The ML3DHS 3D hierarchical semantic segmentation framework, developed by Assistant Professor Zhao Na's team at the Singapore University of Technology and Design (SUTD) with collaborators from Southwest Jiaotong University, was presented at the 43rd International Conference on Machine Learning (ICML) in 2026, and assigns each 3D point multiple related labels at once, from "furniture" to "chair" and "table".
Stial has launched an AI flexible grinding solution for grinding weld seams on welded boxes, achieving uniform removal of weld seams, smooth transitions between welds and base material, and no damage to thin-walled base material. The solution uses the self-developed SeeV vision system to locate weld seams, and NextBrain™ AI Mushroom Cloud computing to calculate grinding allowance and generate grinding paths. After digital twin simulation verification, a six-axis robot works with the PolishX flexible force control system to execute the grinding, with force control accuracy of ±0.5%FS and output force as low as 4N, suitable for energy storage, electrical, new energy battery housings, and other multi-category box scenarios.
STIAL Robotics recently won an AI flexible grinding project from a leading company in China's rail transit sound barrier industry. The grinding targets sound barrier columns up to 9 meters long. In the solution, the SeeV vision system uses 3D line laser scanning to extract weld seams, and NextBrain™ AI automatically matches grinding trajectories and force control parameters. The end effector is equipped with the PolishX flexible force control system, achieving contact force control accuracy of ±0.5% FS, capable of removing 2mm weld spatter residue and adapting to changeover between multiple column specifications.
Mech-Mind will present 20 intelligent robot applications at the core of the robotics exhibition area at the 26th China International Industry Fair, held October 12–16 at the National Exhibition and Convention Center in Shanghai. Booth 7.1H-A225, with 300 sq m of exhibition space. On-site demonstrations will include dual humanoid robots working together on loading/unloading and bin handling, sorting of massive numbers of objects, and generalized grasping of transparent objects, covering more than ten industries including automotive, new energy, and 3C/semiconductors. Its general-purpose eye-brain-hand systems have been deployed more than 30,000 units globally, serving 100+ Fortune 500 customers.
At IROS 2026, under the theme "From Touch to Intelligence," Daimon Robotics publicly demonstrated its tactile intelligence infrastructure on real hardware for the first time, covering three layers: the DM-Tac series of vision-based tactile sensors (more than 12 tactile modalities and over 110,000 sensing units), the Data-Nexus data collection system, and the Daimon-TWM world model released in August.
Jinrong Robotics launched its new vision-based welding product "Zhihanxing" on September 29, using the welding-specific 3D camera from Weijing Intelligent's Starlight Eye series. The camera offers high-precision, high-frame-rate 3D imaging, resists interference from strong light, reflections, light-absorbing surfaces, and arc light, supports combined dynamic and static modes, and reduces hand-eye calibration through zero-error positioning at the coordinate origin. The system can preset photo positions, then use vision to locate the workpiece and weld seam before welding, reducing reliance on fixed positions and preset trajectories.
Professor Hong Chul Moon's team at KAIST published in Advanced Materials on Ionograsper, an ionic soft material inspired by the Venus flytrap that integrates sensing, actuation, and shape retention into a single material.
The 2026 Third Smart Driving "Fengshen Bang" Challenge was held in Changsha. JD Logistics' "Lone Wolf" 6th-Generation Smart Delivery Vehicle Plus Edition won the "Best All-Around Performance Award" with the highest overall score, achieving the best performance in the field across multiple scenarios in its first appearance and passing all difficult scenarios with full marks.
IROS 2026 will be held in Pittsburgh, USA, from September 27 to October 1. Mair Microvision will exhibit 3D visual perception solutions for mobile robots at Booth 436. The company's core technical team was formed in 2016 and focuses on vision sensors dedicated to mobile robots. It has delivered more than 10,000 3D vision sensors cumulatively, supporting autonomous navigation robots and embodied intelligence applications with its 3D+AI deep vision system.
The EBO SE 2 FamilyBot is a small indoor robot from Enabot that moves freely around the home, records video, and lets users interact with their pets remotely through an app; it is recommended as a Halloween gift. It features a 2K (2304×1296) camera with a 137° field of view, supports 24/7 recording and motion alerts, runs for 4–7 hours before automatically recharging via route memory and infrared docking, can climb over 6mm obstacles, and can slip under low furniture as low as 9cm.
Hanwei Technology Group has launched a full-scenario gas safety solution for the semiconductor industry, covering specialty gas monitoring, cleanroom AMC micro-contaminant monitoring, system integration, dynamic inspection, and platform operations and maintenance. The GD-HS10 gas detector offers detection precision down to the ppb level, with an optional GD-PU10 pyrolysis unit enabling detection of hundreds of electronic specialty gases; the trace water cavity ring-down spectroscopy analyzer can achieve quantitative measurement at the ppt level. The solution also includes the GMS programmable control system, the FGS fire and gas alarm control system, and an intelligent inspection robot equipped with a Fourier-transform infrared scanning imaging system, with data transmitted back to the semiconductor facility intelligent operations and maintenance system.
A University of Florida study found that simple black-and-white stripes or checkerboard patterns can cause errors in the stereo camera depth estimation of autonomous vehicles and robots, leading them to misjudge obstacle distances and even trigger sudden braking or collisions. The research team tested a variety of sensors, algorithms, and AI models and found that this fundamental vulnerability, which is based on repeating textures, is widespread, and that an attacker need only place a small-area pattern in a scene to affect depth computation; such patterns may also appear in natural environments.
On September 22, the embodied tennis robot TENNIIX ULTRA MAX, jointly developed by LimX Dynamics and Inhans Power, appeared at the Billie Jean King Cup Shenzhen Finals, where it engaged in live, real-time rallies with tennis star Wang Qiang and completed multiple rounds of human-versus-machine play.
The article analyzes the trend of warehouse inventory drones moving from PoC to multi-warehouse replication, noting that their core value lies in increasing the frequency of inventory data generation rather than simply replacing manual labor. It lists different technology routes including Corvus One, Switzerland's Verity, Gather AI, France's EYESEE, Germany's doks. innovation, Singapore's Infinium Robotics, and the US's B GARAGE, as well as Corvus's application data in cold chain, lights-out inventory counting, and audit compliance scenarios.
A team led by Assistant Professor Zhang Shuyang at Shenzhen University's College of Mechatronics and Control Engineering used the PaXini PX6AX multi-dimensional tactile sensor as its validation platform to address cross-operator generalization in tactile material recognition, proposing the lightweight tactile model Lite-TacFormer, which improves cross-operator generalization through a few-shot calibration mechanism and features an extremely small parameter count, minimal sample dependence, and a very low computing threshold. Built around the self-developed FUSE tactile sensing chip, the PX6AX maintains millinewton-level accuracy and microsecond-level response after tens of millions of force measurements, and its integrated sensing-computing-transmission design shortens experiment cycles.
Jiuwei Technology officially released its full-stack autonomous navigation system Jiuwei Zhihang MIC Navi on September 22. The navigation code runs directly on the robot dog's mainboard, requiring no additional add-on computing kit, and upgrades are completed via OTA through the MICBOT MOVE app.
The East China Sea voyage organized by Tsinghua University in collaboration with the Qingdao Institute of Marine Geology has returned successfully. Using the independently developed deep-sea controllable visual sampler (DCVS), the team completed dive operations at 7 stations in areas including the Xianchi cold seep area and the Qiaoyue hydrothermal field, with a maximum dive depth of 1260 meters. DCVS operated stably under extreme conditions such as high temperature and high pressure, collecting multiple physical samples and acquiring large amounts of high-definition seabed imagery. The equipment was developed under the leadership of the Department of Automation at Tsinghua University, with participation from Shanghai Jiao Tong University and Xiamen University.
STIAL has launched an AI flexible grinding solution for grinding weld scars, weld seams, and burrs on welded enclosures, covering sheet metal scenarios such as energy storage enclosures, computing power cabinets, and electrical enclosures.
Swinging. Turning. Jumping. @GenrobotAI's DFM converts six-camera ego videos into continuous full-body mesh and skeleton data, maintaining body structure and temporal consistency even during fast motion and complex poses. See the whole body. Learn the whole skill set. #GenRobot #WholeBodyData #egodata #PhysicalAI #Robotics
Under the guidance of teachers Zhang Wenqiang and Zhao Qing, a Fudan University student team of Wang Jianan, Li Jinghui, and Jiang Kaixun won a team first prize in the business plan special track at the national finals of the 21st China Graduate Electronics Design Contest (held in Wuzhen, Tongxiang, Zhejiang) for their project "Fangwei Dudian: the world's first industrial-grade embodied intelligent advanced power operation and maintenance equipment."
JAKA Robotics presented multiple applications at the 2026 International Manufacturing Technology Show (IMTS) in the United States, covering scenarios such as loading and unloading, grinding and polishing, palletizing and beverage packaging. The JAKA Zu12, integrated with JAKA Lens series vision cameras, can automatically identify and pick randomly stacked incoming materials; the JAKA Mini performs loading and unloading of phone modules on drive shafts, and the JAKA Zu7 carries out grinding and polishing of curved workpieces. Its solutions have served more than 100 countries and regions worldwide.
Bend. Squat. Move. DFM can reconstruct continuous full-body meshes from egocentric video, even with partial occlusion. Average full-body error is about 3 cm. See the Whole Body. Learn the Whole Skill. #GenRobot #WholeBodyData #ego data #PhysicalAI #Robotics
RealSense has introduced Min-Z, an improved close-range depth feature in Perception Studio that extends usable stereo depth to about 2 cm on D401/D405 and about 12 cm on D43x/D45x without changing cameras.
Why it matters: The original explains how Min-Z extends the near-field usable range of stereo depth and the trade-offs between two deployment paths, which can be used to evaluate grasping and close-range perception solutions.
In an interview, Xspark AI co-founder and chief scientist Ding Wenbo proposed that vision has already solved about 99% of robot perception problems, and that touch should form a low-latency, body-adjacent "spinal cord"-level model rather than being directly integrated into vision large models. He proposed a three-layer architecture of "slow brain VLM—fast brain VTLA+TWAM—spinal cord VTA," arguing that touch mainly serves collision-avoidance safety and post-training for dexterous manipulation, and noted that current bottlenecks lie in sensor consistency, tactile data representation, and unification across sensors, modalities, and embodiments.
EBKernel released its brain-inspired cognitive world model Cog-WM 1.0 on September 14 at the Brain-Inspired Embodied Intelligence sub-forum of the Shanghai Pujiang Innovation Forum. It includes a navigation branch, Cog-WM Nav 1.0, and a manipulation branch, Cog-WM Manip 1.0, and uses memory, prediction, and active exploration to help robots understand the physical world.
Junao Panshi released the brain-inspired cognitive world model Cog-WM 1.0 on September 14 at the Brain-Inspired Embodied Intelligence sub-forum of the 2026 Pujiang Innovation Forum, covering the navigation branch Cog-WM Nav 1.0 and the manipulation branch Cog-WM Manip 1.0.
Unitree Robotics announced that the G1+ has completed 6 major upgrades across three dimensions: motion performance, perception and interaction, and intelligent experience. The upgrades cover these three areas; the original text did not provide specific parameters or real-machine verification results.
Hikrobot appeared at the 2026 Tokyo International Logistics Exhibition (LTT), showcasing three major product lines: machine vision, mobile robots, and articulated robots. Exhibits include the STRP case storage solution, which supports double-deep storage up to 12 meters and handles 200 cases per aisle per hour; the F1-300 forklift robot with a rated payload of 300 kg; and the PlantMirror 3D logistics simulation software.
AgiBot has released AGILE 2.0, a second-generation perception-driven locomotion and whole-body control framework that brings visual perception into the motion control loop and introduces a unified whole-body motion representation, supporting manipulation while moving, real-time replanning, and cross-robot coordination. In the demo video Acrobat BOT, the AGIBOT X2 completes tasks such as ball balancing, three-robot jump rope, and collaborative carrying; the framework also supports risk perception and autonomous recovery from disturbances in human-robot shared environments.
🐾🎥 The ultimate dog's-eye view experience Achieve hands-free low-angle tracking shots with Beni
PKU EPIC Lab proposes UniPart, a feed-forward cross-modal 3D Transformer conditioned on CLIP text embeddings that uses free-text phrases to select functional parts on point clouds, achieving zero-shot language-driven 3D part segmentation.
Why it matters: The paper presents a text-driven 3D part segmentation method and a self-built dataset, and validates transfer to real-world language-conditioned grasping.
The paper "GeoVLA: Empowering 3D Representations in Vision-Language-Action Models," jointly published by RealForce and Tianjin University and Tsinghua University, has been nominated for the IROS 2026 Cognitive Robotics Best Paper Award.
PaXini AI, drawing on the tactile sensing logic of the human Pacinian corpuscle, has released the PX6AX GEN4 product matrix, equipped with the GEN4 FUSE true 6D tactile sensing chip, and completed a new funding round of 1 billion yuan.
Tennant's X4 ROVR and X6 ROVR autonomous floor-cleaning robots use RealSense depth cameras as their core perception component, achieving millimeter-level depth accuracy and real-time 3D perception for obstacle detection, recognition of people and moving objects, environment mapping, and safe navigation.
Boston Dynamics has released Spot 5.2, positioned as a middle layer connecting the various AI systems in a factory. The Orbit software has been restructured to organize data around assets, can import asset hierarchies and link them to inspections, supports connecting external data sources such as PLC sensors and security cameras, and adds an MCP layer for AI agents to autonomously schedule Spot deployments.
Why it matters: The original article details the specific changes in Spot 5.2 regarding data connectivity, inspection modalities, and predictive maintenance, allowing readers to assess whether to update their own deployments.
At WRC 2026, Youren Intelligence made the global debut of a robotic dexterous hand that can perform pulse diagnosis: three fingers rest on the wrist, and on-device processing analyzes 28 pulse types and draws on traditional Chinese medicine knowledge to offer suggestions. The angel-round company, founded less than a year ago with a team of about 20 people, believes the tactile bottleneck lies in data trustworthiness rather than volume. Its CrediTac framework uses heterogeneous fusion sensors based on multiple mechanisms—capacitive, piezoresistive, and piezoelectric—and aims to deliver tactile data with 99.9% trustworthiness.
iRobot showcased the Roomba Duo concept at the IFA trade show in Berlin. The main floor-washing unit navigates using lidar and cameras, its steam-mopping roller brush applies 46 newtons of downward pressure, and it carries a Roomba Plus 757 on top, which a robotic arm lowers to clean under tables and in corners. In the demo, the main unit picked pasta up from a carpet and cleaned the carpet. New CEO Hao Li introduced the machine; as a concept product it has no price or launch date, and its specifications are still in development.
The D-Robotics community released a full-pipeline PyTorch vision model image for RDK X5 and RDK S100/S100P/S600, covering training, validation, Model Zoo ONNX export, BPU quantization, and online inference. On 2026-09-30, the S series will be upgraded to v1.1, adding YOLO26 detection and YOLOv8/YOLO11/YOLO26 instance segmentation and pose estimation.
Mondo Robotics' all-terrain camera robot Beni follows with one tap and autonomously shoots 4K footage, with a top tracking speed of 18 mph and about 1 hour 45 minutes of battery life per charge. More than 3,000 households and creators have backed the project, which raised over $2 million within two weeks of launch; assembly begins in September 2026, with shipping in the fall. All footage is stored locally, and the robot can right itself after a fall.
Huan'er Technology details the complete ROS 2 agent pipeline of the ROSOrin Pro robot: voice commands are recognized by WonderEcho Pro, then passed to the OpenClaw gateway for understanding and broken down into subtasks such as navigation, detection, and grasping.