At IROS 2026, University of Queensland researcher Jen Jen Chung argued in a keynote that data volume is not the same as information, and that blindly piling up data can degrade robot policies. Drawing on drone soaring, multi-robot communication and contact-rich assembly work, she said data value is unevenly distributed and that sharing or collecting only decision-changing information beats full broadcasts or uniform sampling.
At IROS 2026, running September 27 to October 1 in Pittsburgh, Leiphone reports embodied AI research is converging on four trends: foundation models internalizing physics, "brain-cerebellum" architectures becoming standard, touch as core to dexterous manipulation, and cheaper data collection. Papers cited include ULTRA, LITHE, VOFA, KineFuse, VTAP Gripper, ViHaTeleop, and HALO, alongside workshops on physical world models and uncertainty.
Mifeng Technology launched Mifengpai, which it calls the world's first full-category high-quality physical AI data crowdsourcing platform, at a Shanghai event on September 23. The platform pairs MEgo capture devices and the MEgo Engine processing platform with a collector network, and covers 22 categories, over 5,000 task types and more than 50,000 real scenarios. The company also announced a robot trainer program and a 100 million yuan subsidy plan.
The 2026 China International Exhibition on Public Safety and Security and the Third Intelligent and Security Industry Development Conference will be held from November 3 to 6 at the China International Exhibition Center (Shunyi Hall) in Beijing, under the theme "Intelligent Engine Reshapes Security." This year's security expo will feature an embodied intelligence core exhibition area for the first time and will showcase innovations in AI large models and security intelligent agents, with companies such as Huawei, China Telecom, Tiandy, and Vimicro participating.
Can today's frontier language models control a real robot arm, zero-shot, with no coaching? AA-Robotics v0.1 coming soon.
Edinburgh-based Danu Robotics is bringing its recycling sorting robot H.E.R.O. to market; the robot grasps items with grippers rather than a suction system, and uses AI to continuously improve its software. The company has raised $5 million in late-seed funding, holds $500,000 in signed contracts and letters of intent from two major customers, and has more than 200 customers in its sales pipeline.
Aether AI releases the official demo of its causal intelligent robotic system CRIS-0, showcasing its disturbance resistance and long-horizon task capabilities on a real robotic arm. In a coffee-making disturbance test, CRIS-0 identified changes in causal variables and completed replanning within an average of 2 seconds, effectively recovering in 9 out of 10 random disturbances; in dozens-of-steps long-horizon tasks such as finding and tidying objects in a living room, the system decomposes goals into verifiable causal stages and progressively checks preconditions and postconditions.
RobotEra's in-house world action model VPP2 ranked first on the RoboDojo simulation leaderboard, with the highest overall average success rate (32.26%) and average score (39.26 points), ahead of GPT-6-Astra at 22.48% and 28.97 points.
At IROS, Sharpa unveiled three products: D01, a humanoid robot for dexterous manipulation with integrated tactile sensing; W02, a next-generation fully tactile dexterous hand; and AE01, a high-fidelity exoskeleton haptic data glove.
The MirroS team launched AgentGarten, connecting an executable code environment to a real-time neural renderer: code defines the physics, while the model generates visual feedback, enabling continuous agent interaction at more than 30 fps. In a recreated hide-and-seek experiment, hiders learned to move barriers into cover by round 4, while seekers learned to use a ramp to climb over a wall by round 10; comparable OpenAI experiments in 2019 took about 25 million and 100 million rounds, respectively. The team also published a blog, technical report, code repository, and project site.
NVIDIA developers are using frontier models such as GPT-6 Astra and Claude Fable 5 to power the Omniverse libraries and build simulation apps from natural-language instructions. Examples include a warehouse humanoid-robot simulator, a Market Street autonomous-driving test environment, digital-twin sensor validation, and a Unitree G1 humanoid hurdle-jumping experiment that succeeded in 64 of 100 simulations.
The California State Athletic Commission issued a cease-and-desist order to startup Rek, which stages human-versus-machine fights. The commission said a September 18 bout between human Frankie LaPenna and a humanoid robot controlled remotely by a person using a VR system constituted an unapproved boxing or mixed martial arts event. Rek CEO Cix Liv posted the letter on X on September 30. Rek plans to hold a fight between two armed robots this Friday.
In an interview, Periodic Labs' Liam Fedus and Ekin Dogus Cubuk proposed a "synthesis superintelligence" approach: reinforcement learning in physical experiments, AI-driven materials characterization, density functional theory (DFT), and high-throughput labs. The aim is for models to learn the full process of doing science, rather than only studying published results.
In the digital world, AI has made almost anything possible. Before long, the same will be true of the physical world (also thanks in large part to AI).
Is there anything Reachy mini can’t do?
At APRCE 2026, Zhengxing Innovation released a Physical AI solution for open retail scenarios, simultaneously launching the bipedal humanoid robot H1 and the wheeled-arm robot C1 for handling and shelf stocking, supporting rapid "zero-modification" store deployment. The solution consists of the robot bodies and the unified management platform M1. The robots can autonomously complete 99% of tasks. Commercial services are planned for 2027, with support for direct purchase or RaaS subscription.
Nvidia is extending its Halos safety system from autonomous driving to robotics. Announced in June 2026, Halos for Robotics targets warehouse autonomous mobile robots, humanoid robots in factories, and surgical robots. It provides hardware and software tools to help developers build safety guardrails into physical AI devices. Amit Goel, who leads Nvidia's robotics ecosystem and edge computing, said safety is becoming the next bottleneck as AI models and robot hardware improve.
An article produced by MIT Technology Review with Aventine says that while Elon Musk claims Tesla Optimus could eventually automate nearly all human labor for as little as $20,000 per robot and predicts public sales before the end of 2027, many robotics researchers are skeptical. Yann LeCun said at Davos that no humanoid robotics company knows how to make robots smart enough to be truly useful.
Benewake founder Li Yuan has spent eleven years building lidar for drones, trains, cars, and robots. In the first half of this year, Benewake delivered 250,000 lidar units for robots; robotics accounted for more than half of company revenue, and 2025 robotics revenue doubled from 2024. Li believes measurements such as distance and position from lidar should become learning material that helps machines understand the physical world.
Arti Garg, AVEVA's chief technology expert, says industrial AI is shifting from predictive analytics to foundation models, physical AI, and agentic AI, and requires connected data, governance frameworks, and human oversight for safety. AVEVA's responsible AI framework emphasizes safety, efficiency, and human oversight, arguing that AI should augment rather than replace people in critical decisions. Garg also participates in an IEEE working group developing a standard method to measure AI's environmental impact across power, energy, resources, water, and carbon.
At the IROS 2026 venue, Feng Zongbao, head of reinforcement learning at Zhijian Dynamics, presented a solution that enables a robot to autonomously tend two CNC machine tools, completing grasping, loading, insertion, removal, and placement under the condition of only a 0.5-millimeter gap between the workpiece and the chuck, with training using only 600 real robot trajectories.
Li Jianxiong, the first PhD student at Tsinghua University's Institute for AI Industry Research (AIR), founded embodied AI company BASAL Intelligence. 5Y Capital led the round, with Ivy Capital, Linear Capital, and Huashan Capital participating. Tsinghua AIR associate professor Zhan Xianyuan is the chief scientist.
The GaLbot team at Galaxy General, together with Tsinghua, Peking University, and others, proposed LATENT, which trains humanoid robots to play tennis using about five hours of imperfect motion data from five amateur players. The method first trains a motion tracker and compresses it into a latent action space with a Latent Action Barrier constraint. On a Unitree G1, it achieved a 90.90% success rate on real forehands and 77.78% on backhands.
IROS 2026 accepted 1,933 papers. A multi-label analysis counted about 809 papers in Robot Learning / Embodied AI and 564 in Navigation / Planning.
The IROS 2026 Best Paper Award went to Yumin Lee and colleagues for LT-Mem. The work addresses temporal forgetting in long-running robots with a three-layer structure of Live Memory, Delta Memory, and Meta Memory, and the concept of volatility.
The Centre for Artificial Intelligence and Robotics (CAIR) at the Hong Kong Institute of Science and Innovation, Chinese Academy of Sciences, launched CARES 4.0, a multimodal clinical agent system, at the 4th Hong Kong Embodied Intelligence and Medical Technology Forum. Built on the Harness agent framework, it uses in-house foundation models for CT, MRI, ultrasound, endoscopy, and EEG, and has been validated at several top-tier hospitals. The CARES 4.0 website is open. Its framework and general toolkit are available, medical toolkits are being released in stages, and its in-house open-source ultrasound and surgical-video models will be integrated and made freely available.
KargoBot announced on September 22 in Qipanjing Town, Ordos, the launch of a large-scale AI transport network, putting manned heavy trucks, driverless trucks with retained cabs, and cabless transport robots into the same transport system. Its driverless transport fleet has reached the scale of about one hundred vehicles, with operations covering 6 provinces including Inner Mongolia, Xinjiang, Gansu, Shaanxi, and Shanxi, and it cooperates with 30 customers, 8 automakers, and about 30 ecosystem partners.
ONTIME Mobility COO Han Feng believes that as world model capabilities grow stronger, the importance of real data will not diminish: synthetic data is responsible for expanding known long-tail scenarios, while real data is responsible for discovering unknown problems and providing calibration and validation. ONTIME Mobility does not train large models; instead, it collects real data through routine mobility services, Robotaxi, and offline operations and maintenance networks, and delivers training materials to autonomous driving, world model, and embodied intelligence companies through a data closed loop.
At the RoBoWoMo workshop at IROS 2026, Stanford assistant professor Wu Jiajun presented 'Building Physical Agents via Structured Representations'. He argued that structured representations are not answers to hard-code into a system, but scaffolding that helps robots learn efficiently and communicate with people.
It is increasingly clear that humanity will solve the Riemann hypothesis before we pass the Physical Turing Test - you come back to a clean home after a wild Sunday party, and you can't tell whether a human or a robot did the job. We are not Moravec's paradox-pilled enough.
Researchers at PhAI Labs, Stanford, Oxford, and Princeton proposed JEPA-Anything. It splits JEPA's predicted state into four orthogonal factors, assigns a prediction module to each, and recombines them. It outperformed standard JEPA with the same architecture and data on ten test tasks in physics, robotics, weather, and other areas.
Reka AI released a research preview of Rho-1, a 19B-parameter general-purpose model that can process and generate text, images, video, and robot-control actions in a single neural network. Unlike most systems that route tasks to specialized models, Rho-1 places all modalities as tokens in one shared context window without calling tools or external models. It can generate continuous video in real time, respond instantly to new instructions without restarting, and use the same weights that predict camera images to drive robot actions.
Safeworld was founded by Zhao Ding, director of Carnegie Mellon University's Safe AI Lab, together with Kyle Wong and Simo Rachidi. It uses simulation environments such as Genesis and MuJoCo to evaluate the safety of robot-control systems driven by generative AI.
Paril Jain, co-founder and CTO of The Bot Company, and Quan Vuong, co-founder of Physical Intelligence met on the Runway AI Summit stage to talk real-world evals, deploying robots in the field and where robotics will be in three years.
Destro emerged from stealth with an $8 million seed round. Its AI intelligence layer lets robots work in logistics settings while directing human workers, instead of building its own robot hardware.
Stanford's HomeBody project connects GPT-6 Astra to a Unitree G1 in a kitchen, where it follows language instructions to gather coffee bags, throw away a carton, and get medicine from a drawer to hand to a person. The system works in three stages: it first explores the kitchen and saves camera images, lidar, and joint poses as spatial memory; then reconstructs a digital kitchen in Isaac Sim; finally Astra uses images, maps, and memory to call navigation, grasping, placing, and drawer-opening skills.
At an ECCV 2026 workshop, NVIDIA Research scientist Li Boyi presented two projects, MultiGen and DexImit. MultiGen uses diffusion models to add synchronized audio to videos generated by physical simulators.
In the first keynote at ECCV 2026, University of Texas at Austin professor Kristen Grauman argued that visual AI should move from understanding to enabling, becoming an AI Guide that helps people learn real-world skills.
ECCV 2026 received 10,473 submissions and accepted 2,834. The 3D topic fell from a leading area to third place, but questions around 3D representation, geometry, and physical consistency are spreading into video generation, embodied AI, and world models. Of 93 workshops, about 10 were directly related to 3D, 11 to spatial intelligence, nine to physical AI, and five to world models. Best-paper and honorable-mention awards also focused on 3D surface representation, object reconstruction, and geometry perception.
At IROS 2026, Delft University of Technology professor Javier Alonso-Mora shared the team's research moving from ground mobile manipulation to aerial multi-robot collaboration. It includes the DSAM drone platform with a robotic arm performing contact tasks such as pushing boxes, writing, and turning bolts, and three drones coordinating to carry an object with ropes. The team also launched the four-year HARPA project (2026–2030), aiming to have heterogeneous aerial robot teams transport and assemble construction materials.
