Skip to content
#Safety and reliability
  1. arXiv cs.RO80

    Learning-augmented robotic automation achieves cable insertion and soldering on motor production line

    The study proposes Learning-Augmented Robotic Automation, a hybrid system that integrates learned task controllers and a neural 3D safety monitor into conventional industrial workflows to automatically perform flexible cable insertion and soldering on a motor production line. Using less than 20 minutes of real-world data per task, the system ran continuously for 5 hours 10 minutes without physical fencing, producing 108 motors with a 99.4% pass rate on product-level quality inspection, a takt time close to that of human workers with smaller cycle-time variation, and solder-joint quality comparable to that of skilled operators.

    Why it matters: The paper reports data from continuous operation of learning-augmented industrial automation on a real production line for 5 hours 10 minutes, with 108 motors and a 99.4% pass rate, serving as a practical reference for learned control entering manufacturing.

  1. NVIDIA62

    NVIDIA releases OpenShell reference architecture for building governed AI agents

    At GTC Berlin, NVIDIA announced that the OpenShell reference architecture is now available for building, isolating, observing, and governing secure AI agents on enterprise infrastructure and edge systems.

    Why it matters: The original text explains the reference architecture's layered design and millisecond-level isolation mechanism, which helps readers understand how it brings agent security down to the hardware layer.

  2. HuggingFace Daily Papers78

    Mechanistic Study of VLA Models: Not All Features Are Created Equal

    The study proposes an interpretability pipeline and finds that injecting task A's activations into the action expert causes pi0.5 to execute task A's motor trajectory in 99.6% of episodes (n=1,968), and that the injected program is bound to absolute workspace coordinates while ignoring the visible scene, offering a mechanistic explanation for the perturbation brittleness reported by benchmarks.

    Why it matters: The study uses intervention experiments to provide a mechanistic explanation for the brittleness of VLA policies, helping readers understand why models fail after slight scene changes.

  1. 3D视觉工坊 · 微信公众号76

    SenseTime and joint team evaluate GPT-6-Astra embodied capabilities, open-source SenseNova-RoboRSI framework and 27B model

    A joint team including SenseTime and The Hong Kong University of Science and Technology (Guangzhou) released the report "Has GPT-6-Astra Achieved Embodied Intelligence?", systematically evaluating GPT-6-Astra's embodied capabilities across 12 benchmarks, and simultaneously open-sourced the SenseNova-RoboRSI evaluation framework and a 27B embodied model.

    Why it matters: The report breaks model capabilities into testable conditions and provides harness ablation data, so readers can judge the real bottlenecks in embodied capability evaluation and cost.

  1. 人形大讲堂 · 微信公众号78

    Interpretation of the first six national standards for humanoid robots: quantified metrics, recommended status, and implementation gaps

    China's first six national standards for humanoid robots have been approved and released, all taking effect on April 1, 2027. They include GB/T 48316—2026, Technical specification for humanoid robot simulation test platforms, and GB/T 48317.1 to 48317.5, Technical requirements for humanoid robots, in five parts (general provisions, environmental perception, decision planning, motion control, and task operation). The standards are administered by the National Technical Committee 591 on Robots of Standardization Administration of China (TC591).

    Why it matters: The article breaks down the six national standards item by item to identify the metrics that are actually quantified, and, comparing them with conditions on production lines and in shopping malls, explains what they govern and what they do not.

  1. RealSense74

    RealSense D585 Pro On-Device Human Detection: Ranging Principles and Safety Architecture Usage

    RealSense explains D585 Pro on-device human detection: the detection model runs on its Gen5 Vision SoC and outputs JSON metadata at 30 fps with class, confidence, pixel box, and millimeter-level distance, without needing a host GPU.

    Why it matters: The article breaks down why on-device detection does not use neural-network ranging and provides a detection distance table for different postures, which can be used directly for safety zone design.

  1. CMU Robotics Institute62

    CMU and Siemens Partner to Propose PRISM, a Safety-Stop Monitoring Method for Humanoid Robots

    CMU and Siemens partnered to develop a method for evaluating the "safe stoppability" of humanoid robots, and used simulation to define the conditions under which a robot can be safely stopped. The result is PRISM (Proactive Refinement of Importance-sampled Stoppability Monitor).

    Why it matters: The article introduces PRISM, a method that uses simulation to define the conditions under which a humanoid robot can be safely stopped, and it is a useful reference for readers interested in humanoid robot safety monitoring and certification.

  1. Leiphone Robotics76

    AgiBot's 20,000th embodied robot delivered to Chimelong, with over 300 robots to be stationed at the park in the first phase

    On September 24, AgiBot's 20,000th embodied intelligent robot, the Expedition A3 Ultra, rolled off the production line and was delivered to Chimelong Group. In the first phase, over 300 AgiBot robots will be introduced at Hengqin Chimelong Spaceship Park, deployed across more than 100 interactive experience points and covering seven major scenarios including entertainment performances, guided tours, retail, and hotels. The collaboration also includes the implementation of a 5G-A embodied intelligence innovation application, a one-stop operations and maintenance management platform, and a security solution covering six major dimensions. The Expedition series offers up to 10 hours of battery life, and the robots support 360-degree omnidirectional perception and compliant control. The overall deployment scale is expected to reach several thousand units.

    Why it matters: The report documents the specific operational support for the large-scale, routine deployment of robots at the park, helping readers understand the real-world requirements of deployment.

  1. AgiBot67

    AgiBot and Chimelong launch large-scale deployment of over 300 robots, deliver 20,000th humanoid robot

    AgiBot and Chimelong Group launched their first large-scale embodied intelligence deployment at Chimelong Universe Park, with over 300 robots connected across seven scenarios: entertainment, science education, guided tours, retail, AI companionship, hotel services, and sports competition. The deployment includes a dedicated 5G-A network supported by China Telecom and China Mobile, multi-robot coordination, and safety operations support. The 20,000th robot to roll off the AgiBot production line was also delivered, and the two parties announced a joint institute for embodied intelligence in cultural tourism.

    Why it matters: The deployment covers seven scenarios and operational details such as the 5G-A network, offering insight into how embodied intelligence is deployed at scale in commercial cultural tourism environments.

  1. Wandercraft63

    Wandercraft's self-balancing personal exoskeleton Eve officially launches commercially in the US

    Wandercraft announced the official US commercial launch of its personal exoskeleton Eve, calling it the first FDA-cleared self-balancing personal exoskeleton that does not restrict the neurological level of spinal cord injury, allowing users to stand and perform daily activities hands-free without crutches. The access process includes clinical evaluation, a physician's prescription, individualized customization, and training; Medicare covers it under HCPCS code K1007, NSM supports delivery through more than 180 locations, and it is backed by a 16-participant training study published in JoVE.

    Why it matters: The original text is an official announcement of Eve's US commercial launch, explaining its crutch-free self-balancing capability and the access, reimbursement, and training process, so readers can understand the product's positioning and path to real-world adoption.

  1. Agility Robotics66

    Agility Robotics Releases Next-Generation Humanoid Robot Digit 5

    Agility Robotics has released Digit 5, a next-generation general-purpose humanoid robot built around a safety architecture that allows it to work in close proximity to people without traditional physical safety fencing. It uses an independent safety controller, AI-based human detection, and audiovisual cues, and is the first launch partner for the NVIDIA Halos for Robotics platform, built on NVIDIA IGX Thor.

    Why it matters: The official announcement details Digit 5's safety architecture, payload, battery life, and other specific upgrades, along with order background, giving readers insight into the product approach for humanoid robots working near people.

  1. Harbin Institute of Technology Robotics73

    HIT's Li Tianlong team publishes research in Science Advances on magnetically controlled CD5+ dendritic cell microrobot immunotherapy

    Li Tianlong's team at Harbin Institute of Technology (HIT) developed an engineered microrobot system based on CD5+ dendritic cells that can achieve precise intravascular navigation through external magnetic field control and activate immune cells in situ; the results were published in Science Advances. Under a conical magnetic field, the system self-assembles into a ribbon-like swarm that migrates downstream; under a rotating magnetic field, it assembles into a chain-like swarm that moves upstream against blood flow; combined with tumor chemotaxis, it achieves targeted enrichment and deep infiltration. In an orthotopic mouse colorectal cancer model, a single injection combined with magnetic navigation reduced the tumor bioluminescence signal by more than 90%. HIT is the first affiliation of the paper.

    Why it matters: The HIT team presents the construction strategy and magnetic navigation results of the CD5+ dendritic cell microrobot, giving readers a look at the specific approach of this therapeutic path of in situ immune activation.

  1. Harbin Institute of Technology Robotics65

    HIT's Wu Zhiguang team develops adaptive magnetically controlled nanorobot swarm platform, achieving navigation in the spinal canal of mice and non-human primates

    Professor Wu Zhiguang's team at Harbin Institute of Technology (HIT), together with collaborators, developed an adaptive magnetically controlled nanorobot platform that can form swarm motion patterns such as vortex, chain, and ribbon under external magnetic field drive, and achieved targeted navigation within the spinal canal in mice and non-human primates. The results were published in Science Advances.

    Why it matters: The research team released experimental results on navigation of magnetically controlled nanorobot swarms in non-human primates. Readers interested in the medical translation of micro- and nanorobots can learn about its technical approach.

  1. LeadeRobot77

    Mila and Polytechnique Montréal team open-source ~$2,000 robot guide dog Milo, built on Unitree Go2 for autonomous indoor and outdoor navigation

    Mila-Quebec AI Institute and a Polytechnique Montréal team have released Milo, an open-source robot guide dog that they say can cut the roughly $50,000 cost of a guide dog to about $2,000. The hardware base is a Unitree Go2 Air/Pro plus an NVIDIA Jetson Orin Nano; a magnetic-encoder telescoping handle is used to inversely solve the user's 3D position, a bird's-eye-view map is built first and then a navigation policy is run, and the policy is trained in about 10 minutes in GPU-accelerated simulation and deployed zero-shot on the real robot.

    Why it matters: The original article fully breaks down Milo's perception, policy training, and real-world testing details, so readers can reuse its engineering trade-offs in their own navigation projects.

  2. LeadeRobot79

    Italy's IIT and INAIL unveil ergoCub humanoid robot; human-robot collaborative lifting can reduce load on the human lumbosacral joint

    The Italian Institute of Technology (IIT) and INAIL proposed a shared embodied intelligence architecture and built the humanoid robot ergoCub; the research was published in Nature Machine Intelligence.

    Why it matters: The original article lays out methods for building ergonomics into robot hardware optimization along with real-robot data, letting readers understand how design and control are organized around human-robot collaboration.

  1. Deep Robotics/云深处科技60

    DEEP Robotics Lynx S10 prototype completes Arctic Ocean ice-surface and ice-water tests aboard research vessel

    DEEP Robotics' small wheeled-legged robot, the Lynx S10 prototype, completed a 76-day, 11,852-nautical-mile Arctic Ocean research test aboard Sun Yat-sen University's Polar Research Vessel Zhongshan, marking the first time a quadruped robot has set foot on Arctic Ocean ice. The unit, weighing under 20 kg including its battery, was fitted with anti-slip bionic feet mimicking polar bear paws in collaboration with Sun Yat-sen University, Westlake University, and Hangzhou Dianzi University, and completed crawling and motion-control tests with crampons; the body uses an IP67 integrated sealed design, and after increasing the surface area of its limbs for paddling, it completed swimming tests in ice water.

    Why it matters: The original article provides the specific test conditions and hardware modifications for the Lynx S10's walking on Arctic Ocean ice and operation in ice water, helping readers understand how quadruped robots are adapted for polar scenarios.

  1. Dexterity62

    Dexterity launches Instinct tactile intelligence system

    Dexterity announced the launch of Instinct, a tactile intelligence system that reads force at 500–2000 Hz and responds in 0.5–2 ms to perform force-guided bin packing on production lines, with a median placement time of 4.55 seconds and an overall team success rate of 81% (95% CI 75%–86%, sampled from 854 placements). Placements that cannot be completed are quickly identified and handed back to the Foresight world model to retry with a different angle or position; the same skill delivers consistent performance across different robotic arm hardware and generalizes to objects of different materials.

    Why it matters: The official release provides measured production-environment data and details of the force-guided manipulation architecture, which readers can use to understand how tactile skills coordinate with the Foresight world model for recovery.

  1. Fauna Robotics63

    Fauna Robotics exits stealth mode to launch Sprout humanoid robot platform

    Fauna Robotics has released Sprout, its first humanoid robot platform, shipping the same day in a Creator Edition aimed at developers, researchers, and creators building embodied intelligence applications.

    Why it matters: The company has released an official lightweight humanoid platform for developers, detailing its safety design approach, configuration list, and SDK capabilities, which can help readers assess its usability in shared-space scenarios.

  1. Wandercraft61

    Wandercraft Personal Exoskeleton Launches Pivotal Clinical Trial

    Wandercraft announced the launch of a pivotal clinical trial for its Personal Exoskeleton, evaluating the safety and effectiveness of this self-balancing personal exoskeleton in improving walking independence for patients with spinal cord injury (SCI).

    Why it matters: The original article provides the clinical trial location, enrollment criteria, and contact information, offering insight into a key milestone as this exoskeleton moves toward personal use.

You’re all caught up