D-Robotics is exhibiting for the first time at the 26th China International Industry Fair, showcasing Sunrises intelligent computing chips, the RDK robotics developer kit, and the RoboGo software platform, covering visual inspection, industrial data acquisition, and industrial embodied solutions, with a computing power lineup of 5–560 TOPS.
Relying on its self-developed 3D SLAM laser navigation, multi-sensor fusion positioning, and RMS fleet scheduling system, Xunji Zhixing has built a product matrix of outdoor handling robots spanning counterbalance forklifts, tow tractors, and custom AMRs.
Saisuode Intelligent released its new-generation industrial-grade embodied intelligent robot, SynRobot 03, at the 26th China International Industry Fair, held October 12–16, at booth 8.1H-A016. The robot features 18-DOF whole-body coordinated control, a combined dual-arm payload of 10 kg, repeat positioning accuracy of 0.1 mm, force-sensing sampling frequency of no less than 5 kHz, 2000 TOPS of AI computing power, binocular RGBD vision, dual-lidar SLAM navigation, and a 360° omnidirectional chassis.
The NeurIPS'26 Spotlight paper MoSE3 has been open-sourced; it can predict, in a feed-forward manner from in-the-wild monocular RGB video, dense SE(3) motion in the world coordinate frame for each pixel (rotation, translation, and rigid-body grouping).
MEgoVista reconstructs hand motion when an object occludes the hand: when the demonstrator naturally grips a cup and some fingers are blocked by the cup, the system can still extend the reconstructed hand into the occluded region. The method optimizes hand motion over the entire video, combining reliable visual observations with temporal continuity and anatomical constraints, thereby using information beyond a single occluded frame. For small teams collecting robot demonstration data, this means demonstrators do not need to change their grip to expose their fingers.
AIRclub-UdeSA released a ROS 2 bag recorded on a real robot for jar_workshops Week 10: a 2-minute capture in an arena filled with boxes, containing lidar, camera, odometry, IMU, and TF data. wget download and extraction commands are provided, and the empty-arena map comes with the Week 10 mapa_laberinto package.
A developer who has just spent four weeks getting started in robotics completed a differential-drive robot simulation in Gazebo and ROS 2 Jazzy, using SLAM for mapping and localization and Nav2 to autonomously navigate to target points in a maze map.
RealSense has released SDK 2.59.1 beta, which includes the built-in Ask RealSense AI chat assistant in RealSense Viewer and the web-based viewer, and supports on-device depth-to-color alignment.
Marvelmind demonstrates high-precision 3D positioning in indoor GPS-denied environments: a mobile beacon moves quickly along a 30×30 cm square, with positioning updated at 80 Hz and latency of 12–20 ms, achieving centimeter-level XYZ tracking. The system uses 3 fixed beacons plus 1 mobile beacon and, through ultrasound and IMU sensor fusion, raises the ultrasound update rate of about 8 Hz to 80 Hz; it is suited to indoor autonomous drones, mobile platforms, and industrial automation scenarios.
JAKA Robotics has launched the AL Agile series of vision cobots, comprising the A5L, A12L, and A20L models and covering applications from light payloads up to 20 kg. The series embeds an industrial camera and light source in the end effector, routes cables inside the robot arm body, and pre-calibrates hand-eye coordinates at the factory, eliminating on-site calibration. Image acquisition and six-axis motion planning run on the same self-developed controller platform. Eight vision functions are completed in a closed loop on a unified platform, with 2.5D spatial positioning accuracy of up to ±0.5 mm. It supports graphical and drag-and-drop programming and targets fields including 3C electronics, automotive parts, and metal machining.
Huaray Technology exhibited multiple machine vision products and solutions at the Vision Stuttgart 2026 trade fair in Stuttgart, Germany.
The author built a robot-simulation room twin using an Ouster OS0 scan and a 13-second iPhone video, producing 116,492 room points, 78 aligned frames, a Gaussian splat, and 17 objects annotated in 5 cm voxels.
A robotics developer asks OEMs and machine builders: when facing multiple product specifications, should they keep mechanical structures such as guide rails and feeders to ensure determinism, or simplify the mechanics and hand the variability to robot-plus-vision software?
A complete walkthrough of the e-skin bonding process: starting from a piece of Ecoflex, it covers material preparation, surface treatment, and adhesive selection across three layers, then mounting the e-skin on a robotic arm and getting it running, and finally validating the bond through bending, damp-heat, and peel tests to help developers avoid half a year of detours.
As a Gold Sponsor of ROSCon China 2026, Infineon is presenting four robot demo solutions, including a robot head solution based on the Real3™ iToF sensor IRS2976C (capable of recognizing objects ≥0.5 mm).
Hikrobot presented its machine vision product lineup at VISION 2026 in Stuttgart, Germany, from October 6 to 8, including the third-generation CT area scan camera, a 105-megapixel area scan camera, the SC3000X and SC5000X smart cameras, the ID5000X smart code reader, the DP2000 series laser profile sensor, and the RC series structured light 3D camera.
Zhejiang University and other institutions propose Pumpire, a unified benchmark that directly evaluates the ability of image-level and video-level 3D foundation models to estimate point-to-point metric distances in reconstructed geometry, noting that previous approaches of separately evaluating depth and camera intrinsics cannot reflect this ability.
px4-ros2-drone-nav releases v0.5.0, enabling flight in degraded visual conditions that rely on the vehicle's onboard light: when the light is unreliable or the charge is insufficient to return, it abandons the goal and heads home; when completely blind, it performs a level blind descent at 0.5 m/s on the IMU and lands.
ROMA proposes a real-world object-centric multi-sensor active perception system, in which an LLM planner determines missing information, actively collects it, and provides attribute estimates once evidence is sufficient. Unlike fixed pipelines that perceive first and then act, the system uses real multi-sensor hardware, and its code has been open-sourced.
AutoTrust AI Lab has open-sourced JEV-27B-VL, a multimodal decision model built on Qwen3.8-27B that extends System 1 decision-making and System 2 reasoning to image inputs, under the Apache 2.0 license.
On October 8, the US Patent and Trademark Office published Tesla patent application US 2026/0310299 A1, covering a three-dimensional flexible tactile sensor array with multimodal sensing and scalable manufacturing methods. The patent presents two manufacturing paths: first printing conductive areas on TPU, PET, or polyimide film and then thermoforming, or first forming the curved surface and then printing directly; overlapping conductive areas form independent sensing pixels, covering both capacitive and resistive sensing modes, and it states that when there are enough pixels, object type can be detected by touch alone, with target applications extendable to the robot's whole body.
Kingston University has been selected to lead the UK's first waste and recycling robotics adoption hub, NORAH, as part of the government's 40 million pound (approximately 53 million USD) robotics adoption program. The NORAH project is funded with 3.6 million pounds, of which Kingston University receives 1.5 million pounds, funded by Innovate UK.
FANUC America will demonstrate AI and collaborative robots for packaging and palletizing at this month's Pack Expo International 2026. The demonstrations include the CRX-20 i A using generative AI to convert natural-language instructions into Python code and robot actions, dynamic proximity monitoring on the CRX-10 i A/L, three collaborative robots integrated into an end-to-end packaging process, and heavy-payload palletizing with the 40 kg payload CRX/30-18A.
In this episode of the podcast, Claire talks with Kushant Patel of Brookhaven National Laboratory about robots that work in scientific facilities such as particle accelerators. Patel develops software that helps robots and instruments perceive, map, and move through the physical world. He previously taught quadruped robots to map and inspect nuclear materials at UC Berkeley, and now develops control systems for advanced synchrotron light beamlines at Brookhaven.
A team led by Professor Sejoon Lee at Dongguk University in South Korea combined a triboelectric nanogenerator (TENG) with a flexible graphene-channel ionic-gel-gated transistor to create a flexible device that needs no battery, harvests energy from human movement, and enables neuromorphic sensing and learning.
AMR robots rely on four onboard technologies for trackless navigation: laser SLAM mapping, real-time LiDAR perception, dynamic obstacle avoidance, and real-time path replanning, with no need for floor tracks, magnetic strips, or QR codes. REEMAN delivery robots achieve laser navigation accuracy of ±5 cm, with an optional reflector code accuracy of ±3 cm; the Fly Boat model achieves QR code accuracy of ±2 cm, and a single map covers about 40,000 square meters. Its product line covers payloads from 100 kg to 1.5 tons, and since 2015 it has delivered to more than 55 countries and regions.
An 18-day joint Western Pacific hydrothermal expedition by Tsinghua University, the Qingdao Institute of Marine Geology, Shanghai Jiao Tong University and other institutions conducted seven dives and identified a new large high-temperature active hydrothermal field, preliminarily confirming its potential for large-scale polymetallic sulfide resources, with associated gold content of up to 15.4 ppm and associated silver content of up to 1,271 ppm.
Helm.ai announced it has signed $70 million in commercial contracts over the past 12 months, with customers including global automotive OEMs, Tier 1 suppliers, and industrial automation companies, and said the company is nearing break-even. Its foundation models use an unsupervised "deep teaching" method that hierarchically decouples perception and decision-making to improve data efficiency, and the technology has been used in SAE L2 to L4 autonomous driving, mining, construction, and delivery drone projects.
Leiphone's Zuo Lin You Li channel interviewed Benewake founder Li Yuan, reviewing his transition from optoelectronics scholar to entrepreneur. In the first half of 2025, Benewake delivered 250,000 LiDAR units for robots, with robotics business revenue accounting for more than half of the company's revenue; this revenue grew 100% in 2025 compared with 2024.
The CMU Robotics Institute's Howie Choset team provided analysis and modeling support for Northrop Grumman's Mission Robotic Vehicle (MRV). The new docking technology inserts a probe into the target satellite's liquid apogee engine nozzle to complete docking, and has been successfully executed multiple times in orbit. The MRV launched from Cape Canaveral in July this year and, in addition to refueling, can also support in-orbit repair, component replacement, and upgrades.
Overview AI has released the OV Spark (1.2MP) and OV Spark Pro (5MP) AI inspection cameras, featuring the built-in agent Sparky, which uses natural language to recommend inspection tools, retrieve snapshot records, and generate operator interfaces. The cameras come with ten built-in inspection tools; model training and inference both run on-device. The unit has an IP66 sealed design and measures 70 x 116 x 53 mm, and has been deployed for end-of-line inspection at Mansfield Engineered Components.
Inbolt announced it has raised $12.5 million, bringing its cumulative funding to $34 million. The round was led by new investor Shift4Good, with participation from Bridges Climate Transition Partners, BNP Paribas Développement, and Ora Global.
CMU researchers designed human-hand-inspired fingertips that can be mounted on the DeltaHands flexible robotic hand. They combine strain gauges to measure pinch pressure with a microphone that captures micro-vibrations through a plastic fingernail, enabling the hand to count paper cups, identify materials, and grasp fragile items such as tofu cubes and potato chips. Each fingertip costs less than 100 US dollars, compared with about 800 US dollars to build a similar setup from off-the-shelf products. The work was presented at the 2026 Intelligent Robots and Systems Conference in Pittsburgh.
Why it matters: The original text provides details on the multimodal tactile design of the human-inspired fingertips and their cost of less than 100 US dollars, helping readers understand how force and vibration sensing support manipulation control.
During the National Day holiday, DEEP Robotics deployed its DR02 humanoid robot in real traffic scenarios in Hangzhou's Xihu District for real-world testing, covering functions such as intersection traffic direction, multilingual tourist inquiries, and identification and persuasion of traffic violations. The DR02 features IP66 protection, enabling it to navigate autonomously and dynamically avoid pedestrians and vehicles in outdoor environments such as sustained rainfall. DEEP Robotics said it will continue to advance the robot's application in more scenarios, including traffic management and public services, based on real business needs.
Almost all dual-arm robots are equipped with wrist cameras, but these add bulk, cost, and complexity. EyeRobot 2.0 instead uses an active gaze approach with a head-mounted stereo camera to replace wrist cameras. —Do dual-arm robots really need wrist cameras? Maybe not!
The IN²Bot team from the School of Automation at Beijing Institute of Technology won championships in embodied intelligence competitions at two major international conferences: it took first place in the CVPR 2026 ManiSkill-ViTac Visuotactile Embodied Dual-Arm Manipulation Challenge with a 700-point lead, using a solution that includes a self-developed tactile encoder and an action memory module; and it won the ICRA 2026 Fifth Mobile Robot Autonomous Navigation Challenge with the highest score overall, using methods covering a spatial traversability cognitive reasoning field, local reactive planning, and safety-barrier trajectory control.
PaXini hosted the Global Embodied Ecosystem Night (GEEN) dinner during IROS 2026, inviting nearly 200 scholars, industry leaders, and investors, and supported the conference as its highest-level sponsor.
As the highest-level sponsor and partner of IROS 2026, PaXini presented its full product lineup at the conference, including sensors, dexterous hands, PXCAP series data-collection gloves, and complete humanoid robots, demonstrating its "PIE" (Perception-Intelligence-Execution) full-stack closed-loop system.
IROS 2026 was held in Pittsburgh, USA, from September 27 to October 1, with papers, workshops, and exhibition discussions focusing on how touch is entering robot learning, fine manipulation, and real-world applications. A paper from a University of Chinese Academy of Sciences team achieved a 67% success rate in a 0.05 mm clearance insertion task, reducing peak interaction force by 60% and torque by 44%; the experiments used PaXini PX6AX GEN3 sensors.
On September 30, the first "zone-port linkage" drone sealing project in Hunan Province, jointly developed by Sinotrans and China Merchants Communications Technology, went live and completed live-container verification at the customs supervision operation site of the Changsha Zhongnan Consolidation Center. A drone carrying an electronic customs lock reader flew to the designated container slot; its onboard camera automatically identified and verified the container number and slot via a visual large model before completing sealing. The system then automatically submitted the arrival declaration to Xingsha Customs and received the receipt back through the Changsha Single Window, with the entire process digitalized.