A research team proposes uni-robot-agent, a unified framework to support further training of open-weight robot-use agents. Without privileged inputs, the out-of-the-box Qwen3.8-Flash-Next achieves a RoboDojo score of 31.39, surpassing the published GPT-6 Astra-RoboProbe result of 28.97, and performs competitively on 19 real-world tasks.
Tencent Hunyuan used the open-source world model Hy-Nomas to reproduce 8 demos from the official Google Genie 4 website and released a side-by-side comparison; it is close in interactivity and content completeness, but still lags in native audio, camera smoothness, and visual style consistency.
Xuanji Dynamics' Hypertron series quadruped robots won the "High-Rise Fire Rescue Track" championship and first prize in the "Load-Bearing Climbing and Material Transport" category at the 2026 Guangzhou Safety Emergency Rescue Robot Competition. The Hypertron-SW01 is equipped with a self-developed high-torque joint motor module and a multimodal fusion perception system, features IP67 protection, and continued stable motion control, perception, and communication during a 3-minute continuous bake test at 120°C–130°C.
WorldOracle presents a case where the real gripper failed to grasp the fruit, but the generated rollout showed the fruit being picked up, despite the model receiving the same initial observation and recorded actions. When evaluating predictions, check separately whether the robotic arm follows the trajectory and whether the object's final state is correct; for an empty grasp, predicting an empty grasp is the correct answer, while a seemingly successful grasp video may actually be wrong.
The Qiyuan Embodied Research Center has released PrimeSim, a world-model simulator for robot manipulation tasks, built by adapting a pretrained video generation model through mid-training and post-training stages. It ranks first on EWMBench with an overall score of 6.09, and took first place in three of five WorldSimProbe evaluations: global trajectory coverage, action-source behavior preservation, and interaction dynamics.
Why it matters: The original text provides the two-stage training recipe and results across multiple leaderboards, which readers can use to judge the feasibility of adapting video generation models into simulators.
At the dexterous manipulation sub-forum of AgiBot Innovation Day, Critical Point released the OmniGlove Pro data-collection glove and began mass-production delivery. It focuses on "real-robot-ready" data that aligns human hands with real robots using high-precision EMF positioning and full-hand tactile collection.
AgiBot has proposed a robot workload measurement framework called Worken, which multiplies three factors—the number of active robots, autonomous working hours, and output efficiency ratio—to estimate the effective intelligent work hours completed by robots in real deployments. The framework was proposed by co-founder, president, and CTO Peng Zhihui, and emphasizes distinguishing powered-on time from autonomous operating time, and measuring qualified output against task benchmarks.
Humanoid robots are moving from launch-event stages to real-world settings such as the Canton Fair and cinemas for public testing. Guangzhou Zhidong Future deployed robots across five scenarios at the Canton Fair simultaneously, while Dobot's Atom worked 14 consecutive hours in a single day at a Shenzhen cinema and sold over 1,000 cups of popcorn.
At the Dexterous Manipulation sub-forum of AgiBot Innovation Day, Critical Point connected live to the Wolong Hill factory to present industrial POC results for a robotic arm plus dexterous hand. The system continuously completed a 72-step process including loading and unloading, barcode scanning, grasping a screw gun, tightening and removing screws, and pressing buttons. Phase-one task success rate reached 98%, with 8.5 minutes per operation, close to the roughly 8 minutes for a human, and it can run continuously for 8 to 12 hours.
Qiyuan Robotics has launched PrimeSim, a world-model simulator for robot training, and topped the WorldSimProbe LIBERO leaderboard. PrimeSim builds an interactive, trial-and-error, continuously learning virtual world for robots, used for policy validation and reinforcement learning. In tests, reinforcement learning with it raised policy success rate from 61.7% to 73.2%.
X-Era AI Tuoyuan Zhihui released the open-source native 4D world action model E0.5, which uses a 3D foundation model as its backbone and integrates future 3D geometry prediction directly into action learning. On the AgileX Piper dual-arm robot and the Zerith H1 humanoid robot, after pretraining on about 160,000 robot trajectories, E0.5 achieved overall average success rates of 85.0% and 88.3%, respectively, improving on π0.5 by 13.3 and 26.7 percentage points; its out-of-distribution average success rates were 73.3% and 76.7%.
Why it matters: The original text provides a comparison of E0.5 and π0.5 success rates on out-of-distribution real-robot tasks, which readers can use to judge the effect of 3D geometry supervision on manipulation generalization.
Agibot launches the embodied model GO-3 Preview, built on an embodied-native WRAM (World Reasoning Action Model) architecture that connects task reasoning, future prediction, and action generation, supporting full-pipeline scaled learning across pretraining, post-training, and post-deployment.
The PhAIL team and Nebius have launched a robot policy competition. Participants package their policies as Docker submissions, which are first scored in two simulations: MolmoSpaces and RoboLab starting November 16. In December, the top 10 will run the same image on a real Franka. The tasks resemble the DROID benchmark but use a different dataset.
AgiBot has released WITA-Omni 1.0, an omni-modal interaction model that supports full-duplex dialogue with simultaneous speaking and listening, runs on-device with an end-to-end interaction latency of about 400ms, and integrates vision, speech, real-time dialogue, facial expressions, and physical actions into one continuous system. It currently ranks highly on multiple interaction benchmarks.
The Chongqing University "BOBOCUP" team won the national championship in the agricultural robot—harvesting robot category at the 2026 China Robot Competition and RoboCup China Open, which concluded on October 18. This is the university's first championship at the event since it began participating in 2013. The harvesting category requires robots to complete autonomous navigation, intelligent obstacle avoidance, target recognition, fruit and vegetable grasping and harvesting, and collection at a designated location. Another team from the university placed third in the five-finger dexterous manipulation artistic skills challenge, which assesses abilities such as multi-finger fine manipulation, force control, and hand-eye coordination.
At its 2026 Innovation Day, AGIBOT proposed the Worken framework, which measures a robot's effective intelligent working hours in real environments by multiplying three factors: the number of active robots, autonomous working hours, and output efficiency ratio.
The author made a caniuse-style comparison table showing which robot learning policies can run on which robots, covering 24 robots and 21 policies (including ACT, Diffusion Policy, SmolVLA, π0/π0.5, GR00T N1.7, X-VLA, OpenVLA-OFT, and others).
A research team from Aether AI and UC San Diego has proposed RobotAgentBench (RAB), a comprehensive benchmark for open-world robot autonomy, building 50 household layouts, 90 atomic tasks, 18 short-horizon tasks, 8 long-horizon tasks, and 4 types of open-ended scenarios.
AgiBot has released the embodied model GO-3 Preview, focused on full-pipeline scaled learning across pretraining, post-training, and post-deployment, and it ranked first in three evaluations: RoboDojo, RoboTwin, and RoboColiseum.
Why it matters: AgiBot disclosed GO-3 Preview's three-stage learning pipeline of pretraining, post-training, and post-deployment, along with specific benefit data from mixed training using real-machine clusters and world models.
X-Era AI has open-sourced its native 4D world action model E0.5, achieving overall average success rates of 85.0% and 88.3% on the AgileX Piper dual-arm robot and the Zerith H1 humanoid robot, respectively, 13.3 and 26.7 percentage points higher than π0.5.
Why it matters: The original text provides real-robot success rate comparisons between E0.5 and π0.5 on dual-arm and humanoid platforms, which can be used to assess the manipulation generalization performance of the 3D geometric backbone approach.
Ma Chen, associate professor at City University of Hong Kong, has joined Yingjie Technology to lead pre-research on the data direction. The company has completed two consecutive seed funding rounds over the past few months, totaling tens of millions of yuan. Yingjie is building a data pipeline spanning multimodal collection, spatiotemporal alignment, spatial structuring, and dynamic tracking and prediction validation, while advancing spatial intelligence benchmarks and post-training environments, feeding model failure modes back into data collection and training. Ma Chen proposes comparing different supervision methods under the same foundation model and the same budget, focusing on performance in unseen scenarios and task success rates.
Under its Gravity embodied intelligence framework, Geek+ has launched the Gravity Control causal intervention model, which can correct local action deviations in robots with a single natural-language prompt, without updating the original policy weights or adding new demonstration data.
On October 29, AgiBot released its next-generation embodied foundation model GO-3 Preview, which uses the WRAM (World Reasoning Action Model) architecture to connect task reasoning, future prediction, and action generation. It has ranked first on three major evaluation leaderboards: RoboDojo, RoboTwin, and RoboColiseum. AgiBot plans to fully open-source it by the end of 2026.
Apex Intelligence has released and open-sourced its first large model, Apex-35B-V1.0, with 35 billion total parameters, about 3 billion parameters activated per token, and support for a 262,000-token context; it is based on Qwen3.6-35B-A3B and optimized through mid-training and post-training.
In the 50 days since GPT-6 Astra was released, the embodied AI industry has conducted intensive evaluations and discussions around it, and general-purpose large models have begun directly taking over robot operation. Tests by SenseTime and several institutions concluded that Astra is basically capable of completing embodied tasks, but still has a clear gap from practical use in real environments, underperforming in millimeter-level contact, dynamic control, and response speed. The article also reviews industry uses such as using Astra directly for policies, reviewing VLA actions, building simulation environments, and data annotation, as well as chain reactions including cooling financing and major companies stepping up data collection.
Why it matters: It reviews evaluation results, company reactions, and industrial chain changes in the embodied AI industry within 50 days of GPT-6 Astra's release, presenting the differentiated landscape after general-purpose large models entered robot operation.
The Xynova team used the KaRMA simulation to compare 20 dexterous hand models; under the test conditions, its Prima 1 ranked 2nd in the translation category and 1st in the reorientation category. The team says a gap remains between simulation and real-world operation and is seeking more practical dexterity metrics. Xynova will attend CoRL in Austin and Web Summit in Lisbon with dexterous hand demonstrations and has spare tickets for people willing to produce technical content for its dexterous hands.
On October 28, Unitree rose more than 10% during intraday trading, with its stock price returning to 700 yuan. Its 2026 third-quarter report shows revenue of 1.817 billion yuan for the first three quarters, up 55.71% year on year, and net profit attributable to shareholders of 387 million yuan, up 270.54% year on year, but net profit after deducting non-recurring gains and losses was 349 million yuan, down 18.78% year on year.
On October 28, Unitree rose more than 10% during intraday trading, with its stock price returning to 700 yuan. Its third-quarter report shows that revenue for the first three quarters was 1.817 billion yuan, up 55.71% year on year, and net profit attributable to shareholders was 387 million yuan, up 270.54% year on year; however, net profit after deducting non-recurring gains and losses was 349 million yuan, down 18.78% year on year.
Unitree Technology's first earnings report since going public shows that in the first three quarters of 2026, operating revenue was 1.817 billion yuan, up 55.71% year on year, and net profit attributable to shareholders was 387 million yuan, up 270.54% year on year. Third-quarter operating revenue was 665 million yuan, up 69.93% year on year, but net profit attributable to shareholders was 113 million yuan, down 17.37% year on year. R&D expenses rose 191.48% year on year, mainly due to a sharp increase in model R&D and computing power investment.
AgiBot has released the embodied foundation model GO-3 Preview, built on its self-developed WRAM (World Reasoning Action Model) architecture that unifies task understanding, future state prediction, and action generation, and it ranks first on three embodied intelligence benchmarks: RoboDojo, RoboTwin, and RoboColiseum. It is planned to be fully open-sourced by the end of 2026.
Why it matters: The original article fully discloses GO-3 Preview's three-stage design of pretraining, post-training, and post-deployment learning, which can be used to understand the engineering path for scaling embodied models across their full lifecycle.
The 2026 Global Developer Pioneer Conference and International Embodied Intelligence Skills Competition concluded on October 25 at the Shanghai Zhangjiang Science Hall. More than 800 teams and over a thousand robots competed across 9 tracks and 23 events in the industrial, service, and special-purpose domains, testing continuous operation capability in real-world scenarios, with 16 first prizes awarded.
Princeton University, UT Austin, the University of Toronto, NVIDIA, and AgiBot jointly launched WorldOracle, an action-conditioned world model benchmark that uses the same initial frame and action sequence to compare model predictions with real execution.
Why it matters: The benchmark compares world model predictions frame by frame against real execution of the same actions and releases failure samples, making it easier to identify sources of model deviation over long sequences.
AGIBOT's GO-3 Nano Preview has topped the latest RoboDojo embodied intelligence benchmark leaderboard with an average score of 61.90 and an average success rate of 56.42%, surpassing GPT-6 Astra. The model ranked first in all three sub-dimensions: fine manipulation (78.65), long-horizon execution (68.21), and open-ended semantic understanding (61.71).
Chaoyan Intelligence, founded by Chen Yongchao, an assistant professor at Tsinghua University's School of Artificial Intelligence, has released its first 35B model, Apex-35B-V1.0. It was trained and post-trained on Qwen3.6-35B-A3B, balancing general capabilities such as coding and agents with AI research, computational science, and robot simulation tasks.
The You Yangwei team presents three tactile manipulation results: PACE was accepted as a CoRL 2026 Oral and achieved an average success rate of 90.7% across five real-world robotic insertion tasks; ForeTac uses future tactile prediction for action generation.
In an interview, Raphael Memmesheimer, head of the University of Bonn's NimbRo@Home team, said his team won the RoboCup 2026 best paper award for its paper "Efficient Image Annotation via Semi-Supervised Object Segmentation with Label Propagation."
The latest RoboTwin leaderboard shows AgiBot GO-3 Nano Preview in first place with an average success rate of 85.01%. The leaderboard mainly evaluates dual-arm coordinated manipulation and environmental generalization in complex scenarios, with assessment dimensions including target recognition, motion coordination, and adaptation to environmental changes.
GenRobot.AI announced that both of its open-source datasets, Gen-HumanWholeBody and Gen-HumanEgo, have entered the top five of the Hugging Face robotics dataset trending list. Gen-HumanWholeBody is a first-person whole-body manipulation dataset containing 600 hours of whole-body manipulation data, 16,976 real-world scenes, and 1,281 human skills, with whole-body MPJPE of about 3 cm; the authors say it targets whole-body motion learning, human-to-robot transfer, and embodied foundation model research.
Why it matters: Two open-source datasets entering the top five of the Hugging Face robotics dataset trending list can serve as a reference for observing the level of attention to whole-body manipulation data.
The authors propose BiDex-100, a simulation benchmark for bimanual dexterous manipulation comprising 106 tasks organized along five capability dimensions: generalization, instruction following, limited demonstrations, long horizon, and memory. Each task is paired with an out-of-domain variant featuring controlled visual and spatial changes for evaluating robustness. The benchmark supports two embodiments and provides 8,080 success-validated teleoperated demonstrations totaling 6.25 million frames. The authors evaluate seven recent vision-language-action models under a unified protocol, establishing baselines for capability-specific performance, robustness, and data efficiency.
Zhengshu Zhang proposes in an arXiv paper that for diagnosis to improve behavior when a robot faces unfamiliar physical conditions, six ordered empirical conditions must hold: a meaningful reference, identifiability of the physical condition, use of the acquired evidence, decision value, selection value relative to a fixed alternative, and safe realization.