Industry Briefing
A single destination for timely, editor-curated robotics news from around the world.
Florent Delgrange Wins AAMAS 2026 Blue Sky Award for Foundation World Models Research
Florent Delgrange received the Best Blue Sky Paper Award at AAMAS 2026 for his research on Foundation World Models for agents that adapt in dynamic environments. His work addresses the challenge of ensuring that autonomous agents continue to learn effectively while maintaining reliable behavior in changing conditions. This research is significant as it proposes a structured approach to agent learning that integrates reinforcement learning and formal methods. Delgrange emphasizes the need for agents to adapt to evolving environments while ensuring that their decision-making remains trustworthy, highlighting the importance of continuous reliability in real-world applications. Looking ahead, Delgrange envisions a foundation world model that is reusable across various tasks and adaptable to changing conditions. This model aims to enhance decision-making by providing agents with a comprehensive understanding of their environment, which is crucial for maintaining safety and performance in multi-agent systems. No further timeline was disclosed at the time of publication.
Former Huawei CTO Li Yin Establishes Startup Focused on World Models Technology
Li Yin, the former CTO of Huawei's Pangu model, has launched a startup named Xirang Kaiwu Technology, focusing on world models. The company, registered on July 1, 2026, in Shenzhen, has an initial capital of 1 million yuan and a valuation of 1.5 billion yuan following its recent funding round. Li holds a 55% stake, making him the largest shareholder and controlling person. The startup aims to develop world models that enable AI to learn and predict the physical world's temporal and spatial evolution, providing a simulation training foundation for embodied intelligence and autonomous driving. During his tenure at Huawei, Li highlighted the challenges of insufficient training data and fragmented task scenarios as significant barriers to the commercialization of embodied intelligence. Li's departure is part of a broader trend, with several key figures from Huawei's Pangu model team leaving to pursue their ventures. Notably, in March 2026, Wang Yunhe, another leader in the Pangu model, founded Yuanlilv Dong, focusing on AI agents. The world model sector has seen over 20 new startups registered in the first seven months of 2026, with Li being the sixth Huawei executive to enter the embodied intelligence field this year.
World Models Embodied Intelligence AI Autonomous Driving
NVIDIA Advances Physical AI with Open World Models and Omniverse Technologies
NVIDIA has joined over 200 organizations in signing an open letter advocating for AI leadership through open ecosystems. This approach emphasizes the importance of open models in physical AI, which require understanding and predicting environmental behaviors rather than just appearances. The significance of open world models lies in their ability to generate training data, simulate future states, and provide a foundation for specialized applications in robotics, autonomous vehicles, and vision AI. NVIDIA's Cosmos 3 integrates these capabilities, achieving leading benchmark results and widespread adoption across various sectors. Looking ahead, the focus will be on enhancing the data collection process for physical AI, particularly in rare events and long-tail scenarios. The need for diverse environments and adaptable models is crucial for improving the performance of specific robots and systems. No further timeline was disclosed at the time of publication.
Contactile Introduces Friction-Based Tactile Sensors for Enhanced Robot World Models
Contactile has developed tactile sensors for robotic hands and grippers that enhance world models in robotics. These sensors address the limitations of current conditioning methods, which primarily rely on visual inputs and joint encoders, by incorporating friction as a critical input. This advancement is essential for improving robots' ability to generalize across various surfaces and objects. The significance of this development lies in its potential to transform robot learning. By making friction a first-class input, robots can better predict the outcomes of their actions and adapt to novel tasks without extensive memorization. This shift could lead to more capable robotic systems that can operate effectively in contact-rich environments, where traditional conditioning methods fall short. Looking ahead, the integration of Contactile's tactile sensors into robotic systems could revolutionize how robots interact with their environments. As the field of robot learning continues to evolve, the emphasis on accurate conditioning will be crucial for deploying effective world models. No further timeline was disclosed at the time of publication.
Artificial Intelligence Artificial Intelligence / Cognition Assembly Design / Development End Effectors / Grippers Grippers
Gravity 4D Launches First Module to Enhance World Models for Robotics
On July 17, Extreme Intelligence unveiled Gravity 4D WAM, the first module of its embodied intelligence framework, at the 2026 World Artificial Intelligence Conference (WAIC). This launch addresses a critical issue in current visual language models: the discrepancy between visually plausible predictions and physical reality. Gravity 4D aims to transition world models from merely predicting visuals to accurately anticipating three-dimensional physical evolution. The significance of Gravity 4D lies in its ability to enhance robotic operations by ensuring that actions are based on physical realities rather than just visual appearances. The framework introduces a 4D latent model that allows the World Action Model (WAM) to learn essential information about RGB appearance, spatial structure, and motion dynamics. This paradigm shift is crucial for improving the reliability of robotic tasks, as it ensures that robots can effectively grasp objects and navigate environments based on true physical interactions. Looking ahead, Gravity 4D's approach could redefine how robots interact with their environments, moving from visual-based predictions to a deeper understanding of physical laws. The framework's dual-brain architecture and integration of various sensory inputs will be further detailed in upcoming technology releases. No further timeline was disclosed at the time of publication.
Robotics AI Machine Learning Automation
World Models at WAIC 2026: Bridging Understanding and Execution in Physical AI
The World Artificial Intelligence Conference (WAIC) 2026 highlighted the significance of world models as a crucial pathway for physical AI to transition from laboratory settings to real-world applications. The event featured a summit focused on the integration of world models and embodied intelligence, showcasing discussions led by Nobel laureates and industry leaders. This summit, organized by the WAIC committee and hosted by Daxiao Robotics, emphasized the need for machines to not only express but also understand and reliably act within the physical world. Key presentations included insights from Nobel laureate Sargent on the gap between AI expectations and human rationality, and a demonstration of the ACE technology stack by Daxiao Robotics' chairman, Wang Xiaogang, illustrating practical applications across various industries. The forum also introduced the PIQ platform, a unified evaluation benchmark for embodied physical intelligence, aimed at addressing industry challenges related to self-assessment and standardization. As industry representatives engaged in discussions about technical routes and implementation pathways, the event underscored the journey from theoretical frameworks to tangible industry applications, marking a pivotal moment in the evolution of physical AI.
World Models Physical AI AI Standards Industry Applications Technological Breakthroughs
Chinese Companies Explore World Models for AI Simulation of Environments
Artificial intelligence is evolving with a focus on 'world models,' which simulate environmental responses to actions. This shift is gaining traction among Chinese companies, expanding the application of these models beyond traditional physics and robotics. The technology is still developing, with no clear consensus on its final form, indicating a significant area of exploration for AI advancements. The significance of world models lies in their potential to enhance AI's predictive capabilities, allowing systems to anticipate changes in both physical and digital environments. This could lead to improved decision-making processes across various sectors, as companies leverage these models to better understand and interact with their surroundings. The growing interest from major tech firms highlights the competitive landscape surrounding this emerging technology. Looking ahead, the development of world models is expected to progress, although specific timelines for advancements or implementations remain undisclosed. As the industry continues to explore this frontier, stakeholders should monitor the evolution of standards and applications that will shape the future of AI simulation technologies.
BYD AI Team Unveils HyWorldVLA Hybrid Model Achieving 90.59 PDMS on NAVSIM Benchmark
The BYD AI team has introduced the HyWorldVLA, a hybrid pixel-latent world model that utilizes VLA architecture. This model has achieved a score of 90.59 PDMS on the NAVSIM benchmark, marking a significant milestone for BYD in the field of autonomous driving foundation models. This achievement is noteworthy as it highlights BYD's commitment to advancing autonomous driving technologies. The collaboration with researchers from HIT robotics underscores the importance of interdisciplinary efforts in developing state-of-the-art models that can enhance vehicle autonomy and safety. Looking ahead, the performance of the HyWorldVLA on the NAVSIM benchmark sets a high standard for future developments in autonomous driving models. No further timeline was disclosed at the time of publication.
Technology
Ant LingBot Unveils Six Open-Source AI Models Amid Data Challenges
Ant LingBot, a subsidiary of Ant Group, has launched six open-source embodied AI models as part of its dual-track strategy focusing on Visual Language Agents (VLA) and world models. This initiative aims to enhance AI capabilities while addressing the growing demand for advanced AI solutions. The significance of this release lies in Ant LingBot's commitment to fostering an open-source ecosystem, which is crucial for collaboration and innovation in the AI field. However, the company is contending with challenges related to data scarcity and competition within the ecosystem, which could impact its development and deployment efforts. Looking ahead, it will be important to monitor how Ant LingBot navigates these challenges and whether it can successfully leverage its dual-track strategy to establish a strong presence in the AI landscape. No further timeline was disclosed at the time of publication.
Technology
Moonshot AI Launches Kimi K3, the World's First Open 3T-Class AI Model
Moonshot AI has introduced Kimi K3, a groundbreaking 2.8 trillion-parameter open-source AI model, marking it as the largest of its kind to date. This model is designed to handle complex workflows and features a one-million-token context window, enabling it to perform tasks significantly faster than traditional methods. For instance, Kimi K3 can complete a task in about two hours that would typically take one to two weeks for an experienced researcher. The significance of Kimi K3 lies in its potential to enhance scientific research workflows, allowing for the creation of interactive reports and presentations. It incorporates advanced features such as Widgets and Dashboard capabilities, which facilitate persistent, interactive workspaces. Despite its impressive performance, Kimi K3 still falls short compared to proprietary models like Claude Fable 5 and GPT 5.6 Sol, indicating that while progress is being made, there is still a competitive gap to close. Looking ahead, the focus will be on how Kimi K3 can further evolve and compete with leading AI models. Moonshot AI's innovative architecture, including Kimi Delta Attention and a Mixture-of-Experts framework, has improved scaling efficiency significantly. No further timeline was disclosed at the time of publication.
AI and Robotics Innovation
WAIC 2026 Highlights Six Key AI Trends Shifting Focus from Models to Systems
The World AI Conference 2026 (WAIC 2026) showcased six significant trends in artificial intelligence, emphasizing a shift from model competition to system efficiency. Notably, the integration of robots into real factory environments marks a pivotal moment for industrial automation, indicating a growing reliance on intelligent systems. This transition is crucial as it reflects the industry's evolution towards more efficient and effective AI systems, moving beyond traditional model-centric approaches. The emergence of domestic chips reaching a tipping point further underscores the importance of localized technology development in enhancing AI capabilities. Looking ahead, stakeholders should monitor how these trends will influence the deployment of AI systems in various sectors. The increasing presence of robots in factories and advancements in chip technology could reshape operational strategies and drive innovation in the coming years. No further timeline was disclosed at the time of publication.
News
PHANES AI Launches TouchWorld Tactile Model for Enhanced Robot Dexterity
PHANES AI, established by 28-year-old Yang Shuo from HIT, has introduced TouchWorld, a tactile foundation model designed to enhance robotic manipulation capabilities. This model enables robots to perform precise physical tasks by incorporating a sense of touch, marking a significant advancement in robotic dexterity and interaction with their environment. The introduction of TouchWorld is significant as it allows robots to predict and react to tactile stimuli, which is crucial for applications requiring fine motor skills. This development could lead to improved performance in various sectors, including manufacturing and healthcare, where dexterous manipulation is essential for tasks such as assembly or surgical procedures. Looking ahead, the impact of TouchWorld on the robotics industry will be closely monitored, particularly regarding its adoption in real-world applications. No further timeline was disclosed at the time of publication, but the potential for this technology to transform robotic capabilities is substantial.
Technology
ByteDance Prepares AI Model for Real-Time Spatial Video Generation Competing with Meta and Alphabet
ByteDance Ltd. is developing an AI model focused on real-time spatial video generation, positioning itself against major players like Meta Platforms Inc. and Alphabet Inc. This initiative highlights the growing competition in the AI sector, particularly in applications relevant to robotics and autonomous systems. The significance of ByteDance's efforts lies in its potential to enhance capabilities in robotics and autonomous systems, areas that are increasingly reliant on advanced AI technologies. By entering this competitive landscape, ByteDance aims to carve out a niche in a market that is rapidly evolving and attracting significant attention from industry leaders. Looking ahead, stakeholders should monitor ByteDance's progress in this AI model development, as it could influence trends in spatial video applications and their integration into robotics. No further timeline was disclosed at the time of publication.
RoboColiseum Unveils Evaluation Platform for Assessing Embodied Intelligence Models
RoboColiseum has launched an evaluation platform for embodied intelligence models, aimed at providing a multidimensional assessment system. This platform, set to be fully operational by August 2026, allows developers to identify strengths and weaknesses in their models through simulations that closely mimic real-world performance. The significance of this platform lies in its ability to bridge the gap between simulated and real-world environments, with a reported correlation of 89.5% between simulation assessments and actual deployments. This high fidelity in simulation helps developers quickly filter models and pinpoint issues before moving to more costly and time-consuming real-world testing. Looking ahead, RoboColiseum plans to continuously update its tasks and assessment dimensions based on industry needs. The platform currently features four capability subcategories and 78 high-fidelity simulation tasks, focusing on instruction following, spatial reasoning, robustness, and general manipulation, ensuring comprehensive evaluation of model performance across various scenarios.
Embodied Intelligence Simulation Evaluation AI Development Robotics Model Assessment
StarSeaTech Advances Embodied Intelligence at World Robotics Conference 2026
During the 2026 World Robotics Conference, StarSeaTech hosted a forum titled 'From Models to Productivity' at the Beiren Yichuang Exhibition Center. The event gathered experts from academia, developers, and corporate executives to discuss innovations in embodied intelligence models, machine evolution, commercialization, and technological boundaries, showcasing the latest thoughts on technology iteration and industrial application. StarSeaTech's Chief Scientist Zhao Xing highlighted the evolution of embodied intelligence technology through instinctive, operational, and evolutionary intelligence levels. The industry is transitioning from isolated technological breakthroughs to systemic competition involving models, data, entities, infrastructure, and applications. StarSeaTech has established a comprehensive productivity flywheel for embodied intelligence, focusing on real machine data, model iteration, and physical task execution supported by infrastructure for continuous optimization. Key announcements included the upgrade of the G0.5 foundational model, the introduction of the Fast-WAM world model pre-training plan, and the launch of the G-Fleet distributed reinforcement learning system. These developments aim to enhance model capabilities, speed up inference times, and support continuous learning in real robot clusters. No further timeline was disclosed at the time of publication.
Embodied Intelligence Robotics AI Technology Model Innovation Industrial Automation
Qingche Intelligent Showcases Comprehensive Data Systems and Robotics at WAIC 2026
From July 17 to 20, 2026, Qingche Intelligent participated in the World Artificial Intelligence Conference (WAIC) in Shanghai, focusing on bridging models and real-world intelligence. The company showcased a complete technology system from real-world data collection to model training and robotic applications, emphasizing the importance of real-world data for embodied intelligence. Qingche's RoboPocket system, a no-body robot data collection platform, allows users to gather data without direct interaction with robots, significantly lowering the barriers for data acquisition. The DM3 data management platform complements this by enabling visual retrieval, quality control, and data asset management, handling up to 10,000 data entries daily and accumulating substantial real-world data assets. The company also presented its new generation of embodied intelligence pre-training models, which do not rely on teleoperation data, demonstrating a viable path for model training using only field-collected data. Qingche's solutions have already been deployed in real-world scenarios, such as retail pharmacies and hotel laundry services, showcasing the practical applications of their technology and the continuous evolution of their data-driven model development.
Embodied Intelligence Data Management Robotics AI Applications
Thirteen New Embodied AI Models Released in June 2026 by BAAI and Alibaba
In June 2026, the landscape of embodied AI saw the introduction of 13 new models, including significant contributions from BAAI and Alibaba's Qwen-Robot. This rapid development indicates a shift from traditional hardware benchmarks to a focus on software intelligence, highlighting the competitive nature of the field. The emergence of these models underscores the growing importance of software capabilities in embodied AI, as companies strive to enhance their offerings and differentiate themselves in a crowded market. This trend reflects a broader industry movement towards prioritizing intelligent software solutions over hardware specifications. Looking ahead, industry observers should monitor the ongoing advancements in embodied AI models, as the pace of innovation suggests that new releases may continue to emerge frequently. No further timeline was disclosed at the time of publication.
Technology
Small-AI Models Gain Traction Around the World
One morning in 2019, Adebayo Alonge was in a Cape Town hotel room, preparing to demonstrate his startup’s AI answer to a serious problem in African health care: counterfeit medication, which kills thousands of people across the continent every year.The RxScanner is a handheld spectrometer that scans a pill with infrared light, then sends the item’s molecular profile to an AI model equipped with a pharmaceutical database. In seconds, the AI identifies the medication from its molecular profile—or reports that it’s phony.Pharmacies were using the system in more than a dozen countries, including Ghana, Kenya, Myanmar, and Alonge’s native Nigeria. But that morning in South Africa, it didn’t work. “I was shocked,” Alonge says.The spectrometer connected to the AI model—but the data center was 14,000 kilometers away and bandwidth was limited. “Our server was in the United States, and just to get the result of a single scan was taking me over 5 minutes.”So Alonge immediately asked his engineers to shrink the AI model down to a smaller, low-power, unconnected version that could run entirely on his Android phone. They produced it 2 hours later, and that saved the demo.More importantly, the work birthed a new version of his device, which can authenticate a pill in places without broadband, computers, or even reliable electricity. It also turned Alonge into an advocate for this kind of “small AI.”Small AI for Global Health Care AccessSmall AI is a far cry from wealthy nations’ colossal large language models (LLMs), hyperscale data centers, multibillion-dollar investments, and debates about AI consciousness. But for millions of people around the world, the only AI that matters, and often the only kind available, is small. (According to a World Bank Report issued in November, only 0.7 percent of internet users in the world’s poorest countries have used ChatGPT, compared to a quarter of all internet users in the most developed nations.)“Most people are discussing AI from the LLM/generative side. But that needs a lot of computing power, electricity, massive data, and skilled people to manage it,” Ajay Banga, president of the World Bank, said last January at the World Economic Forum, in Davos. “Outside the developed world, other than maybe India and China, very few countries have that combination.”By contrast, small AI can deliver useful, even life-saving services to people in areas that have none of those things, Banga said. In India, where the government’s AI plans call for more development of small AI, many such systems are working for farmers.For example, a drone-based system developed by Bala Murugan and colleagues at the Vellore Institute of Technology, in India, takes photos of cashew plants and quickly identifies those with splotches that indicate disease. All the processing takes place on the drone itself, so there’s no need for a computer on-site, nor for a connection to a central server.Using small language models trained for a specific problem, and sometimes running on cheap, low-power devices, other small-AI implementations have been developed to identify ant infestations in a Uruguayan vineyard, detect the presence of malaria-carrying mosquitoes in a number of nations, and run electrocardiograms from an Arduino device in parts of Brazil that lack access to more complex equipment.“This is the most important area in AI nowadays,” says Marcelo José Rovai, a professor at the Institute of Engineering and Information Systems at the Federal University of Itajubá, in Brazil, who was involved in all three projects. “It’s growing very fast.”Low-Power, Small-AI Models on Devices Small AI models can run on a variety of low-power devices, including [from left to right] an Arduino Nano 33 BLE Sense, a Seeed Wio Terminal, and an Arduino Portenta.Moez AltayebFor Alonge, Rovai, and other advocates, small AI is not just “a promising trend,” as that November World Bank report calls it. It may be, in the long term, the form of AI that will touch the most lives and remain sustainable after some of the giant models become too costly for most users.“I think the future of AI is not like one giant model, at a center. I think it’s millions of small, precise models deployed at the edge, each one solving like a specific problem, a specific context,” Alonge says. This is partly because much of humanity—including people in parts of rich countries as well as the developing world—lives without access to cutting-edge frontier models. But, he says, it’s also because those models are not sustainable.“If someone is not subsidizing it, most people will not be able to afford those models. So those of us who are said to be small-AI developers are the ones who will have to build for the majority of the world,” Alonge says.There is no strict definition of “small AI,” but people often use the term for language models with at most a few billion parameters. (Compare that to cutting-edge models, which can include more than a trillion.) That’s small enough to run directly on a phone or a Raspberry Pi. That’s what allows these applications to run on devices without a connection to a data center and use only a few watts of power, often supplied by a battery or a solar panel.Despite their small footprint, these models aren’t fundamentally different technology from that of gigantic AI models, Rovai says. Many instances of small language models were created the same way the phone-based version of Alonge’s pharmaceuticals scanner was—by “pruning” large models, or removing the parameters that weren’t involved in the task. The result is a system that’s less capable generally but still very good at the specific job it was pruned for, Rovai says. A lighter version of RxAll’s RxScanner spectrometer sends its results to an AI model run locally on a phone to check that a drug’s molecular signature is genuine.RxAllOther small models are created by “distillation.” They are trained to mimic a large model, until their performance approaches that of their “teacher,” Rovai says. In other cases, a larger model’s precision is reduced, for example, so that a model run on 32-bit architecture can run on 8-bit designs. In situations where the machine learning application is being used to classify data or predict patterns (like an ant infestation), it’s trained from the beginning on a small device, not derived from a larger model at all. Running all these small, specialized systems is becoming easier, Rovai says, for two reasons.The first reason is that hardware is getting better and more capable while using less power, he says. This means more and more phones can run small AI—especially those equipped with neural processing units, which are specialized chips that handle AI tasks like facial recognition and changing the brightness, shadows, or contrast in a photo.In 2025, slightly more than a third of all smartphones shipped worldwide were capable of running generative AI, and that figure will reach 45 percent by the end of this year, according to the technology research firm Counterpoint. By the end of next year, slightly more than half of all smartphones will be able to run a small AI model.The second reason Rovai cites is the shrinking footprint of language models. Both Google DeepMind’s Gemma 4 (released in April) and Alibaba’s Qwen 3.5 are “fantastic” for small AI, Rovai says. Both models are “open weight,” meaning users can adjust the connections between parameters to suit their needs. This makes it easy, for example, “to take a lot of data from, say, the milk industry and retrain the model specifically on that,” Rovai says.Rovai illustrated these reasons on a Zoom call, using one of his most recent experiments. Holding up a device, he says, “This is the new Arduino UNO Q—a US $50 device with a Qualcomm chipset. I’m running a language model here, which collects data from sensors and analyzes that data to detect tiny pools of water where mosquitoes might be breeding. It takes 3 watts to run it.”Support for Small-AI DevelopmentConvinced that millions of people are already benefiting from these kinds of applications, the World Bank now actively promotes small AI with grants, mentorship programs, financing, technical advice, and models of government policies that are friendly for small-AI development. For example, in Rwanda, the World Bank is backing a government program to help low-income households get devices that can run AI.All that said, no one claims that large language models are going away entirely. To create a generative AI that can run on a phone or other small device requires the architectural insights, data processing, and results of a larger model, Rovai says. “We need the big models to create these smaller models.” And for all that small AI can benefit people without access to big AI, the technology can’t solve the larger problems of development and digital inequality, Alonge says. Implementing small AI won’t allow nations to escape the challenge of creating an ecosystem to support AI: reliable power, a supply chain that works, and an educational system that develops the talents needed to create AI tools.Though his drug-scanning system can run for days on a phone with no connection, “you still want to be able to enable periodic syncing for updates with new signatures for the medications and analytics,” Alonge says. “And even when you are using batteries, reliable power is important. That phone battery is not going to last forever.”In many parts of the world, the future of small AI isn’t assured, he says. “It works, and many places will eventually need to use it. The question is whether or not the political actors are wise enough to invest in infrastructure to support it long term.”
Small-language-models Artificial-intelligence Llms
From WorldArena Champion to 1500+ Models: Kuawei Intelligence Proves World Models are Business, Not Just Demos
Kuawei Intelligence has secured 1 billion RMB in a Series B funding round, elevating its post-investment valuation to over 10 billion RMB. This significant financial milestone underscores the company's rapid growth and innovation in the realm of physical AI and world models. Established as a unicorn, Kuawei is now poised for an initial public offering (IPO). Their recent triumph in the WorldArena competition further cements their status as a leader in world modeling and robotic training, showcasing their advanced technological capabilities on a global stage.
Physical AI World Models Robotics AI Technology
Large World Models & Software-Defined Automation: A Schneider Exec's Look at the Future
In a recent industry discussion, experts highlighted a significant challenge facing businesses: the issue of vendor lock-in. This problem, which restricts companies to a single supplier, limits their flexibility and innovation potential. The conversation took place during a technology conference held in San Francisco on October 15, 2023, where industry leaders gathered to address current trends and obstacles in the market. Participants emphasized that reliance on a single vendor can hinder competition and stifle creativity, as companies may feel compelled to continue using a service or product that does not fully meet their evolving needs. The motivation behind this concern stems from a desire for greater adaptability and the ability to leverage multiple solutions to enhance operational efficiency. To combat vendor lock-in, experts suggested strategies such as adopting open standards and promoting interoperability among different systems. By encouraging a more collaborative environment, businesses can mitigate risks associated with being tied to one provider and foster a more dynamic marketplace. The discussions underscored the importance of addressing these challenges to ensure that companies can thrive in an increasingly competitive landscape.
Factory / Control
What Exactly is Being Modeled by World Models?
Recent discussions in the field of embodied intelligence have brought to light the concept of 'world models,' revealing significant confusion regarding its definition and the diverse methodologies being employed across the industry. Experts are examining the limitations of existing modeling techniques and the challenges posed by data quality, underscoring the necessity of analyzing failures within training data. The discourse emphasizes that the size of parameters alone does not guarantee success in developing effective world models. This exploration is crucial as the industry seeks to enhance the understanding and application of embodied intelligence, paving the way for more robust and reliable systems.
World Models Embodied Intelligence Robotics AI Data Challenges
Wujie Power completes over $200 million in angel round financing, accelerating the development of embodied general intelligence and world models.
Wujie Power has successfully secured over $200 million in angel round financing, a significant boost aimed at advancing its research and development in embodied general intelligence and world models. This funding round, completed recently, underscores the growing interest and investment in artificial intelligence technologies. The financial support will enable Wujie Power to enhance its capabilities and accelerate its projects, positioning the company at the forefront of innovation in the AI sector. As the demand for sophisticated AI solutions continues to rise, this investment is expected to play a crucial role in the company's efforts to develop cutting-edge technologies that could reshape various industries.
Robotics Automation AI
How World Models and VLA Can Be Implemented: Insights from Top Experts in Embodied Intelligence
At the 2026 Zhangjiang Embodied Intelligence Supply Chain Conference, a roundtable discussion brought together leading experts in robotics to explore the critical role of world models in embodied intelligence. The event highlighted various industry challenges, particularly the necessity for robust data infrastructure and the integration of visual-language-action models with world models. Experts emphasized that high-quality data and innovative technological solutions are essential for advancing the field. The conference served as a platform for addressing these pressing issues, aiming to foster collaboration and drive progress in robotics and artificial intelligence.
Embodied Intelligence World Models Robotics Data Infrastructure AI Integration
Tsinghua Ecosystem Sets Its Sights on World Models as the Next AI Frontier
Tsinghua University-affiliated companies, including Zhipu AI, Shengshu Tech, and Momenta, are advancing their research and development efforts in world models, focusing on applications in video processing, robotics, and autonomous driving. These initiatives are part of a broader push to enhance artificial intelligence capabilities and improve the efficiency and effectiveness of automated systems. The companies aim to leverage cutting-edge technology to address real-world challenges and contribute to the rapidly evolving landscape of AI. With a commitment to innovation, these firms are positioning themselves at the forefront of the AI revolution, seeking to establish a competitive edge in the global market.
AI
Understanding World Models: Diverging Paths of Fei-Fei Li and Yang Likun
Fei-Fei Li and Yang Likun are at the forefront of artificial intelligence research, each adopting unique methodologies in the development of 'world models.' Li is concentrating on the creation of editable 3D environments aimed at practical applications, which could enhance user interaction and real-world utility. In contrast, Likun is focusing on internal simulations designed to improve predictive capabilities in autonomous systems, a crucial aspect for advancing AI reliability and functionality. Their differing approaches underscore the complexities and challenges inherent in AI problem-solving. By exploring these methodologies, both researchers contribute to a deeper understanding of how to effectively define and tackle issues within the field. This ongoing discourse reflects the broader landscape of AI development, where diverse strategies are essential for innovation and progress.
World Models 3D Environments Autonomous Systems AI Research
Imagining Consequences Before Robot Actions: The Next Intersection of Xingyuan's ω-EVA and Embodied World Models
At the 8th Beijing Zhiyuan Conference, Xingyuan unveiled its innovative ω-EVA model, marking a significant advancement in the field of embodied intelligence. This model represents a shift from traditional world models, which have typically acted as passive observers, to a more dynamic role in robotic decision-making. By integrating real-time feedback into action generation, the ω-EVA model emphasizes the necessity of predicting outcomes prior to executing movements. This development highlights a broader industry trend towards the practical application of artificial intelligence capabilities, showcasing how robotics can evolve to become more responsive and effective in various tasks.
Embodied Intelligence Robotic Decision-Making AI Models Real-Time Feedback Technology Innovation
Alibaba eyes physical world with its first suite of AI models for robots
Alibaba Group Holding has unveiled its inaugural suite of artificial intelligence models designed for robots, positioning itself in the competitive landscape of advancing AI beyond traditional chatbot applications. On Tuesday, the Hangzhou-based technology leader introduced the Qwen Robot Suite, a significant step into the realm of "embodied AI," which enables machines to perceive, reason, and engage with their physical surroundings. This innovative suite has been developed by Alibaba's AI research division, Tongyi Lab, and is currently undergoing pilot testing with select partners within the company. This move reflects Alibaba's commitment to expanding the capabilities of AI in real-world applications, aiming to enhance the interaction between machines and their environments.
AI’s next frontier, world models, and why China is ahead of the pack
In the rapidly evolving field of artificial intelligence, world models that simulate physical environments are gaining attention as the next frontier, surpassing traditional large language models. Recent developments indicate that China is leading the way in this area, outpacing the United States in the deployment of these advanced systems. These world models, which comprehend the physical laws governing the universe, are already being utilized to enhance AI applications, including robotics and autonomous vehicles. As of October 2023, China has integrated these technologies more extensively than its American counterparts, marking a significant advancement in the global AI landscape. This trend highlights the growing competition between the two nations in harnessing AI's potential for practical applications.
Interview with Wang Zhongyuan: VLA will survive, but world models are the future.
In recent months, the concept of "World Model" has gained significant traction within the AI and robotics sectors, driven by underlying industry anxieties. As AI technology has rapidly evolved over the past two years, limitations in embodied intelligence have become apparent, revealing that while robots can recognize objects, they struggle to understand physical interactions and causal relationships. The World Model aims to bridge this gap by enabling robots to learn the laws of the physical world. At the forefront of this exploration is Wang Zhongyuan, the director of the Beijing Academy of Artificial Intelligence, who identifies four distinct paths in the development of World Models. These include language-centered models, pixel-centered models, 3D structure-centered models, and visual representation-centered models. The Beijing Academy is pioneering a fifth approach that integrates language and visual data into a unified latent space representation, allowing for more complex interactions and predictions. Wang emphasizes that the World Model's potential lies in its ability to enhance embodied intelligence, enabling robots to understand and predict physical interactions over time. He envisions a future where World Models serve as the foundational brain for robots, capable of complex reasoning and decision-making in real-world scenarios. However, he cautions that achieving this goal will require significant advancements in data collection and model training, with a timeline of three to five years anticipated for substantial progress. As the field continues to evolve, the competition will focus on the ability to create models that accurately reflect the complexities of the physical world.
AGIBOT holds World Challenge 2026 to see how AI models perform on real tasks
AGIBOT has announced the launch of the World Challenge 2026, an initiative aimed at evaluating the performance of artificial intelligence models through closed-loop testing on actual robots engaged in real-world tasks. This event marks a significant shift in the robotics industry, moving away from traditional simulation scores to a more practical assessment of AI capabilities. The challenge is set to take place in 2026, providing a platform for developers and researchers to showcase their advancements in AI technology. By focusing on real tasks, AGIBOT aims to enhance the reliability and effectiveness of AI applications in robotics, ultimately driving innovation and improving performance in various sectors.
Artificial Intelligence Artificial Intelligence / Cognition Design / Development Development Tools / SDKs / Libraries Humanoids Mobility / Navigation
ICRA Highlights World Models: New Opportunities for Intelligent Space Solutions
At the ICRA 2026 conference, experts gathered to discuss the evolving role of world models in robotics, focusing on the necessity for advanced spatial perception hardware. The event, which took place recently, underscored the challenges the robotics industry faces as robots increasingly operate in varied environments. Key issues highlighted included the importance of data quality and perception systems, which are essential for developing autonomous capabilities. This shift towards a deeper understanding of the physical world marks a significant advancement in robotic technology and its applications, signaling a transformative period for the industry.
World Models Spatial Perception Robotics AI Data Quality
Chinese Company Kuawei Intelligence Tops WorldArena Global Benchmark in Embodied World Models
Kuawei Intelligence, a leading Chinese company in embodied artificial intelligence, has secured the top position in the WorldArena Track 2 (Data Engine) global benchmark for May 2026. This accomplishment places Kuawei ahead of notable international competitors such as WoW and BLM. The ranking not only highlights Kuawei's advancements in embodied AI but also signifies a pivotal moment for China's presence in the realm of world model research, reflecting the nation's increasing competitiveness in this cutting-edge technology sector.
EmbodiedAI
ZhiYuan Releases First Open-Source Dataset for World Models Focused on Rich Interaction
On June 3, 2026, ZhiYuan unveiled the second phase of the AGIBOT WORLD 2026 dataset, which centers on the theme of 'Rich Interaction.' This innovative open-source dataset is pioneering in its focus on physical interactions, meticulously documenting both successful and unsuccessful scenarios between robots and their environments. By offering a comprehensive range of data, the initiative seeks to improve world model training, thereby advancing the capabilities of robotic understanding and physical intelligence. This development marks a significant step forward in the field of robotics, as it aims to better equip machines to navigate complex real-world situations.
World Models Robotic Interaction Physical Intelligence Open-Source Datasets
LiberAI: Redefining World Models in AI
LiberAI, a company founded by Tsinghua University alumnus Liu Songming, is making strides in artificial intelligence by developing an innovative physical world model. This initiative aims to enhance AI's predictive capabilities and decision-making processes by fostering a deeper understanding of the physical environment, rather than relying solely on imitation. Recently, LiberAI has attracted significant investment from leading investors, which will support its mission to advance AI's ability to engage in causal reasoning. The company's efforts are positioned to transform how AI interacts with the world, marking a pivotal shift in the technology's evolution.
World Models AI Development Causal Reasoning Machine Learning
Startup Revolutionizes Restaurant Kitchens with World Models and AI Technology
Yuanjie Intelligent, a startup founded in March 2026 and led by former Meituan executive Dr. Wang Dong, has successfully raised millions in seed funding within just two months of its inception. The company aims to tackle the pressing labor shortages in the restaurant industry by developing intelligent kitchen robots designed to enhance operational efficiency and reduce costs. Utilizing advanced world action models, these robots are engineered to navigate the complexities of kitchen environments effectively. By addressing critical pain points in food delivery operations, Yuanjie Intelligent is positioning itself to revolutionize the culinary sector, promising a smarter and more automated future for restaurants.
Restaurant Automation AI Technology Robotics Food Delivery Solutions
Jubrain Stone Secures New Funding Round to Advance Cognitive World Models in Embodied Intelligence
Jubrain Stone has successfully secured a substantial funding round, spearheaded by leading investors in the industry, to advance its development of cognitive world models aimed at enhancing embodied intelligence. This funding will be directed towards bolstering core technology research, expanding the team, and increasing global market outreach. The initiative seeks to address current challenges in robotic learning and adaptability within real-world environments, positioning Jubrain Stone at the forefront of innovation in the field.
Cognitive World Models Embodied Intelligence AI Research Robotics Machine Learning
Former Kepler CEO Hu Debo Launches 'Sota Unbounded' to Achieve Scaling Law with World Models
Hu Debo, the former CEO of Kepler and a prominent figure at Huawei, has launched a new venture called Sota Unbounded. This company is focused on creating an advanced brain system designed to enhance robots' ability to comprehend and engage with the physical world. By employing innovative data collection and modeling techniques, Sota Unbounded aims to redefine the concept of embodied intelligence. The initiative seeks to address the needs of various industries, positioning itself as a leader in the development of intelligent robotic solutions.
Embodied Intelligence Robotics AI Data Collection Automation
AI Company Aims to Be the 'Brain Supplier' for Embodied Intelligence with World Models and Hardware Adaptation
An innovative AI company is making strides in the development of advanced world models and adaptable systems aimed at enhancing embodied intelligence. Rather than concentrating solely on physical robotics, the company prioritizes the capabilities of its models, reflecting a shift in focus within the industry. Founded by a team of seasoned professionals with extensive experience in artificial intelligence, the company has effectively implemented its solutions in a range of real-world applications. This approach not only showcases the versatility of their technology but also highlights the growing recognition of the significance of model performance in the evolving landscape of AI.
Embodied Intelligence AI Models Robotics Industrial Automation
Comprehensive Survey on World Models for Robot Learning Published by NTU, Berkeley, Stanford, and ETH
A recent collaborative study conducted by prominent research institutions examines the advancement of world models in robotics, highlighting their significance in allowing robots to forecast and simulate actions prior to execution. The paper reviews different paradigms for merging world models with robotic strategies, illustrating how these models serve a dual purpose as both predictive tools and learning environments. This exploration is crucial for enhancing the capabilities of robots, enabling them to operate more effectively in complex scenarios. The findings contribute to the ongoing discourse on improving robotic intelligence and adaptability, paving the way for more sophisticated applications in various fields.
Robot Learning World Models Machine Learning Robotics AI
Gradient-based planning for world models at longer horizons
A team of researchers, including Mike Rabbat, Aditi Krishnapriyan, Yann LeCun, and Amir Bar, has introduced GRASP, a new gradient-based planning method designed for learned dynamics in world models. This innovative approach addresses the challenges of long-horizon planning, which has proven to be fragile and inefficient with existing models. GRASP enhances planning by lifting trajectories into virtual states, allowing for parallel optimization across time, and incorporating stochastic elements to facilitate exploration. The development of GRASP comes in response to the limitations of current world models, which, despite their ability to predict complex sequences in high-dimensional spaces, struggle with optimization and can easily fall into local minima. The researchers emphasize that while powerful predictive models exist, effective control and planning remain significant hurdles. By utilizing a collocation-based approach, GRASP optimizes both actions and states, improving computational efficiency and robustness against adversarial vulnerabilities inherent in state gradients. The method also introduces exploration through Gaussian noise in state updates, enhancing the ability to navigate complex planning landscapes. Preliminary results indicate that GRASP significantly outperforms traditional methods in success rates and time efficiency for long-horizon planning tasks. The researchers view GRASP as a foundational step towards more advanced world model planners, with future work aimed at integrating the method into reinforcement learning systems and exploring diffusion-based world models. The full details of the study can be found in their published paper.
World's Best! Zhongke Fifth Epoch Tops WorldArena Rankings, Redefining Embodied World Models
Zhongke Fifth Epoch has secured the highest ranking in the WorldArena assessments, marking a significant milestone in the field of embodied world models. This achievement, announced recently, underscores the company's commitment to innovation and excellence in technology and product development. The recognition not only reflects Zhongke Fifth Epoch's advancements but also sets a new benchmark for the industry, showcasing the potential of their cutting-edge solutions.
Embodied AI World Models Technology Innovation Artificial Intelligence
NVIDIA Launches Ising, the World’s First Open AI Models to Accelerate the Path to Useful Quantum Computers
NVIDIA has unveiled the world's first family of open-source quantum AI models, known as NVIDIA Ising, aimed at empowering researchers and enterprises to develop quantum processors that can effectively execute practical applications. This announcement, made today, marks a significant advancement in the field of quantum computing, as it provides accessible tools for innovation and exploration in quantum technology. By fostering collaboration and knowledge sharing, NVIDIA hopes to accelerate the development of quantum applications, addressing the growing demand for powerful computing solutions. The initiative reflects the company's commitment to advancing AI and quantum research, positioning itself at the forefront of this emerging field.
The Death of the Label: Generalist AI Rejects 'World Models' in Favor of First-Class Physical Foundation
Pete Florence, CEO of Generalist AI, has expressed his views on the evolving terminology within the artificial intelligence sector, specifically criticizing terms such as 'VLA' and 'World Model' as mere temporary solutions. During a recent discussion, he emphasized that the architecture of GEN-1, which boasts a 99% scratch-trained framework, represents a strategic investment in the future reliance on purely robotic data. Florence's insights reflect a broader industry trend towards embracing more advanced and foundational approaches to AI development, suggesting a shift away from conventional terminologies as the field matures. This commentary comes as the AI landscape continues to evolve rapidly, with companies seeking to establish more robust and effective models for the future.
US GEN-1 World-Models Generalist AI
AGIBOT Unveils Genie Envisioner 2.0, Advancing World Models into Scalable “World Simulators” for Embodied AI
AGIBOT has unveiled its latest innovation, Genie Envisioner 2.0, a significant advancement in embodied artificial intelligence. This new platform transforms traditional world models into scalable and interactive simulators, enabling robots to learn and optimize their performance within environments generated by these models. The launch, which took place recently, signifies a pivotal shift from merely understanding the world to actively engaging with it, enhancing the robots' training capabilities and facilitating real-time interactions. This development aims to improve the efficiency and effectiveness of robotic learning processes, positioning AGIBOT at the forefront of AI technology.
Embodied AI World Models Robotics Simulation Technology Artificial Intelligence
Embodied Large Models: Aligning Evaluation First, Then Aligning with the World
Researchers in the field of robotics are grappling with the significant challenges posed by embodied intelligence, particularly the disparity between simulated environments and real-world applications. In response to these issues, a new benchmarking platform called RoboChallenge has been launched. This initiative aims to provide standardized evaluations for robotic models, addressing the pressing need for objective assessments to propel advancements in the industry. By establishing a consistent framework for evaluation, RoboChallenge seeks to bridge the existing gap and enhance the practical deployment of robotics in various settings.
Embodied Intelligence Robotics Benchmarking AI Evaluation RoboChallenge Simulation to Reality
GenEgoData: The Industry's First Dataset for Embodied World Models Officially Released
JianZhi Robotics has unveiled GenEgoData, the first multimodal dataset specifically designed for embodied world models. Launched recently, this innovative dataset captures high-quality, natural human interactions from an ego-centric perspective. The primary goal of GenEgoData is to improve the understanding of physical world dynamics and human behavior, providing valuable insights for researchers and developers in the field of robotics and artificial intelligence. By focusing on realistic interactions, the dataset aims to bridge the gap between human experiences and machine learning applications, ultimately enhancing the development of more intuitive and responsive robotic systems.
Embodied Intelligence World Models Human Behavior Data AI Robotics
Beyond the VLA: NVIDIA’s DreamZero and the ‘GPT-2 Moment’ for Robotic World Models
NVIDIA GEAR Lab has introduced DreamZero, an advanced World Action Model (WAM) featuring 14 billion parameters. This innovative model employs video diffusion technology to provide robots with a form of physical "imagination," allowing them to complete tasks without prior training and adapt quickly to various robotic forms. The unveiling of DreamZero marks a significant advancement in robotics, showcasing the potential for enhanced flexibility and efficiency in robotic applications. By leveraging this cutting-edge technology, NVIDIA aims to revolutionize how robots interact with their environments and perform complex tasks autonomously.
Dr Jim Fan NVIDIA World-Models Research embodied-ai
NVIDIA Launches Earth-2 Family of Open Models — the World’s First Fully Open, Accelerated Set of Models and Tools for AI Weather
At the American Meteorological Society’s Annual Meeting, NVIDIA introduced its groundbreaking Earth-2 family, a suite of open models, libraries, and frameworks designed for weather and climate artificial intelligence. This initiative marks a significant advancement in the field, as it presents the world’s first fully open, production-ready weather AI solutions. The unveiling took place during the annual event, which gathers experts and stakeholders in meteorology to discuss advancements and innovations. NVIDIA aims to enhance the accuracy and accessibility of weather forecasting and climate modeling through this initiative, responding to the growing demand for reliable climate data amid increasing environmental challenges. By providing these resources openly, NVIDIA seeks to foster collaboration and innovation within the scientific community, enabling researchers and developers to build upon their work and improve predictive capabilities in weather and climate science.
DeepMind CEO Demis Hassabis: World Models and 'Infinite Training Loops' are the Keys to AGI
In the season finale of the Google DeepMind podcast, Demis Hassabis discussed the limitations of language models in advancing robotics. He emphasized that while language models play a crucial role, they are insufficient on their own for the development of physical AI. Hassabis highlighted the importance of integrating world-generators, such as Genie, with agents like SIMA to create a more effective synergy that can enhance robotic capabilities. This collaboration aims to address the challenges faced in the field of AI, particularly in bridging the gap between virtual understanding and real-world application. The insights shared during this episode reflect ongoing efforts to innovate and improve the functionality of AI in practical settings.
DeepMind Google embodied-aiRobotics needs a service framework.
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Join the RobotToday community on LinkedIn
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.
