Associate Professor Yu Chao's team at Tsinghua University, together with Striding AI, jointly proposed the self-supervised frame-level advantage modeling algorithm STEAM, which learns frame-level progress signals from the temporal order of expert demonstrations and scores expert data, robot rollouts, and human correction data frame by frame.
A UC San Diego team completed two live pig gallbladder removal surgeries in an ordinary operating room using the teleoperated humanoid robot Surgie, with surgeons directing and controlling the procedures rather than the robot operating autonomously. Surgie is a modified commercial Unitree G1 that can quickly switch between standard surgical instruments and is smaller in size and footprint than conventional surgical robots such as Da Vinci. Participating surgeons reported intuitive control and reduced physical strain, but also noted overheating and the need for frequent repositioning.
Lingchu Intelligence and the PKU-Lingchu Joint Lab released and open-sourced the dual dexterous hand general manipulation model and full-stack system EgoSteer, comprising the data processing pipeline EgoSmith, the robotics technology stack Robot-Stack, and a world-model-enhanced VLA model.
Why it matters: The original text presents a complete chain from human-hand data processing and real-robot correction to model architecture and experimental figures, allowing readers to assess its reproducibility value.
PKU EPIC Lab proposes WAM-TTT, a test-time training framework that steers world-action models from raw human videos without robot actions, human-side annotations, or task-specific fine-tuning. The method absorbs human videos into a lightweight adaptive memory inside a frozen WAM through self-supervised video prediction, and uses a meta-training stage with paired human-robot data and a key-value memory reconstruction objective to align human demonstrations with robot behaviors. Experiments show it consistently outperforms in-context human-video conditioning baselines across diverse manipulation tasks and generalization settings.
Why it matters: The original work proposes a method for adapting a frozen world-action model at test time using only unlabeled human videos, without robot demonstrations or fine-tuning, offering methodological reference value for researchers who want to reuse it.
BAAI has released the multimodal representation world model Wujie·RoboBrain Orca. Its core idea is to shift from predicting the next word, next frame, or next action to predicting the next world state: it first learns a unified latent world representation, then adapts to downstream tasks through three lightweight readout modules for text, images, and actions.
Why it matters: The original text explains the training paradigm shift from predicting the next output to predicting the next state, as well as the approach of a frozen backbone plus lightweight readout modules, which is transferable to readers following the world model route.
If you want a robot to do something well, you need to know how to talk to it. If you don't, you can learn—with Semantic Action RL! In our paper, @JagdeepBhatia8, @ajwagenmaker, and @verityw_ show how to do reinforcement learning on VLA prompts to achieve new tasks and learn extremely fast in the real world!
The Chinese University of Hong Kong, Shenzhen and other institutions propose GeoPredict, which explicitly predicts end-effector and joint-level future trajectories before the VLA generates actions, and uses 3D Gaussian modeling for the geometry of the future workspace.
Why it matters: GeoPredict incorporates future trajectory and 3D geometry prediction into VLA training, giving readers a spatial reasoning approach that does not increase inference overhead.
The team of Hirotaka Sato at Nanyang Technological University developed 3D-printed waterproof diving suits for remote-controlled Madagascar hissing cockroaches (Gromphadorhina portentosa), enabling them to walk underwater for up to 3 hours at depths of up to 50 centimeters. The suits supply oxygen through a reaction between hydrogen peroxide and manganese dioxide; five suited cockroaches remained healthy after three days, and reached an underwater speed of 78.4 millimeters per second (87.5 millimeters per second on land). The team next plans to test the approach under extreme space environmental conditions.
The article explains LocateAnything's Parallel Box Decoding (PBD) mechanism, which treats a complete bounding box as a generation atom of length 6, predicting the class and four coordinates in parallel in a single model forward pass, while blocks remain autoregressive between one another.
Can a rough action intent plus a VLA prior be turned into successful behavior? Flow Reversal steering: 1) Run the VLA's flow ODE in reverse to infer the noise closest to the rough trajectory 2) Run the flow forward to obtain the closest favorable behavior Paper: http://flow-reversal-steering.github.io
We can learn a model that provides shaped "process rewards" for robotic RL, and this model evolves automatically as the policy improves. This improved performance on benchmarks and also works in the real world! Some interesting new work with Raymond Tsao & @ajwagenmaker.
Professor Zhu Yanhe's team at Harbin Institute of Technology (HIT) published a paper in the International Journal of Extreme Manufacturing (IJEM), proposing an autonomous fault-tolerant repair strategy for modular robots based on cooperative reconfiguration, which restores joint fault damage by changing morphology without backup module replacement. The method combines topological description and kinematic self-modeling, particle swarm optimization to search for target configurations, and an improved RRT cooperative reconfiguration algorithm. In experiments, the fault-tolerance upper limit was 7 times higher than that of fixed-configuration redundancy methods. HIT is the sole completing institution, and doctoral student Qi Jian is the first author.
Why it matters: The paper presents a cooperative reconfiguration fault-tolerance strategy that requires no backup modules and data showing a 7-fold increase in the fault-tolerance upper limit, allowing readers to understand a concrete implementation path for self-repair in robot swarms under extreme environments.
Our latest research on octopus-inspired soft robotics has been featured by @Reuters. The study was published in @NatMachIntell, conducted by researchers at @BSRiit under #IIT, coordinated by @BarbaraMazzolai. 🔗 Watch the Reuters report: https://www.reuters.com/video/watch/idRW135222062026RP1/?chan=technology
Adam's new paper proposes Ambient Diffusion Policy, which selectively learns robot policies from suboptimal data through a data-usage approach that depends on noise level. The authors note that when training denoising diffusion models, each data point is used multiple times at different noise levels, and a data point's value may differ by noise level. The method can extract value from a considerable portion of the OXE dataset that was previously discarded, and is complementary to pi0.7's approach of learning conditioned on data quality.
Researchers at Nvidia GEAR lab, Carnegie Mellon University, and UC Berkeley developed an agent framework called ENPIRE that lets AI coding agents autonomously schedule and iterate robot training; the robots learned to cut zip ties and insert GPUs into narrow, thin slots on a motherboard.
JAXA researchers reported in Science Robotics that the palm-sized spherical lunar rover LEV-2 (nicknamed SORA-Q) operated autonomously for more than 100 minutes after landing near Shioli Crater during the January 2024 SLIM mission, traveled an estimated 24 meters, and returned multiple images.
This article is an interview with Dr. Jianyuan Wang, first author of VGGT, which won the CVPR 2025 Best Paper Award, and of VGGT-Ω, a CVPR 2026 Best Paper finalist. He is currently pursuing a PhD through a joint program between the Oxford Visual Geometry Group and Meta AI.
The 2026 8th Beijing BAAI Conference closed on June 13, with more than 10,000 in-person attendees, bringing together 2 Turing Award laureates, 8 academicians, and more than 40 AI company CEOs. Centered on two main themes, world models and agents, the conference held an AI×Neuroscience forum for the first time, and the number of agent forums increased from 1 to 3.
How does test-time scaling affect robots? We found that larger models, more thinking, and more context significantly help some prompts but not others. Similar to large language models, we can also train a router to achieve a better performance/latency trade-off! Paper: https://jadee-dao.github.io/direct/
Chelsea Finn's team fine-tuned a VLA so that a single policy can control any robot in a robot team, enabling multi-robot collaboration without explicit communication (the system described in the quoted tweet is named CHORUS). Results show that this method matches or outperforms training separate models for each robot or training a unified centralized model, and it scales easily to large teams. Paper: https://arxiv.org/abs/2606.12352
Investor Shi Ma speaks with Stanford PhD student Huang Wenlong (advisor Li Feifei) about his PointWorld work and 3D world models. PointWorld uses 3D point flows to unify state and action representations, mapping different robots' joint actions into point cloud changes via URDF to address inconsistent actions across embodiments; training used the DROID and BEHAVIOR datasets, with annotations relying on FoundationStereo, VGGT, and CoTracker3.
Flow reversal steering can "steer" diffusion-based vision-language-action models (VLAs) through high-level actions, such as high-level actions inferred from vision-language models (VLMs). It also lets us run reinforcement learning (RL) in the diffusion noise space, with high-level reasoning guiding exploration: think about the task first, then practice! 👇
PKU EPIC Lab proposes KPGrasp, a dexterous grasp generation method that combines an all-Euclidean 3D hand-keypoint parameterization with a Transformer flow-matching model to learn grasp priors from large-scale data, without contact losses or contact-based test-time optimization.
Why it matters: The paper presents a new contact-loss-free grasp generation method and specific results on two benchmarks, letting readers compare it with existing methods and assess its approach.
PKU EPIC Lab released LIMMT (Less Is More for Motion Tracking), which it describes as the first data-centric study for physics-based simulated humanoid motion tracking. The method defines motion data quality along three dimensions: physics feasibility, diversity, and complexity. Training on less than 3% of AMASS achieves better tracking performance than training on the full dataset, and the team also cleaned web-sourced mocap data. The paper has been accepted to ICML 2026; see arXiv:2606.06953.
Why it matters: The paper presents a three-dimensional data quality framework and results from training on a small amount of data, offering a methodological reference for researchers working on humanoid motion tracking.
Through reinforcement learning post-training, EXPO-FT brought π0.5 to a 30/30 success rate on all 8 manipulation tasks tested, using an average of only 19 minutes of reinforcement learning data per task. The method targets challenging manipulation tasks, including routing string lights and inserting a power connector to light the string, potting a billiard ball, and placing a flower into a wine bottle. It is released as an efficient, reliable, open-source VLA fine-tuning method; the paper and video are at https://pd-perry.github.io/expo-ft/.
Zhang Wanruo, a 2021 direct-entry PhD student at Peking University's Open Laboratory of Intelligent Robotics, has had the paper "MXPose: Multiplex interactive learning for multi-view 3D Human Pose Estimation" officially accepted by Pattern Recognition (PR), a journal in the pattern recognition field. Her research direction is multi-view 3D human pose estimation.
This article is an interview with Dr. Jianlan Luo about his new work LWD (Learning While Deploying), which enables general-purpose robot policies to continuously perform online reinforcement learning from their own experience during real-world deployment. LWD combines offline RL pretraining with online RL fine-tuning to unify offline and online optimization objectives; the value function is trained separately with a VLM and uses both positive and negative samples (roughly over 60% positive and over 30% negative), and deployment data flows back into the training pipeline in about 40 seconds on average.
The Human Interactive Driving team at Toyota Research Institute (TRI) presented a paper at the AutoUI conference introducing a data-driven, user-centered scenario development method for studying the application of drift-capable control to road safety.
An EPFL research team published the Kinematic Intelligence framework in Science Robotics, enabling robots to safely transfer skills to robots with different configurations from a single human demonstration. The paper is available at http://dx.doi.org/10.1126/scirobotics.aea1995.
Genrobot has open-sourced the Phase 3 UMI dataset, containing 13K+ hours of dual-arm data that is ready to use and supports sub-1ms frame-level synchronization and high-fidelity trajectory reconstruction. The data was collected by 3K+ contributors in 10K+ real homes, covering diverse everyday manipulation skills. Download at https://huggingface.co/datasets/genrobot2025/10Kh-RealOmin-OpenData.
Why it matters: The data scale and real-home collection background help readers judge whether this kind of UMI data can be used to train their own manipulation tasks.
Sony AI's autonomous table tennis robot Ace competed under the rules of Japan's professional table tennis league, defeating multiple elite and professional players; the research was published in Nature (DOI: 10.1038/s41586-026-10338-5).
Why it matters: The article explains Ace's competitive record and three key technologies, and places it alongside the milestones of Deep Blue and AlphaGo, helping readers judge where physical sports competition currently stands.
Sanctuary AI demonstrated zero-shot in-hand manipulation in the real world with its proprietary hydraulic hand: a policy trained entirely in simulation continuously reoriented a lettered cube to a target pose ten times in a row without dropping it. The operation relied solely on the fingertips with no palm support, requiring simultaneous stable grasping and approach to the pose target. The original text states that the hand has a relatively high number of active degrees of freedom enabling finger abduction, and that hydraulic actuation and compact hydraulic valves offer a path toward industrial-grade humanoid dexterous manipulation.
Global views on robots are not uniform, and this is shaping adoption. 🌎 Where is trust highest? 🤖 What is holding people back? ⚙️ Who is ready, and who remains cautious? Attitudes are determining who acts first. Explore the data from our study: https://hxgn.biz/4rOBrpD
Teams including Berkeley AUTOLAB propose CaP-X, an open framework for systematically studying Code-as-Policy agents in robot manipulation, centered on the interactive environment CaP-Gym and CaP-Bench, which evaluates 12 frontier models.
Why it matters: The original text presents a systematic evaluation framework for Code-as-Policy agents and the finding that test-time scaling can mitigate abstraction dependence, offering reference value for research on coding-agent manipulation control.
Physical Intelligence proposes a new method called RL tokens (RLT), in which a vision-language-action model (VLA) outputs a special token to interface with a lightweight online reinforcement learning policy, improving the speed and success rate of fine manipulation phases with only a small amount of real data.
Why it matters: The original clearly explains the interface design between the VLA and lightweight online reinforcement learning and provides speedup data on four tasks, so readers can assess whether this type of method applies to their own fine manipulation tasks.
Ignacio Montoya, a US Air Force veteran and biomedical engineering doctoral student, reported years of self-study at the Society for Neuroscience annual meeting: transcutaneous spinal cord stimulation (tSCS) combined with adaptive task-specific training may reactivate dormant neural pathways 12 years after a clinically complete spinal cord injury. The training used Wandercraft's fully self-balancing, hands-free Atalante X exoskeleton, with robotic assistance gradually reduced; his gait consistency and muscle activation improved, and endurance, posture, and bladder and bowel sensory function also improved.
Major announcement! For whole-body humanoid robot teleoperation Project homepage: https://zhutengjie.github.io/CLOT.github.io/
Toyota Research Institute (TRI) reviewed eight years of polymer electrolyte research conducted by its AMDD project at MIT, producing about 40 papers and 7 doctoral dissertations and building a robotic experimentation platform that shortened tests that originally took about a month to completion within a day. The team also developed a multitask graph neural network to accelerate MD simulations and released the largest polymer simulation dataset, but the authors believe that AI can currently accelerate optimization within a class of materials, and that breakthrough polymers relying on new physical mechanisms still require human creativity.
Berkeley AUTOLAB proposes the STITCH 2.0 suturing robot system, with seven improvements including improved EKF needle pose estimation, new thread untangling methods, and an automated 3D suture alignment algorithm.
The paper "Generative Visual Foresight Meets Task-Agnostic Pose Estimation in Robotic Table-top Manipulation," a collaboration between SUSTech CLEAR Lab, LimX Dynamics, and the University of Hong Kong, has been accepted to CoRL 2025.
Why it matters: This research presents a concrete technical approach and experimental figures for learning manipulation from human video, helping readers understand the feasibility of training robot manipulation with video data.