots become more of a part of our daily lives.

Welcome to the jungle – the robot exhibition floor

Every time you entered the exhibition hall, you were greeted by one of the child-sized Booster robots, either playing football, dancing, or demonstrating some kung fu. They were always an endearing welcome to the sea of robots. 

The robots ranged from the endearing to the uncanny, but the common thread was their technical capabilities were astounding. Veteran attendees consistently remarked on how much the robots had improved year on year.

The cute.

The below clips show the robots that most caught my eye. After admiring a phosphorescent, Stranger Things-esque robotic flower display, I was blown away by the D1-modular robot from Direct Drive. Unlike most robot dogs, it can split into two halves, with the ability to jump, twist, and traverse difficult terrain. 

<span data-mce-type="bookmark" style="display: inline-block; width: 0px; overflow: hidden; line-height: 0;" class="mce_SELRES_start"></span><br />

Sharpa’s North was always a friendly face, waving and making love-hearts at visitors to its booth. Around the back of the booth, you could even challenge it to a round at blackjack. We saw humanoids zipping up rucksacks and trying to fold laundry. Enchanted Tools’ social care robot, Mirokaï, was an unusual sight among the mass of black and steel, with a bright orange body, feline ears and an orange, furry face. Tesollo’s humanoid spent much of its time at ICRA using its long, wavering arms to pick up fruit and drop it into baskets, with impressive dexterity. Vietnamese company Vinrobotics’ humanoid offering was reminiscent of the Cybermen, with its gently wheezing joints, but the team assured me it was much friendlier. The pint-sized Boosters were almost always playing football, not far from their similarly sized Agibot cousins. 

<br />

However, humanoids didn’t steal the show this year, as they have done previously. The big trend this year was robotic hands, and the levels of dexterity were truly impressive. Closing the gap between human and robot abilities here would unlock whole new swathes of tasks to automate – and who wouldn’t want a robot folding their laundry? 

<br />

Industrial challenges: solving dexterity 

Tackling the dexterity challenge really defined the industrial talks for me this year. A talk by ARIA’s program director, Prof. Jenny Read, demonstrated how the UK government is already laying the groundwork here.

They outlined their funding proposals as part of their Smarter Robot Bodies program, which is split into two branches: robot locomotion and robot dexterity. The robot locomotion branch aims to enable robots to traverse messy, unpredictable physical environments, while the robot dexterity branch will try to break the bottleneck of adept physical manipulation by robotic hands. With the programme set to launch in early 2027, it’ll be exciting to see what kind of innovations this attracts. 

One standout innovation in the realm of dexterity was TARS, a record-setting newcomer in the Chinese robotics market. Co-founded by Dr Ding Wenchao just 18 months ago, TARS has already raised the most funding in angel and pre-seed rounds of any company in the Chinese embodied intelligence sector, and achieved a Guinness World Record for robotic flexible wiring-harness insertion completed in one hour. 

TARS’ DexHand. Image credits: TARS.

DexHand is a 1:1 model of the human hand, even replicating the 21 degrees of freedom we have in the wrist joint and hand. According to TARS, “it can interpret tactile data to distinguish slipperiness, roughness, and hardness in real time and perform 26 English alphabet hand gestures with high-precision finger control.” I was given the chance to control the hand using my own, and was impressed by how well it emulated my own movements. I was also impressed by their humanoid zipping up a backpack – despite the technical abilities of all the robots on the floor, few were able to perform tasks which required such fine motor skills. 

TARS robot zipping up a rucksack.

On Thursday, Dr Wenchao Ding delivered an industry keynote where he presented TARS’ roadmap, charting the path from academia to industrial deployment. With an impressive academic team behind them, TARS may be one to watch. 

Plenary talks

Aside from the wealth of invention and innovation taking place in the exhibition hall, the breadth and depth of academic research at ICRA was fantastic. The plenaries especially gave an insight into the research trends that are currently defining the field. 

Ken Goldberg delivered an electrifying plenary, titled “A Tale of Two Cultures:  Can Agentic Coding Close the Gap?” In this talk, he called for a step change to close the data gap faced by robot manipulation. With the rise of diffusion models and LLMs, it is clear that big data has solved computer vision and language. He challenged the audience – when will the ChatGPT moment for robotics come? With state spaces larger than 50 dimensions in robotics, there is not enough training data to close this gap. Currently, the data required to train vision-language models is equivalent to 100,000 years of real physical experience. 

According to Goldberg, the two dominant cultures in engineering – model free “good old fashioned” engineering (GOFE), and, the currently more popular model based engineering.  GOFE encapsulates rigorous engineering methods that pre-date AI, but may have been slightly forgotten about in the AI wave of recent years. He highlighted how in his career he has always been working to bridge the gap between two cultures: from science and art to robotics and automation. 

He outlined four possible solutions to the data gap:

Prof Goldberg described how he used the fourth approach in his robotic delivery packing company, Ambi Robotics, for 22 years. Picking up bags is an example of variational automation –  one task is done repeatedly, but with different initial conditions each time. From this rich dataset, they created a generative model to train robots in the best way to pick up bags. Here, they close the gap between model-free and model-based methods – Ambi exploits both to achieve industry-leading results. This plenary served as a call for other researchers to use their own production data to do the same. 

During Thursday’s keynote on Robot Learning, Planning & Foundation Models, Stefanie Tellex from Brown University gave a compelling talk titled “Towards Complex Language in Partially Observed Environments”. While current research is bounded in known, predictable scenarios, using action-based language, this does not reflect what the real world is like, nor how people would naturally communicate with robots. Prof. Tellex described her work creating robots that can understand complex, goal-based commands in only  partially observed, dynamic environments, and outlined the grounded Turing test – a reimagining of the Turing test for embodied AI. 

An example of a robot performing a goal-based task in a dynamic environment. Credits: Tellex et al, 2026.

Both plenaries spoke to current pinch points in robotics: data, reasoning, and operating in complex real-world environments. It’ll be interesting to see what solutions are developed in the coming years. 

Science communications crash course

One of my favourite parts of ICRA was delivering the Science Communications Crash Course. Along with Robohub Executive Trustee Sabine Hauert, IEEE Spectrum Senior Editor Evan Ackerman, and IEEE Spectrum Community Manager Kohava Mendelsohn, we gave our guidance on effective science communication to an audience of 100 academics. It was encouraging to see so many people interested in communicating their research effectively – it is a crucial skill, especially in the era of AI and robotics when mainstream narratives can be hijacked by doom-mongering, hype, and corporate interests. More academics communicating their work clearly and neutrally will go a long way to grounding our societal discussion in technical reality, not sci-fi futures. 

Art and robotics 

The arts and robotics section was rich and interesting. There was a lot to visually take in, with constant background music from a robotic saxophone. 

PET – marked by a large “PET ME” sign – was a white and orange mass of connecting, rotating pyramids, that gently pulsed and hummed in response to touch, responding by curling towards or away from you depending on how you touched it, The effect was strangely lifelike. 

Rhombus Research presented “Reptile: A Bio-Mimetic Choreography Engine for V2X and A2X Swarms”. This artistic simulation visualises the contracts negotiated within autonomous vehicle and swarm fleets in dreamy blue, browser-based visualisation. Performance artist and former Cirque de Soleil acrobat Silke Grabinger explored human-robot interaction via her piece, AREYOUARE. 

Silke Grabinger performing AREYOUARE. Image credits: ©Denes Erdos – Your Event Photographer

I’d also like to acknowledge some of the video creators I met there. YouTubers Back to Engineering and the. Amazing, PhD are making some fantastic videos about physical AI. 

Nothing lasts forever – ending with the robot parade 

ICRA 2026 ended with the robot parade, which attracted quite a crowd. See if you can spot the panda, dragon, and headless humanoid!

<span data-mce-type="bookmark" style="display: inline-block; width: 0px; overflow: hidden; line-height: 0;" class="mce_SELRES_start"></span><br />

Seeing all the robots gathered together was a real spectacle. The technology on display was state-of-the-art, and it only improves year on year. What struck me most was the sense that robotics is moving from proving what is possible to tackling the remaining barriers to real-world deployment. Across the exhibition floor, industry keynotes, and plenary talks, the focus was often on the same challenges: dexterity, data, and operating reliably in complex environments. Solving these bottlenecks will open up new avenues to real-world applications of robots. 

While workshops such as WOROBET highlighted important questions around ethics, agency, and governance, the overwhelming emphasis at ICRA 2026 was on capability. Researchers and companies alike are working to close the gap between what robots can do in carefully controlled demonstrations and what they can do in the messy reality of the world outside the lab. Judging by the pace of progress on display in Vienna, that gap may be narrowing faster than many of us expected. I hope that the kinds of conversations around human-robot interaction and ethics that were commonplace at WOROBET and in the Arts exhibition space will become more mainstream – we may need to face the questions that they raise sooner than we think. 

Note: Where image and video credits are not stated, they belong to Ella Scallan.

Cornell engineers have developed a robotic collective that behaves less like a machine and more like a material that flows, reshapes and adapts to its environment without centralized control.

The system, called the Cross-Link Collective, consists of dozens of small robots that have limited mobility individually, but together exhibit coordinated and sustained motion. The research, published May 20 in Science Robotics, demonstrates a robotic system that resembles soft matter, continuously deforming and reorganizing as it moves, driven by what researchers call mechanical intelligence.

“Instead of relying on explicit computation and communication, the system shifts the intelligence into the shape of the robots and their physical interactions,” said corresponding author Kirstin Petersen, associate professor of electrical and computer engineering and the Aref and Manon Lahham Faculty Fellow in the Cornell Duffield College of Engineering. “We’re leveraging the contact dynamics to let useful behaviors emerge, so the system naturally settles into configurations that reduce internal stresses and improve motion.”

Each robotic module measures about 200 millimeters in length and 20 millimeters in width, and contains a small motor that drives it to oscillate between two shapes, an “I” and a “U.” These oscillations generate forces against the ground, allowing the modules to inch forward and jostle into one another. At each end of the module are weak Velcro patches, enabling them to temporarily latch and unlatch onto neighboring modules.

On their own, the modules move slowly and inefficiently. But when they entangle into chains, they begin to move collectively, self-organizing into shifting configurations that prove resilient in challenging environments.

“It doesn’t matter if one module has a compromised battery or fails for other reasons,” said lead author Danna Ma, visiting lecturer in electrical and computer engineering. “The system stays functional because it can adapt. It is redundant and doesn’t depend on any single module.”

Despite the minimal approach, the researchers showed that even a small amount of computation can improve system properties. To enhance cohesion, isolated modules emit an audible distress signal, prompting nearby modules to slow down and allow the straggler to reconnect.

“There is no centralized sensing or control,” Ma said. “Each module can infer when it has lost contact with the group by how much it’s being jostled and then use an audible buzz to slow down nearby modules while it catches up. It’s as simple as that.”

Co-authors at the Georgia Institute of Technology developed the original design of the module, which Petersen and Ma refined over years of experimentation and statistical analysis to improve its ability to entangle and operate in large numbers. That process revealed how even subtle changes in module size and other characteristics can influence how effectively they connect and move as a group.

The Cross-Link Collective draws inspiration from active gels – materials whose molecular links continually form and dissolve while maintaining overall structure. The findings could help inspire new forms of soft-matter engineering, though the researchers mostly see the system as a tool for studying how mechanical intelligence can give rise to resilient emergent behaviors in robot collectives.

“It’s helpful for us to start thinking about what we can encode into the physics of a system itself, as robots are increasingly applied to real-world scenarios that are highly unreliable and dynamic,” Petersen said. “Counterintuitively, by giving up exact control over configurations and coordination, we gain a surprising range of useful behaviors.”

Cornell researchers used stretchable fiber-optic sensors to create a soft robot gripper that can predict the ripeness of strawberries by touch. Credit: Anand Mishra.

By David Nutt

When assessing the ripeness of fruit, sight and smell can tell you a lot, but the best indicator is often how the fruit feels.

Cornell researchers used stretchable fiber-optic sensors to create a soft robot gripper that can predict the ripeness of strawberries by touch, then gently twist them off their branch or vine without causing any damage.

The technology, developed in the lab of Rob Shepherd, the John F. Carr Professor of Mechanical Engineering in the Cornell Duffield College of Engineering, could lead to more resilient and ecological food production and increase the availability of fruit species that are difficult to cultivate.

Shepherd’s Organic Robotics Lab previously demonstrated the potential of stretchable fiber-optic sensors to give soft robotic systems the ability to feel the same dynamic, tactile sensations that enable humans to navigate the natural world. In recent years, the team has expanded into agriculture, designing a soft robotic gripper that injects living plant leaves with sensors that help it detect and communicate with its environment.

“The great thing about Cornell is we’re a really great agriculture school, and a lot of avenues are opening up because of it,” Shepherd said. “It really allows us to uniquely combine our robotics expertise with our agricultural prominence.”

To develop a way to evaluate and handle fruit with care, Shepherd’s team partnered with Marvin Pritts, professor of horticulture and global development in the College of Agriculture and Life Sciences, who specializes in developing sustainable production methods for berry crops.

In order to train and test their gripper, they needed a model fruit. And for that, they turned to the strawberry.

“You can accurately tell when strawberries are ripe by their color,” Shepherd said. “So we could train our model to know if it’s ripe based on touch, then validate our model by looking at the color. And Anand was able to accurately estimate whether it was the right time to pick strawberries based off of the stiffness he measured.”

The soft robot gripper has an equally soft touch. The gripper is equipped with two different fiber-optic sensors, one to measure the curvature of the finger, and the other to measure the pressure at the fingertip. This way the robot can estimate the shape of an object and adjust its grip accordingly to grasp the ripe fruit without bruising it.

“The fiber-optic strain gauges have the same mechanical properties as the grippers that are using them. So it’s kind of like the flesh feels the fruit, rather than having separate sensors,” Shepherd said.

The researchers also included a planetary gear mechanism so, once the fruit is grasped, the robot wrist can rotate and twist the strawberry off its vine, instead of pulling or plucking it, which can strain and damage the fruit.

This soft-gripping technology, developed in the Organic Robotics Lab, could lead to more resilient and ecological food production and increase the availability of fruit species that are difficult to cultivate. Credit: Anand Mishra.

For cases in which touch isn’t enough, the researchers installed a camera in the gripper’s palm to find fruit that are occluded by leaves or other vegetation. However, the device will be particularly handy when ripeness can’t be detected visually, such as for avocadoes, pineapples or – Shepherd’s favorite – pawpaws.

“The problem with pawpaws is you can’t see when they’re ripe, and they ripen so fast that if you’re not there at the right time, you just miss them,” he said. “And you can’t harvest and ship them, because they don’t survive shipping very well, either. That’s one reason we don’t have pawpaws in grocery stores. But this can help with that.”

The technology could have an even greater impact on sustainable farming practices.

“Robots will allow us to do things we cannot do economically right now. We have row crops because row crops fit our machines. But if we have a larger amount of smaller robots, we can have mixed cropping of different species that support each other,” Shepherd said. “Instead of having soy one year and corn the next, you can have them both. Or you could have interspersed species that are resistant to pestilence, that help block infestations and reduce the amount of pesticides and fertilizer. You can have drought resistance from canopy species.

“It’s very complicated to manage a farm that way,” he said, “and robots could allow us to do that.”

The research was supported by the National Science Foundation Center for Research on Programmable Plant Systems (CROPPS) and the Cornell Institute for Digital Agriculture.

Ace rotates its paddle as it prepares to return the ball back to its human opponent, Yamato Kawamata, during a match in December 2025. Credit: Sony AI.

In an article published today in Nature, Sony AI introduce Ace, the first robot to beat elite human players in competitive physical sport.

Although AI systems have shown advanced performance in digital domains and board games (such as complex video games, chess and Go), translating this to physical performance has remained a significant challenge. Such a feat requires perception, planning, and control to work in a high-speed domain on the scale of milliseconds. Table tennis is a demanding and complex real-world test for robotics, requiring rapid decision-making, precise physical execution, and continuous adaptation to an unpredictable opponent. The ball’s high speed, spin, and complex trajectories are central to competitive play.

Director of Sony AI in Zürich, and project lead for Ace, Peter Dürr said “this research has shown that an autonomous robot can, in fact, win at a competitive sport, matching or exceeding the reaction time and decision making of humans in a physical space. Table tennis is a game of enormous complexity that requires split-second decisions as well as speed and power. This research breakthrough highlights the potential of physical AI agents to perform real-time interactive tasks, and represents a significant step toward creating robots with broader applications in fast, precise, and real-time human interactions.”

A complete view of table tennis robot, Ace, including arm and track. Credit: Sony AI.

Ace combines event-based vision sensors and a control system based on model-free reinforcement learning, as well as state-of-the-art high-speed robot hardware. Ace was designed with three novel components:

Members of the Ace research team and table tennis officials pose with the robot and its human opponent, Mayuka Taira, following an official match in December 2025.

For the results reported in the Nature publication, Ace was evaluated in matches against five elite players and two professional table tennis players, under International Table Tennis Federation (ITTF) regulations. Ace achieved three victories in five matches against the elite players, along with competitive performances in the remaining matches.

There were some interesting results from the evaluations, including the fact that Ace was able to return a wide range of spins, consistently achieving over 75% return rate up to spins of 450 rad/s. The control systems behind Ace also allowed for quick reaction to unusual shots, such as balls bouncing off the net. This behavior illustrates the ability of the approach to generalize to situations that are both rare and hard to model in simulation.

Following submission of the Nature manuscript, the team conducted additional competitive matches in December 2025 and March 2026, beating professional players in the process. Compared with earlier evaluations, Ace demonstrated higher shot speeds, more aggressive placement closer to the table edge, and faster-paced rallies, reflecting continued performance gains under competitive conditions.

Find out more about the project in this video from Sony AI.

The Consumer Electronics Show (CES) is one of the largest trade events in the world. Every year, thousands of companies showcase their state-of-the-art technologies to over 100k attendees. It brings together global industry leaders, startups, and media, and is used to launch products and signal future tech trends.

Henry Hickson, a Research Associate at the Hauert Lab, attended CES 2026. In this article he reports on the highlights of the show, exploring how physical AI is moving from experimental labs into consumer technology, and what this shift reveals about the future of robotics, autonomy, and real-world deployment.

All photos and videos in this article are credited to Henry Hickson.

It doesn’t take long after landing in Las Vegas to see the signs of CES 2026. Within minutes, gleaming LED screens announce the show’s presence, and those ubiquitous two letters: AI. One flavour of AI dominated the show, from Jensen Huang’s keynote speech to even the smallest booths on the show floor: physical AI.

While I went there looking for robotics, the phrase ‘physical AI’ was reaching the ears of even the most casual visitor. Brands best known for televisions, laptops, and home appliances were talking confidently about embodied intelligence. At CES 2026, physical AI – in practice, robots – was no longer niche.

As a first-time CES attendee, the sheer scale of the event was overwhelming. Even with ruthless efficiency, it would have been impossible to speak to every robotics exhibitor at the Vegas mega-show. What follows, then, is not a comprehensive survey, but a set of themes that emerged repeatedly over the week – reinforced by either repetition or hype surrounding the exhibits. Together, they offer a snapshot of where consumer-facing robotics stands today, and where it appears to be heading next.

A quick caveat for seasoned roboticists: this consumer space is a far-cry from our daily interactions with lab-based robotics. Much of what’s on display is not state of the art by academic standards – but it is innovative in a different way, shaped by consumer expectations and manufacturability.

Outside CES at the Las Vegas Convention Center. 

It would be impossible to feature robotics at CES and not expect a wave of the biggest trend in recent robotics history: the rise of the humanoid. What once felt like a sci-fi dream is now a serious commercial and research focus, attracting intense interest from investors, companies, and the public. In some circles, the debate is whether humanoids are necessary at all, given the efficiency of other forms. At CES, however, they were possibly the most widespread and discussed representation of physical AI.

The range on display was vast. Booster showcased child-sized humanoids which deliver an education program on robotics. These were lined up in what resembled a classroom and were available to buy then and there – many of the classroom robots had bold red ‘SOLD’ signs attached. One Booster would set you back $10k.

Booster’s silicon class-bots wave in unison. 

Unitree drew crowds with a non-stop show of boxing, dancing, and acrobatics. While lacking in real-world application and still largely remote controlled, their rapid progress in motion capability is impressive. Unitree has achieved this with improved actuators and reinforcement learning. Their team stated that they added instabilities to their simulations, which makes the system more adaptable and gives confidence that each trick will run smoothly during demos. These boxing bots retail for $6k+ and are customisable for user needs.

Beijing-based Galaxea Dynamics are targeting research labs with a high-precision platform for foundation model development. Beginning with a simple robot arm with a camera and pincer gripper, they now offer a humanoid with inbuilt height adjustment for $70k. In addition, they are developing an open-source VLA as a starting point for users developing their own, task-specific setups. The result is an out-the box VLA system, reducing the barrier to entry for universities or R&D companies to develop their own twin armed VLA systems. Galaxea are proving especially popular with US institutions.

Home robots were one of the most popular categories at CES. SwitchBot presented H1, a wheeled humanoid with arms that slide up and down. At this stage it is likely becoming clear that the definition of a humanoid is becoming somewhat blurry – aside from the friendly face and human-dimensioned arms, the H1 isn’t exactly a human form. Trained by observing humans completing household tasks, the H1 has so far mastered tidying around the house and loading the washing machine – if still at a slower pace than a human. Release dates are uncertain, but the company is aiming to undercut the competition by reaching a $10k price point. It’ll be interesting to see how this develops.

SwitchBot’s H1 Humanoid, ready to do your chores (very slowly).

The GR-3 humanoid by Shanghai-based Fourier caught my eye – its turtle-neck styled torso, emotive LED eyes and automotive-upholstered soft skin makes for a standout aesthetic. Its dextrous hands were shown off in close interactions with visitors via tic-tac-toe games, where it never lost. I was happy to pull off a tie. The reason for these close interactions is that GR-3 is targeted for healthcare providers, for social interaction and physical tasks around care homes and hospitals. Retailing at $100k, the primary customers are still research labs.

Fourier’s tic-tac-toe playing humanoid, GR3. 

An unexpected development for many was the entrance of large consumer brands into the humanoid space. LG’s CLOiD, integrated with its smart home ecosystem, embodied the idea of a “zero-labour home.” The CLOiD is a wheeled humanoid with a friendly LED face. Somewhat tentative in its movements, it was able to retrieve items from the fridge and fold a towel. This is still very much in development, but watch this space.

LG’s wheeled humanoid, CLOiD.

Paxini’s offerings sit exactly in the popular intersection between humanoid and wheeled robots, with an adjustable wheeled base, a floating torso adjustable up and down by rotating joints, and a humanoid upper body. Utilising VTLA models, they are targeted to move crates in warehouses or parts on factory floors, with several units already deployed.

GalBot presented another wheeled, VLA-focused platform, with demonstrations of crate stacking in a warehouse-style environment and customer service tasks in a mini convenience store setup. The platform features an adjustable torso with attached arms and head. The company has invested heavily in VLA development for manipulation, navigation, following, and dancing. They highlighted the challenge of tuning models for each platform, again pointing to data collection as a key bottleneck. Despite this, the team is pushing toward near-term deployment.

The aptly named company Humanoid, representing the UK contingent of roboticists, showcased a deployment-focused wheeled system in the form of the HMND 01 Alpha Wheeled platform. The large rectangular industrial base clearly signals a platform designed for real-world factory work rather than show-floor demos. This focus was reinforced by Humanoid’s recently announced partnership with Schaeffler, with whom they were exhibiting at CES. A smooth demo showed the height-variable robot moving metal rings between bins. While near-term efforts are firmly industrial, the team stated that a legged version better suited to home environments remains part of their longer-term roadmap.

Schaeffler’s second humanoid partner at CES, Agility Robotics, is also focused on production-ready systems for the factory floor – this time, with a fully legged platform. Their distinctive backward-knee design has been under development for many years, and the full-height robot on show was clearly built for industrial environments, with exposed metal structures and hard casing. Agility has no plans to enter consumer home robotics.

Perhaps most notable at CES was Boston Dynamic’s Atlas humanoid. It saw 4-5 row deep crowds for the entire show, as well as hitting global headlines. Atlas has been in development for over a decade and has become widely known through videos openly showing both successes and failures. The latest version was unveiled during the opening day. It performed an agile sequence of crouching, sweeping, and pointing motions. For the rest of the show, it performed regular pick-and-place demonstrations in a mock factory environment. The team openly discussed current limitations, noting that much of the system was teleoperated during the show, highlighting that the software stack is still under development.

Standing 1.9 metres tall with a lifting capacity of 50 kg, it is designed for heavy-duty work. Hyundai plans to focus production capacity on internal deployments through 2026, with broader manufacturing customers expected later. Long-term plans include scaling production to tens of thousands of units per year, signalling what may be one of the first large-scale deployments of humanoid robots with real industrial value.

The headline-grabbing Atlas humanoid demonstrates its impressive range of mobility.

Being consumer focused, it is perhaps unsurprising that a standout theme was so consumer oriented, but I was surprised by the sheer number of newly released robots in this space. The strength of this theme lies in human–robot interaction, and in how consumer technology is being reshaped by robotics and AI.

One example was Loona, built by Beijing-based KEYi Tech. This pint-sized, four-wheeled robot pet was surprisingly insistent, identifying potential playmates through an onboard camera and bumping clumsily into their feet to ask for attention. Designed for all ages, the team reported strong interest from adults buying Loona as company for elderly parents. With some life-like qualities, Loona manages to evoke some of the emotional response of a household pet.

Loona’s adorable robotic pet.

KEYi also previewed a desktop robot concept that integrates with a smartphone, which provides access to cloud tools like ChatGPT and acts as a screen for expressive robotic eyes. Positioned as an assistant, the demo showed agentic capabilities, such as booking meetings and analysing screenshots. Its long-term usefulness remains unclear, but the slick interface was fun to interact with.

Next up was an unreleased robotic pet from OlloBot – two-wheeled, dog-sized robot pets with tablet-based faces that reminded me of Sid the Sloth from Ice Age. These child-friendly bots had ChatGPT integration as well as camera modes and learning programs.

Ollobots – your child’s new playmate or favourite toy?

Fourier presented a small, customisable desktop robot with wide digital eyes and head tracking that gave an eerie, Labubu-esque aesthetic.

Fourier’s cute desktop robot is watching you. 

Ludens AI offered INU, a fluffy desk robot. Marketed as a ‘desktop alien’, INU has come to Earth to remind users to take breaks and not work too hard. How could you say no to that? Alongside the cute rocking movements and occasional squeak, it was hard not to feel some attachment to this little robotic being.

The INU desktop alien robots waving hello. 

Rounding out the category was Sentigent’s DuoRover, a rugged, follow-me robot for outdoor adventures. Using a two-wheeled balancing design and vision-based following, it is intended to provide conversation, photos, and video recording on family outings.

Companion robotics is clearly a rapidly growing space. The open question is longevity: will these robots remain companions, or be relegated to people’s cupboards once the novelty wears off? Companies are betting that continual feature expansion, fueled by LLMs, will help them endure – but only time will tell.

While less prevalent at a consumer-focused show like CES, task-driven robots still made a strong impression. Typically found at trade shows, these more industry focused robots are used for warehouse / factory automation, or in hospitality. However, there was an interesting sprinkling at CES too and some great innovations in this space.

One of the most notable examples was from the Hyundai Motor Group, whose acquisition of Boston Dynamics made them one of the most popular exhibitors. Their MobED robot is a modular wheeled platform with four independently controlled wheels that can perform tight manoeuvres and can traverse varied terrain whilst maintaining a perfectly level loading bed. The modular approach means this can be used for a variety of uses cases, from delivery carts to ride-on vehicles. What stood out most was the build quality, which felt more akin to a luxury car than a typical robotics module. It will be interesting to see how this does against its lower cost competition – is there an appetite for luxury robotics platforms yet?

Hyundai’s 4-wheeled modular platforms greet the crowd by waving their wheels. 

At the other end of the spectrum was Nexus Patio Tech from Bolivia, showcasing delivery robots designed for simplicity and rapid route-to-market. With 20+ robots already operating on Bolivian streets, and many more in malls and restaurants, it was great to see a deployment first approach – although this may partly be thanks to a different regulatory landscape.

China’s Neolix demonstrated logistics at scale, with over 15,000 autonomous vehicles deployed across three sizes. By focusing exclusively on logistics, the team has simplified decision-making and accelerated deployment. For example, if there is a choice between crashing into a wall or a pedestrian, in logistics only you always choose the wall, as there is no passenger onboard to account for. This makes for a unique approach compared to some of the other, albeit highly sophisticated offerings on show at CES, such as passenger focused Waymo, Zoox, and NVIDIA. Pilots in Portugal are scheduled for the first half of 2026, which is a promising step – being able to comply with stringent EU regulations would unlock deployments in several other regions.

Neolix’s autonomous robot. 

Construction logistics were addressed by Gole Robotics from South Korea, who are tackling point-to-point material transport on construction sites – a use case so far relatively untouched by autonomous robotics due to its complexity. Their large material-carrying robot is designed to move pallets overnight in apartment buildings, allowing human workers to focus on skilled labour during the day. Here, the innovation lies as much in the redesigned workflow as in the robot itself.

Finally, USX Robotics showcased a modular indoor delivery platform for hotels and events, notable for having been developed from concept to production-ready system in just over a year.

This was not the only robot to be market-ready in lightning-fast time. With the availability of affordable and specific robotics focused components, open-source baseline libraries and easier access to manufacturing services from abroad, both the cost and complexity of building a new robotic system seems to be decreasing, enabling a host of new players into the industry. This raises interesting questions about how the market will evolve in the future. Differentiation may increasingly lie in identifying the right use cases and routes to market.

CES 2026 showed that humanoid and embodied AI systems still have a long way to go before delivering real-world value, particularly in homes. At the same time, there is a growing sense that the path to deployment is becoming clearer. A consensus has emerged across platforms: multi-camera perception, often wrist-mounted, paired with VLA models, is sufficient for most tasks. Increasingly, tactile hands and VTLA software are added.

There was a clear split between industrial and home-care humanoids. The most convincing offerings with the most deployments were in industry, with the firm favourite being logistics tasks in the manufacturing sector. With highly developed platforms like Atlas leading the charge, this industrial humanoid push feels closer than ever. Another general trend here was reaching the market via industry, followed by a longer-term goal of reaching people’s homes in some unspecified time frame. It was sometimes hard to tell whether that home-ready goal is genuine or to please the likes of investors and the excitable general public.

One interesting question for me is capability – at what point is a humanoid considered useful enough for the time-constrained consumer to buy it, not just for an early adopter? Currently, these platforms feel too slow to be useful enough for the average buyer. As development continues, I hope we begin to see more adaptive systems that provide genuine value.

What has changed is momentum. A spark of excitement has captured the imagination of so many at CES, now that what once seemed impossible is for sale, right in front of you. With greater access to compute, the development of foundation models like VLAS and VTLAs, a roadmap is emerging. The race now is for data, data, and more data –it’ll be a fascinating one to follow over the coming years. I left CES with the sense that the future is happening now.

* Watch a compilation of all the clips featured here and more on our youtube channel:

A core objective of RoboCup is to promote and advance robotics and AI research through the challenges offered by its various leagues. The ultimate goal of the soccer competition is that, by 2050, a team of fully autonomous humanoid robots will defeat the most recent winner of the FIFA World Cup. To bring this vision closer to reality, the RoboCup Federation has announced several changes to the leagues. We spoke with Alessandra Rossi, a trustee who has been involved in the humanoid soccer league for many years, to learn more.

I am Alessandra Rossi from the University of Naples “Federico II”, where I am an Assistant Professor of Computer Science. I began working with and collaborating in RoboCup in 2016, when I started my PhD at the University of Hertfordshire in the UK. I am still affiliated with the University of Hertfordshire, as I remain a member of the humanoid KidSize team Bold Hearts, the longest continuously active team in the UK. After a few years, I became the team leader of Bold Hearts.

In 2019, I became a member of both the Technical Committee and the Organizing Committee of the Humanoid League. After serving on the Technical Committee for two years, I was elected to the Executive Committee of the Humanoid League. In 2025, I was elected to the Board of Trustees for the first time.

Over the years, I have steadily increased my involvement and commitment to RoboCup. I have always sought to actively engage the RoboCup community, both during competitions and outside of competition periods. I also work to encourage engagement between the major and junior leagues and to participate in regional RoboCup events.

While working at the University of Hertfordshire as a Visiting Lecturer, we launched an online module that uses RoboCup as a benchmark for teaching robotics to undergraduate students. The module is still running. I initially served as the module leader, and this role has since been taken over by our Bold Hearts teammate, Bente Riegler.

Last year, Maike Paetzel-Prüsmann, Merel Keijsers, and I (as lead authors), in collaboration with several trustees and many members from different leagues, published a paper on the current and future challenges in humanoid robotics. The paper was published in Autonomous Robots and is, to the best of my knowledge, the first to involve such a large and diverse group of contributors from across the RoboCup leagues. It discusses research within RoboCup and the collaboration and synergies between the leagues.

Group photo of the humanoid league teams at RoboCup 2025.

The 2050 goal of the RoboCup Federation, as many people are probably aware, is for a team of humanoid robots to play against the winners of the FIFA World Cup. To achieve this, it is necessary to push further in that direction. One of the key changes, therefore, will be a stronger focus on humanoid robots.

Another major change will be the merger of the Standard Platform League (SPL) and the KidSize Humanoid League. This merged league will have the freedom to define its exact format and to develop a new roadmap that aligns the entire league toward a shared objective. While the 2050 goal itself remains unchanged, the path toward achieving it will need to be adjusted.

It is crucial to continue fostering the engagement of teams in the leagues that will be affected by these changes. At the same time, we must recognize that technology is advancing rapidly. Over the past year, in particular, we have seen significant progress in both hardware platforms and large language models. As RoboCup serves as a global benchmark for robotics research, we should continuously strive to advance technology and research—while still having fun.

Soccer is the complex task and behavior we are studying, and it is complex in many dimensions: from physical control and robot motion, to communication and strategy, and even human-like interactions. These include responding to the referee’s whistle, verbal and non-verbal communication among team members, interactions with the coach, and communication with the referee. All of these aspects will ultimately need to be incorporated into the humanoid league.

I believe we will see a mix of different robots. With the three companies currently sponsoring RoboCup, we have already seen that their robots can achieve a wide range of behaviors, and there have been significant improvements in robot control. Some of these robots can walk very quickly—almost to the point of running.

Initially, there may be the possibility of multiple teams using the same platform. However, we must keep in mind that both hardware and software can become obsolete very quickly, so we need to remain open to multiple options. A robot that is state of the art today may no longer be so in a year or two. As a result, committing to a single standard platform could limit future progress.

For this reason, the current idea is to remain open to multiple platforms. Many teams already have excellent custom-built robots, and further improvements to these platforms should be encouraged. That said, the exact structure has not yet been decided, and these decisions will be made in consultation with the teams. It is important to give the RoboCup community the time it needs to adapt and move forward.

One major change is that we have added extra robots to each team. Previously, teams played with just two robots per side, but matches are now played three versus three.

Another important improvement is the reduced presence of humans on the field. There is no longer a handler assigned to each robot. In the past, a team member had to walk behind the robots in case they fell and risked being damaged.

I have actually played in a match against the winning humanoid team. Naturally, the human team won, but it was an enjoyable and very interesting game, as the robots were surprisingly fast.

Yes, it depends on which league you are part of. The reactions have been a mix of excitement and passion. Of course, everyone is keen to see improvements, and participants have always been prepared for changes to the rules and the league structures. However, there are still some open questions, and teams are waiting to see how things will evolve. Tomorrow, there will be a meeting with the President and several trustees to address questions raised by the leagues.

The overall direction of RoboCup, guided by the 2050 goal has not changed. Each league has been extremely valuable and has contributed in different ways toward achieving that goal. RoboCup has also been immensely valuable for robotics research more broadly. Beyond being fun, the challenges involved in making robots play soccer are extraordinarily complex. The research and solutions developed within RoboCup can be applied to many other fields and applications.

Alessandra is Assistant Professor at the University of Naples Federico II, Italy. Her PhD thesis was part of the Marie Sklodowska-Curie Research ETN SECURE project at the University of Hertfordshire (UK). Her research interests include Human–(Multi) Robot Interaction, social robotics, trust, XAI, multi-agent systems and user profiling. She is Project Manager and co-supervisor of the MSCA PERSEO (955778), TRAIL (101072488) and SWEET (101168792). She is also co-PI of the project ERROR (FA8655-23-1-7060), and part of several national and international projects. Alessandra is also trustee member of RoboCup Federation, and member of the Humanoid League team called Bold Hearts. She is Chair of the IEEE P3108 “Study Design”, and a member of the “Appendix” groups, she is Program Chair of IEEE RO-MAN 2027, she has been Robotic Challenge Chair at ICSR 2025, Special Session Chair of IEEE RO-MAN 2024, Publicity Chair of IEEE RO-MAN 2022 and 2023, Organising Chair of the 26th RoboCup International Symposium 2023, and she is on the program committee of several international conferences on human–robot interaction and artificial intelligence.

Observe your own hand. As you read this, it’s holding your phone or clicking your mouse with seemingly effortless grace. With over 20 degrees of freedom, human hands possess extraordinary dexterity, which can grip a heavy hammer, rotate a screwdriver, or instantly adjust when something slips.

With a similar structure to human hands, dexterous robot hands offer great potential:

Universal adaptability: Handling various objects from delicate needles to basketballs, adapting to each unique challenge in real time.

Fine manipulation: Executing complex tasks like key rotation, scissor use, and surgical procedures that are impossible with simple grippers.

Skill transfer: Their similarity to human hands makes them ideal for learning from vast human demonstration data.

Despite this potential, most current robots still rely on simple “grippers” due to the difficulties of dexterous manipulation. The pliers-like grippers are capable only of repetitive tasks in structured environments. This “dexterity gap” severely limits robots’ role in our daily lives.

Among all manipulation skills, grasping stands as the most fundamental. It is the gateway through which many other capabilities emerge. Without reliable grasping, robots cannot pick up tools, manipulate objects, or perform complex tasks. Therefore, we focus on equipping dexterous robots with the capability to robustly grasp diverse objects in this work.

While humans can grasp almost any object with minimal conscious effort, the path to dexterous robotic grasping is fraught with fundamental challenges that have stymied researchers for decades:

High-dimensional control complexity. With 20+ degrees of freedom, dexterous hands present an astronomically large control space. Each finger’s movement affects the entire grasp, making it extremely difficult to determine optimal finger trajectories and force distributions in real-time. Which finger should move? How much force should be applied? How to adjust in real-time? These seemingly simple questions reveal the extraordinary complexity of dexterous grasping.

Generalization across diverse object shapes. Different objects demand fundamentally different grasp strategies. For example, spherical objects require enveloping grasps, while elongated objects need precision grips. The system must generalize across this vast diversity of shapes, sizes, and materials without explicit programming for each category.

Shape uncertainty under monocular vision. For practical deployment in daily life, robots must rely on single-camera systems—the most accessible and cost-effective sensing solution. Furthermore, we cannot assume prior knowledge of object meshes, CAD models, or detailed 3D information. This creates fundamental uncertainty: depth ambiguity, partial occlusions, and perspective distortions make it challenging to accurately perceive object geometry and plan appropriate grasps.

To address these fundamental challenges, we present RobustDexGrasp, a novel framework that tackles each challenge with targeted solutions:

Teacher-student curriculum for high-dimensional control. We trained our system through a two-stage reinforcement learning process: first, a “teacher” policy learns ideal grasping strategies with privileged information (full object shape and tactile sensors) through extensive exploration in simulation. Then, a “student” policy learns from the teacher using only real-world perception (single-view point cloud, noisy joint positions) and adapts to real-world disturbances.

Hand-centric “intuition” for shape generalization. Instead of capturing complete 3D shape features, our method creates a simple “mental map” that only answers one question: “Where are the surfaces relative to my fingers right now?” This intuitive approach ignores irrelevant details (like color or decorative patterns) and focuses only on what matters for the grasp. It’s the difference between memorizing every detail of a chair versus just knowing where to put your hands to lift it—one is efficient and adaptable, the other is unnecessarily complicated.

Multi-modal perception for uncertainty reduction. Instead of relying on vision alone, we combine the camera’s view with the hand’s “body awareness” (proprioception—knowing where its joints are) and reconstructed “touch sensation” to cross-check and verify what it’s seeing. It’s like how you might squint at something unclear, then reach out to touch it to be sure. This multi-sense approach allows the robot to handle tricky objects that would confuse vision-only systems—grasping a transparent glass becomes possible because the hand “knows” it’s there, even when the camera struggles to see it clearly.

Trained on just 35 simulated objects, our system demonstrates excellent real-world capabilities:

Generalization: It achieved a 94.6% success rate across a diverse test set of 512 real-world objects, including challenging items like thin boxes, heavy tools, transparent bottles, and soft toys.

Robustness: The robot could maintain a secure grip even when a significant external force (equivalent to a 250g weight) was applied to the grasped object, showing far greater resilience than previous state-of-the-art methods.

Adaptation: When objects were accidentally bumped or slipped from its grasp, the policy dynamically adjusted finger positions and forces in real-time to recover, showcasing a level of closed-loop control previously difficult to achieve.

RobustDexGrasp represents a crucial step toward closing the dexterity gap between humans and robots. By enabling robots to grasp nearly any object with human-like reliability, we’re unlocking new possibilities for robotic applications beyond grasping itself. We demonstrated how it can be seamlessly integrated with other AI modules to perform complex, long-horizon manipulation tasks:

Grasping in clutter: Using an object segmentation model to identify the target object, our method enables the hand to pick a specific item from a crowded pile despite interference from other objects.

Task-oriented grasping: With a vision language model as the high-level planner and our method providing the low-level grasping skill, the robot hand can execute grasps for specific tasks, such as cleaning up the table or playing chess with a human.

Dynamic interaction: Using an object tracking module, our method can successfully control the robot hand to grasp objects moving on a conveyor belt.

Looking ahead, we aim to overcome current limitations, such as handling very small objects (which requires a smaller, more anthropomorphic hand) and performing non-prehensile interactions like pushing. The journey to true robotic dexterity is ongoing, and we are excited to be part of it.

The Industrial League arena at RoboCup2025.

RoboCup is an international scientific initiative with the goal of advancing the state of the art of intelligent robots, AI and automation. The annual RoboCup event took place from 15-21 July in Salvador, Brazil. The Logistics League forms part of the Industrial League and is an application-driven league inspired by the industrial scenario of a smart factory. Ahead of the Brazil meeting, we spoke with three key members of the league to find out more. Alexander Ferrein is a RoboCup Trustee overseeing the Industrial League, and Till Hofmann and Wataru Uemura are Logistics League Executive Committee members.

Alexander Ferrein: The idea of the Logistics League is to have robots helping in intra-production logistics. The playing field is set up with different machines and the robots need to bring raw materials and products to the machines and pick products up from them. There are orders coming in for different products of different complexities. The idea is that the robots deliver these products after they have been machined, at a certain handover point, and then the team will be awarded points. The setup is that we have six machines per team, and three robots operating in the smart factory.

There are two teams competing at the same time on different sides of the field. Most of the machines are on the home side of the field, but some machines are also on the opponent’s side of the field. The teams need to show basic robotics skills like navigation skills and collision avoidance. For the production of products we have little discs that have different colours and can be stacked on top of each other, and they have different caps. We have around 550 different items that can be produced. We don’t focus so much on the handling of the parts, therefore the manipulating mechanisms are quite simple and are usually custom built by the teams. They just need to grab these discs and drive them around and put them on conveyor belts on the machines.

The main focus is on the production logistics and the planning phase. The complex products need to be produced or machined by a team of robots – without this planning and team work it wouldn’t be possible to deliver the products within the allotted time in the competition.

Until now, we had been supported by Festo didactics, who supplied the machines. However, they pulled out in February and told us that they won’t support us in bringing the machines to Brazil. Our team in Aachen has a complete field set up, so we are in the (not so easy) process of packing the machines up in pallets and shipping them to Brazil.

Till Hofmann: One important detail is that all the products that need to be manufactured, the orders for those come online, and the number of possible products is very high. Therefore, you can’t do any planning in advance – you can’t just create a big database that contains one sequence that you execute for every possible product, that doesn’t really work. So the robots need to do online planning. Due to the fact that we have multiple robots in the team and then also the opponents teams’ robots on the field, many things go differently than planned, so a very big aspect of the competition is execution monitoring and online replanning. Basically you create an initial plan, but you need to constantly adapt that plan to what actually happens during execution. I just want to stress that in contrast to other RoboCup competitions, it’s really a long-horizon planning task in the sense that we usually need to do actions on a time horizon of five to ten minutes to actually get to an intermediate goal of producing one of those products.

Wataru Uemura: The focus of our league is on how to handle the production line. The three mobile robots are a very important part. At first these were autonomous guided robots, but now they are completely autonomous mobile robots. The robots need to decide on their path to make the product.

Alexander: First of all, one needs to say that it is a really tough problem that we are facing here. So the teams that are starting new, they have to deal with all of the robotic aspects, so mobile robots, autonomous intelligence systems, they need to drive around, map, interact with the machines. As these items that we are pushing around are non-standard things, they also need to build their own manipulating devices. And then there is this massive planning aspect of the league, which is also not so easy. When we started this in 2011, 2012, we were thinking that this must be a solved problem, that one could use scheduling systems, that production is digital, and that everything would be easy. However, we found out it’s not actually easy, and there are no off-the-shelf solutions for a fleet of robots doing planning and production.

Looking at our team (which has become particularly successful over the years) and from observing the other teams as a trustee, I think that the integration aspect of all the different tasks is really a hard thing. Having a software system that is capable of so many things, communicating with the centralized referee box, and making it run within the time limit of a match, is the major challenge. Today, navigation of a robot is not the big issue, basically, but getting it integrated into your software system and building all the rest around the planning components and so on, this is, from my point of view, the major challenge.

Till: As I mentioned, I think the combination of long-horizon planning and execution monitoring is particularly difficult. We do have a lot of failures during one production run, because of hardware limitations and problems with the robots. Sometimes the machines themselves fail and they need to deal with this, without being able to solve the problem itself, because it’s not in their control. So they need to do a lot of reasoning that considers all the different cases that may happen. For example, suddenly you have a product appearing in a machine and you no longer know the configuration because this information was lost on the way. How do you deal with this? Another example is that the robot drops a piece while it is driving around and then tries to feed it into the machine. Then the machine reports a failure and the team needs to keep its world model up-to-date to know that this product is no longer where they thought it was. How do we deal with this?

Alexander: Because the problem we are trying to solve is so hard, we don’t have many teams in the league. We had a new team at RoboCup2023, in Bordeaux, and they are now consistently coming to the RoboCups, which is nice. Apart from this, we have a core of three to five teams that participate in the league. As the challenge itself is hard to get into, the skills only develop slowly, so new challenges are not really introduced. There are slight changes here and there. One of the major changes in recent years was during Coronavirus times where we had to abandon the whole match aspect because we couldn’t play with two teams at the same time. We introduced aspects of the game as challenges. Now we have a challenge track as well, where teams can just focus on certain aspects of the league and do not need to play the full game.

The skills of the teams are not developing in such a way that we really need to add new challenges because it is still unsolved and hard for all the teams to get the robots running, because we have so many different aspects.

Alexander: Yes, we have some ideas for a new Industrial League. As Festo are pulling out we don’t see the need to stick with this particular machine type. We had a workshop earlier this year with all the teams and organising committees and we’re proposing a future challenge, or league. We will discuss this at RoboCup2025.

Till: We actually already had discussions with @work, and agreed that the long-term goal is to merge the two competitions into one big Industrial League. Next year, we will start converging by doing some kind of collaboration challenge or crossover challenge where teams from @work will collaborate with what’s now the Logistics League, but will be the Smart Manufacturing League by that time. Then hopefully, in two or three years, this will be one big league rather than separate industrial competitions. And the idea of the league that we’re currently planning to migrate to is really a broader smart manufacturing scenario where we have different aspects of smart manufacturing. So currently it’s really only the production logistics part. But in the future, we also want to include the assembly itself as part of the competition and also extend this to humanoid robots and also focus on human-robot collaboration in this manufacturing setting.

So this will be very different to the Logistics League as it is right now. How we will do the migration from what we have to that new league without losing all the teams is something that’s still in the making.

Alexander: I also don’t think we shouldn’t restrict ourselves to just one type of robot. As we see, there is something going on with respect to humanoid robots, and the Rescue League is proposing a quadruped robot. In the @Home League there are ideas to introduce the staircases where you need more agility in these settings. So for me, I wouldn’t restrict the type of robots that we’re using. We are just thinking about proposing challenges that are looking towards the future. So far we have been doing things that we thought might be relevant to industry, but industry is not very interested in what we are doing here. At least, they are not knocking at our doors and asking what our solutions are. That’s also another aspect that we will possibly discuss at Robocup 2025 – how we could increase our impact as a league for the outside world.

Alexander: Yes, be relevant, right? I mean, you see so much going on, in particular in China with the robot manufacturing OEMs [original equipment manufacturers] that build these robots in a few years that have quite impressive capabilities. And well, we stand aside and just watch. So maybe we should use those robots, and integrate them into our process. It’s very important that we are opening our minds to envision a future that’s different from today.

Alexander Ferrein received his MSc in Computer Science (Dipl.-Inform.) and his PhD (Dr. rer. nat) from Aachen University in 2001 and 2007, respectively. Between 2009-2011 he joined the Robotics and Agents Research Lab, University of Cape Town, as a postdoctoral research fellow with Feodor-Lynen scholarship granted by the Alexander-von-Humboldt Foundation. He then re-joined the Knowledge-Based Systems Group at Aachen University before he became a professor for Robotics and Computer Science at FH Aachen University of Applied Sciences. He is a heading the Mobile Autonomous Systems & Cognitive Robotics Institute at Aachen Applied Science University. His research focusses on the field of Artificial Intelligence and Cognitive Robotics. Since 2015 he is member of the Advisory Committee of the African-German Network of Excellence in Science whose Vice-president he was between 2019-2023. His research concentrates on the field of cognitive robotics. In particular, he is interested in high-level control and decision making of robots and agents acting under real-time constraints.

Till Hofmann is a Postdoc at RWTH Aachen University. His research focuses on planning, plan execution, generalized planning as well as reactive synthesis, with a particular focus on planning for robotics. He was a participant in the RoboCup Logistics League from 2016 until 2019 and member of the technical committee from 2017 until 2020. Since 2024, he is on the executive committee of the RCLL.

Wataru Uemura was born in 1977, and received B.E, M.E. and D.E. degrees from Osaka City University, in 2000, 2002, and 2005. He is an associate professor in Electronics, Information and Communication Engineering Course, Faculty of Advanced Science and Technology, Ryukoku University in Shiga, Japan. He is a member of IEEE, RoboCup and others. He is a chairperson of the RoboCup Japanese National Committee. He is an executive committee member of RoboCup Logistics League. He was a member of the Industrial Robotics Competition Committee, the World Robot Summit. He was TPC Vice Chairs of GCCE 2012, Conference Chair of GCCE 2016, and Publication Chairs of GCCE (Global Conference on Consumer Electronics). He is a member of the World Skills in Japan organizing committee of Autonomous Mobile Robots.