Renmin University of China's GeWu-Lab and the Beijing Humanoid Innovation Center will livestream a presentation of the Robo-ValueRL millimeter-level VLA reinforcement learning solution on October 21, 2026 at 19:00 Beijing time, with talks by Hu Di and Xia Wenke.
NEURA Robotics and NXAI announced a partnership to jointly develop memory components for cognitive robots, combining NEURA's Physical AI technology stack with NXAI's xLSTM recurrent architecture.
SUPCON held an industrial AI results conference in Ningbo, launching an industrial AI foundation platform centered on the TPT time-series large model and integrating it with the universal control system UCS to build the AOP autonomous operations factory system, connecting the perception-analysis-decision-execution chain. TPT and AOP have been replicated across 29 sub-sectors, and SUPCON signed industrial chain cooperation agreements with Huawei, Alibaba Cloud and others, while the industrial AI data alliance has more than 300 members. Companies including Wanhua Chemical and Zhenhai Refining & Chemical disclosed the results of TPT in early warning and optimized operations at chlor-alkali and refining units.
Recently, the draft for public comment of the national standard "Specification for the Construction of Humanoid Robot Data Training Grounds" (plan number: 20261784-T-604), which Zhujing participated in drafting, was released. The standard is administered by TC591 and executed by TC591SC5, with the China Machinery Industry Federation as the competent authority.
Recently, the draft for public comment of the national standard "Specification for Construction of Humanoid Robot Data Training Grounds" (plan number: 20261784-T-604), in whose drafting Zhujing participated, was released. The national standard is administered by TC591 and executed by TC591SC5, with the China Machinery Industry Federation as the competent authority.
Mobile ALOHA is a low-cost, open-source dual-arm mobile manipulation system proposed by a Stanford research team. It adds a mobile base and a whole-body teleoperation interface to the ALOHA dual-arm teleoperation platform, synchronously collecting vision, robotic arm state, and base motion data through teleoperation, then training policies with imitation learning methods such as behavioral cloning. Among these, ACT predicts a segment of an action sequence at a time.
Pollen Robotics has released Microduck Arena, a virtual arena where Microduck robots race, train, and compete without the need for physical robots. Users train policies with reinforcement learning and publish them to their own Hugging Face accounts to enter challenges; each submission is evaluated uniformly by the Arena, and results can be compared across players. Published policies can be downloaded as .onnx files to watch replays and learn from the best runs.
Why it matters: It connects policy training, unified evaluation, and replay-based learning into a closed loop, letting readers learn how to train and compare reinforcement learning policies without physical robots.
Shanghai AI Lab has launched Shusheng SCP-Lab, an experimental equipment access and collaboration protocol based on the Scientific Context Protocol (SCP), providing research agents with a unified mechanism for device capability discovery, task interaction, and trusted feedback. It has been open-sourced.
Why it matters: The original text presents the specific design of the layered architecture, five control granularity levels from L1 to L5, and the trusted feedback mechanism, allowing readers to understand the protocol approach for research agents to connect to heterogeneous experimental equipment.
Being-H0.7, the general-purpose dexterous manipulation large model independently developed by ZhiZaiWuJie, has been accepted to NeurIPS 2026, marking the Being-H series' completion of acceptances to the three top conferences ICML, CoRL, and NeurIPS within one year.
Sharpa launched three products at IROS 2026: the D01 manipulation robot with integrated tactile sensing, featuring an end-effector speed over 10.5 m/s, repeatability of 0.2 mm, and electronic-skin tactile sampling at 100 Hz; the W02 fully tactile compact dexterous hand, with 21 active degrees of freedom, about 30% smaller than the W01, and fingertip force sensing from 5 mN to 30 N; and the AE01 high-fidelity tactile exoskeleton data glove, which uses 22 encoders for teleoperation and data collection and supports 256-level continuous tactile feedback per fingertip.
Lingchu Intelligence has officially released its embodied model Psi-R2.5, which uses a two-tier architecture with a QwenVL3.5-4B high-level planner plus a Wan2.2-IT2V-5B low-level controller, and proposes the strong Pair Data standard. The model reverse-invokes the Psi-W0 world model to generate human demonstration videos backward from real-robot data, enabling mass production of paired data, and supports in-context learning (ICL) for zero-shot generalization to new tasks. It comes with a HIL+RL post-training framework and is claimed to raise task success rates to 99% after a few iterations taking only 1-2 working days, and has already been deployed at customer sites.
Why it matters: The original article details the model architecture, the strong Pair Data data standard, and the reverse-generation method, letting readers understand how its data approach supports real-robot generalization.
High-jump star Zhang Guowei visited the Beijing Humanoid Robot Innovation Center, where he interacted with Tiangong Omni, Tiangong Ultra, and Tiangong 3.0 in running, high-jumping, and tug-of-war experiences; Beijing Humanoid gifted Tiangong Ultra to Zhang Guowei. Tiangong Omni can run at speeds exceeding 10 m/s, while Tiangong Ultra has broken world records 34 times and set a 3.4-meter high-jump record, supported by the Huisi Kaiwu embodied intelligence platform. Its data base covers nearly 6,000 square meters, is equipped with more than 150 robots, has an annual data capacity of up to 180,000 hours, and has delivered nearly 30,000 hours of real-machine data.
Skild AI announced that it post-trained its foundation model S1 through physical self-play, letting the policy play against a recent version of itself with scoring as its sole objective. In NVIDIA Isaac Sim, over 140 years of simulated time, it learned skills such as dribbling past opponents, shielding the ball, and tackling, and these skills were successfully transferred to humanoid robots for matches. The authors believe self-play can break through the capability ceiling of pretrained human data, and the next step will be to extend it to everyday tasks such as collaborative manipulation and city-scale navigation.
Why it matters: The authors describe using physical self-play to let a foundation model acquire soccer skills in simulation and transfer them to real robots, which lets readers learn about a feasible path for self-play-style post-training.
The Embodied Intelligence and Intelligent Computing Development Forum of the 2026 AI Computing Conference (2026AICC) was held in Beijing on September 21, hosted by Tsinghua University's Institute for Embodied Intelligence and Robotics, under the theme "Breaking Through the Critical Point of Human-Machine Symbiosis." Scholars including Zhang Tao and Zhao Mingguo of Tsinghua University, along with representatives from companies such as Lingyu Intelligence, Shouyi Technology, and Qianjue Technology, shared results on directions including clinical applications of embodied intelligence, brain-inspired approaches for humanoid robots, data infrastructure, and world models. The forum built an exchange platform for industry, academia, and research, promoting the engineering and deployment of embodied intelligence technologies.
Psi Robotics released the Psi-R25 robot manipulation model. Its core upgrades are improved data quality and data attribute compression: it reviewed and extracted high-quality data from 100,000 hours of data, and used strong pair data to align human data with robot data.
Why it matters: The original article details how strong pair data is generated and the data quality standards, giving readers a referenceable path for aligning human data to robot data.
Figure AI released Helix 2.5, which completed three long-horizon full-body behaviors—tidying a living room, folding towels, and making a bed—zero-shot in 30 Bay Area homes where no data had ever been collected, without collecting data or fine-tuning in the evaluation environments.
Why it matters: The original article provides controlled comparison data on Index pretraining for zero-shot generalization and evaluation details, helping readers understand the actual level of humanoid robot pretraining transfer.
In a humanoid surgical assistant robot demonstration hosted by Samsung Medical Center, the lead surgeon issued commands by voice only, and two humanoid robots, acting as an instrument-passing nurse and a camera-holding assistant respectively, collaborated to reproduce a cholecystectomy on a goat liver.
ROBROS unveiled Real-Time FBT, a real-time full-body teleoperation technology for its IGRIS-C humanoid robot that maps the operator's movements onto the robot's full-body motion in real time. The demo includes basic full-body movements such as squatting, kicking, turning, and bowing, as well as box handling, sandbag punching, and ball kicking. The technology extends teleoperation from the previous upper-limb-focused control to the whole body, allowing the operator to coordinate locomotion, posture changes, and manipulation in real time as one continuous behavior.
Dr. Wu Min, CTO of Elite Robots, delivered a keynote speech titled "Optically Connected Intelligence: Large-Scale Deployment of Embodied Intelligent Robots in the Optical Communications Industry" at the Embodied Intelligent Robot Integrated Applications Forum of the 27th China International Optoelectronic Exposition, which opened on September 10.
Researchers at MIT CSAIL and Tianjin University have introduced X-Hinges design software, which automatically embeds two different materials when 3D-printing interactive objects, allowing sensor signals and wire-like traces to avoid interfering with each other, and supports up to three axes of motion. The team used it to make a smart glove that transmits grasp-motion data to a humanoid robotic hand, an origami-style desk lamp that senses pinching to adjust brightness, and a detection device that can identify the objects placed on it, and used a computer vision system to align sensor-reading differences among devices from the same batch. The work will be presented at UIST.
PKU EPIC Lab has released FoldNet++, a large-scale synthetic dataset for robotic T-shirt folding and unfolding, covering 6 robotic embodiments, 1K T-shirts, 1K environmental assets, and 120K richly annotated demonstrations.
Why it matters: The source gives the dataset scale and the real-world deployment success rate, so readers can assess the feasibility of training garment manipulation policies purely on synthetic data.
During the World Robot Conference, Yunqi Capital co-hosted a themed roundtable with SEE Fund, which was included in the podcast Attent!on. Three founders—Xu Huazhe of PokeBot, Li Qiang of Quantum Dynamics, and Liu Nianqiu of Feimo Yi Technology—shared their views on world models, scaling, and deployment.
Why it matters: The three entrepreneurs approached the topic from models, hardware systems, and mobility respectively, specifically breaking down where the bottlenecks in embodied intelligence deployment lie and where they diverge on scaling efficiency.
Songyan Dynamics has launched Scalabot, an embodied intelligence sub-brand, and released its first world model, HERON-World Model. Building on its strengths in robot hardware, the brand focuses on letting intelligence grow in the real world.
The Cell and BioRobotics Group (CBoRG lab) at the Technical University of Munich (TUM) Munich Institute of Robotics and Machine Intelligence (TUM MIRMI) proposes "Embodied Laboratory Intelligence," in which robotic assistants and virtual agents work together to explore the internal mechanisms of cancer cells.
BeingBeyond and PaXini AI have signed a cooperation agreement, making BeingBeyond an official PaXini Inside ecosystem partner. The two parties will collaborate in depth on embodied foundation model R&D, POC validation for deployment scenarios, and co-building embodied data standards. PaXini will provide key multimodal data for Being-H0.8, the implicit tactile world-action model BeingBeyond released on July 28; the model was pretrained on over 500,000 hours of human video data.
PKU EPIC Lab proposes GIF, an agentic generation framework for robot learning that reframes the problem as disentangled reconstruction followed by relative pose recovery. CoGen combines 2D and 3D generative models to produce instance-disentangled meshes and coarse initial poses; GPRM refines the relative poses under joint geometric and physical guidance; and a VLM verifier selects the candidate that best matches the structured specification.
Why it matters: The paper splits interactive and functional object composition generation into disentangled reconstruction and relative pose recovery, and provides a benchmark of eight contact-geometry classes along with collision-rate figures, which can help readers assess this route to simulation data generation.
On September 1, JUBAOPEN, the Greater Bay Area's first embodied intelligence innovation community, officially opened at Kexing Science Park in Qianhai, Shenzhen. It is led by OneRobotics. The community covers a total area of 6,039 square meters and can run up to 100 teleoperation and UMI data collection terminals in parallel, covering 50 to 100 types of task scenarios, using a "1+N" real-world setup. At the launch ceremony, five first-batch innovation projects, including Physen AI, signed to move in.
At the 2026 XAIR Expo Embodied Intelligence Industry Expo, Zhido Future unveiled a technical preview of its differentiable world model "Zhizi World" (PGD World Model) and demonstrated validation progress on its embodied intelligent robot Xiaozhi S3. "Zhizi World" reconstructs particle-based scenes from short multi-view videos and can edit parameters such as object pose, material, mass, and friction to predict the physical outcomes of actions. S3 Xiaozhi features bionic dual arms, lidar-vision fusion navigation, and multimodal voice interaction; on-site demonstrations included scooping popcorn, filling cups and delivering them, and self-service retail shopping.
PKU EPIC Lab published the ZETA study on arXiv (arXiv:2609.02546), proposing a unified definition of zero-shot cross-embodiment VLA transfer and a controlled benchmark covering 14 held-out target embodiments, distinguishing strict zero-shot transfer from pretrain-exposed transfer. Experiments show that local end-effector state-action representations, source embodiment diversity, and auxiliary co-training improve transfer by about 15, 18, and 7 percentage points, respectively, and that adding only 5% target-embodiment data during pretraining raises average target-embodiment progress by 13.4 percentage points.
Why it matters: The paper provides a unified definition of zero-shot cross-embodiment transfer and a controlled benchmark, allowing readers to understand the actual benefits of different representation and data strategies.
BeingBeyond and Shutu Technology recently formally signed a strategic cooperation agreement. The two parties will deepen collaboration around high-quality data construction, model training and validation, and industry data standards, and will explore building an industry-oriented data-sharing ecosystem platform. Shutu Technology had previously served as one of BeingBeyond's core data partners, providing data support for embodied foundation model training. On July 28, BeingBeyond released the embodied foundation model Being-H0.8, which was pretrained on more than 500,000 hours of raw first-person video data.
Pollen Robotics has opened pre-orders for Microduck at $399 (excluding taxes and shipping), with first deliveries targeted before Christmas 2026. The 25 cm-tall robot is built for reinforcement learning, with 15 degrees of freedom, a front camera, an 8x8 LiDAR, two IMUs and a full sensor suite, and supports designing behaviors in simulation and transferring them to the real robot.
Why it matters: It packs the full loop from simulation training to real-world deployment into a small, pre-orderable complete robot, letting readers gauge the barrier to building their own reinforcement learning workflow.
Lingchu Intelligence announced the completion of a new funding round of over $100 million. Investors include Tuopu Group, Ruicheng Fund under Chery Holding, Lens Technology, 37 Interactive Entertainment, Wuhu Municipal Investment Holding Group, and Fosun Chuangfu, while existing shareholder Zhuhai Technology Industry Group made an excess additional investment. The proceeds will focus on embodied world model R&D and commercial deployment in logistics and advanced manufacturing scenarios. The company's self-developed dual-system architecture comprises the Psi-R2 manipulation policy model and the Psi-W0 action-conditioned world model, and it plans to expand the scale of human data collection and accelerate the layout of data collection facilities.
Why it matters: Lingchu Intelligence officially announced the completion of the funding and stated that the funds will go toward world model R&D and commercialization in logistics and manufacturing scenarios, giving readers insight into its industry-finance resource layout.
Figure AI announced it is exiting stealth mode and launching the Index app on Google Play and the App Store to collect real-world physical data from Creators worldwide to train its Helix AI system.
Why it matters: The original article provides scale figures for the Index data pipeline and a five-stage processing workflow, which readers can use to assess the actual progress of its human data collection approach.
The paper on InspecSafe-V1, an industrial multimodal safety evaluation dataset released by Tianchuang Robotics together with Tsinghua University, DAMO Academy, and Southeast University, has been accepted by Scientific Data, a journal under Nature.
At the Beijing World Robot Conference (WRC), held August 19–23, X Square Robot showcased three exhibition areas: home services, logistics sorting, and bodyless data collection. Its fully self-developed end-to-end unified world model WALL-B drives robots to complete dozens of household chores. Logistics sorting reaches an efficiency of 1,816 items per hour with 98% accuracy, at a cost 70% lower than the "humanoid robot + five-fingered dexterous hand" approach, and has already entered routine operation on real production lines. The QUANXTA Zero data collection solution achieves a data-ingestion validity rate above 85% and can reduce training-data costs by 90%.
In August 2026, the 2026 Embodied Intelligence Summer Special Training and Study Camp, hosted by Zhejiang Lingqiao Intelligent Technology Co., Ltd. and co-hosted by Shanghai Artificial Intelligence Research Institute and Zongshen Chuangke (Shanghai) Technology Co., Ltd., concluded. Over the 5-day, 40-class-hour program, all trainees from more than ten institutions and dozens of enterprises and public institutions passed the assessment, with an overall satisfaction score of 96.8 points.
PKU EPIC Lab proposed the HumanTracker benchmark for evaluating humanoid motion tracking, accepted to ECCV 2026. The benchmark contains approximately 153 hours of optical motion trajectories from multiple professional performers, organized into four motion families with text labels; the accompanying HumanScore metric is trained on 12K motion pairs containing 24K motions, aligns with human preferences, and can reveal contact and stability failures that kinematic metrics often miss, such as foot skating and mistimed landings.
Why it matters: The original work proposes a benchmark and metric that align humanoid motion tracking evaluation with human perception, and points out that existing kinematic errors miss contact issues such as foot skating.
Persona AI collaborated with cinematographer Pekka Varis to convert CAD data into USD format within 30 days, using motion capture data and Isaac Lab joint motions on a Dell Pro Max Tower T2 workstation (NVIDIA RTX PRO Blackwell 6000 GPU).
OneRobotics and 58.com have formally signed a strategic cooperation framework agreement, under which they will collaborate on real-world scenario data sharing, robot application deployment, and industry talent development, initially focusing on home, sports and health, and commercial services scenarios.
After completing the deployment of virtual simulation data applications, Topstar has filled in its real-machine data collection capability and launched a real-machine multimodal data collection system. The system is centered on a self-developed high-precision multimodal data collection gripper, which can adapt to various robot models without joint mapping and supports six-degree-of-freedom high-precision pose tracking, with an average positioning error of ≤0.25% under normal operating conditions. It also simultaneously records multidimensional information such as visual images, motion trajectories, end-effector poses, and gripper actions, and is compatible with single-arm and dual-arm process teaching.