The RDK X5 robot duck project completed full assembly and a short-distance real-machine walking demo on September 29, and open-sourced complete engineering materials covering everything from printing and assembly to real-machine integration and debugging (GitHub repository rdk-x5-microduck).
In 1921, Czech playwright Čapek first brought the word "Robot" to the public eye in his play R.U.R. (Rossum's Universal Robots); in 1942, Asimov wrote the Three Laws of Robotics in his novel Runaround.
As a robot platform support partner for the final stage of the 2026 Huawei Cloud CloudRobo Embodied Intelligence Competition, Galaxea provides participating teams with the A1Z robotic arm and related technical support, helping developers complete VLA model deployment, robot control, and real-world task validation. Meanwhile, the G0.5 reproduction and open-source co-creation challenge is ongoing, offering developers simulation reproduction, real-robot deployment, and open-source co-creation.
The Cell and BioRobotics Group (CBoRG lab) at the Technical University of Munich (TUM) Munich Institute of Robotics and Machine Intelligence (TUM MIRMI) proposes "Embodied Laboratory Intelligence," in which robotic assistants and virtual agents work together to explore the internal mechanisms of cancer cells.
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.
Qiuzhi Technology, together with Qihang Information Technology and Xuanji Dynamics, integrated the AIRBOT Play six-axis robotic arm with the Qihang Insight Cloud OS teleoperation system and mounted it on Xuanji Dynamics' quadruped robot platform, building a "mobile + manipulation" inspection solution that was validated in three scenarios: sanitation inspection, construction site operations, and hazardous-material simulation experiments.
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.
The RLChina "Seven Questions" column examines DreamerV3, a general reinforcement learning algorithm proposed by Google DeepMind and the University of Toronto. Based on a world model, the algorithm first trains an actor-critic on imagined trajectories, and with a single set of fixed hyperparameters it matches or exceeds domain-specific algorithms across 8 major categories and more than 150 tasks.
Dexbotic releases Update Brief (1), announcing that its robot VLA development chain is now connected and that it supports fine-tuning for multiple models. Readers can click "Read Original" to jump to the corresponding GitHub repository to get the code.
Seohong Park published a new blog post discussing the "puzzles" of behavior cloning on real robot data (such as overfitting actually being a "good thing"), attempting to demystify them and share his own thoughts. Seohong wrote a novel; you won't believe who the murderer is.
ATDev has updated progress on its autonomous wheelchair development, building a modern power wheelchair equipped with a collaborative robotic arm under the ARPA-H-funded RAMMP project. The wheelchair uses a shared autonomy model in which the user decides the path and the system handles fine adjustments, with plans to add teleoperation and federated learning features. ATDev is working with Meta to apply AI models to the wheelchair, with both parties betting on open source.
Persona AI collaborated with cinematographer Pekka Varis to convert CAD data into USD format within 30 days, using motion capture data and Isaac Lab joint motions on a Dell Pro Max Tower T2 workstation (NVIDIA RTX PRO Blackwell 6000 GPU).
Westwood Robotics releases the fourth installment of its BEAR actuator tutorial series, walking through the control table register architecture via the BEAR Data Control GUI: CONFIG registers are used to set PID gains, baud rate, hardware limits, and safety thresholds, while STAT registers manage runtime states and telemetry feedback.
Westwood Robotics' tutorial walks through the wireless E-STOP setup and actuator logic for BEAR actuators. It covers the 4 health states of BEAR actuators, how Hardware Damping Mode prevents runaway damage, and step-by-step setup of the E-STOP Kit (Receiver, Adapter, Remote) along with the correct use of the two power options, USB-C and EXT_3V3.
The ROS Summer School "Embodied Intelligence VLA Bootcamp" organized by Qiuzhi Technology concluded at Zhejiang Sci-Tech University from July 20 to July 23, with more than 40 trainees from universities, research institutes, and enterprises taking part. The course used the AIRBOT Play six-axis robotic arm as its hands-on platform; participants completed motion planning with MoveIt2 and ran through the Real2Sim2Real pipeline using DISCOVERSE 3DGS rendering and MuJoCo simulation, then carried out data collection, model training, and deployment inference on the real robot, finishing with a group task challenge.
The key to the NexArm robotic arm's smooth start and stop at a 500mm reach is its high-precision magnetic-encoder bus servos. The 12-bit magnetic angle sensors in the core joints provide 360° absolute angle detection and 4096-step position resolution, and their contactless measurement avoids wear.
Agile Robots and its subsidiary Idealworks describe their edge AI deployment stack since 2018, covering five layers: hardware, BSP, inference pipeline, inference framework, and quantization.
Why it matters: Using their own Xavier deployment experience as an example, the authors present a reusable TensorRT plugin method for running VLMs on older hardware that peers can directly draw on.
So how do you train robots? Great question. Here's an insider-view video: https://youtu.be/oRymQyooWjE?si=hoFNf7DyBv8ZQSFA
Wuba Intelligent appeared on the first day of the 2026 World Artificial Intelligence Conference, exhibiting a humanoid robot concept designed for high-risk environments and a complete remote control system for the Tianlang series of quadruped robots. The humanoid robot performs cross-space operations using "VR technology + intelligent cockpit beyond-line-of-sight remote control." At the event, an operator remotely controlled a robot located in a Hangzhou laboratory to complete indoor search-and-rescue demonstrations, including pushing open a gate, pressing the button to shut down a hazardous device, and traversing a 60cm ultra-narrow passage. The solution targets three major pain points: high latency in remote control, compliant manipulation, and image stitching.
RoboParty announced the establishment of RoboParty Lab and open-sourced three humanoid robot toolchains: the supervised learning training framework MimicLite, the unsupervised reinforcement learning control framework UFO, and the motion retargeting tool hhtools.
Why it matters: The three toolchains provide specific figures on training compute, time required, and the policies already integrated, which readers can use to assess their reuse value in their own R&D workflows.
The RoboParty 'LuoBo Top Player' summer camp officially opened on July 6, running for 8 weeks and 40 working days, centered on hands-on practice with real robotics projects.
Mobile robot chassis selection should be evaluated across six dimensions: terrain adaptation, drive and steering architecture, core hardware reliability, modular expansion, software ecosystem, and mass-production delivery, using the AgileX Robotics UMR four-wheel four-steering omnidirectional chassis as a reference.
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.
Shima Notes talks with Wen Bowen, an NVIDIA researcher and first author of FoundationStereo and Fast FoundationStereo, about how stereo depth foundation models are changing robotic 3D perception.
Noble Machines published an article introducing its three-layer robot learning stack for industrial autonomy, along with on-site test results from GTC 2026: over four days, the same model simultaneously drove multiple Moby robots to complete two autonomous tasks at the NVIDIA booth and its own booth.
The evSense system on the Strutt ev¹ provides obstacle detection and assistance features such as Co-Pilot, Co-Pilot+, Waypoints, and Pathfinder. Co-Pilot+ can steer around obstacles along a set direction, but none of these features is autonomous driving; the driver must be ready to take over with the joystick at any time.
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.
Strutt has launched the Resource Hub, offering tutorial videos, setup and travel PDF guides for the ev¹c personal smart mobility device, along with full product specifications (including dimensions, motor performance, and range data) and open-source 3D file downloads for space planning and accessory customization.
Did you see Jensen Huang's keynote at GTC 2026? During the presentation, Jensen Huang had a fascinating conversation with Olaf. But have you ever wondered what makes Olaf move so smoothly and precisely? It's all thanks to actuators! Top-tier robotics technology? It's all here. Have any questions? Join our Discord! https://discord.gg/robotis
1X Technologies disclosed that the AI development process for its humanoid robot NEO is based on NVIDIA's robotics platform, covering the open-source physical AI agent tools released this week at GTC Taipei.
The Toyota Research Institute (TRI) published a tutorial article introducing a statistical A/B testing method based on success/failure labels for robot policy evaluation.
Why it matters: The TRI authors systematically explain statistical pitfalls and reusable tools in robot policy A/B testing, with runnable Python examples.
Pollen Robotics has released an assembly guide video for the Reachy Mini LITE, demonstrating the full build process in three sections: foot, body, and head. In the body section, the 6 motors must be wired in sequence, with cables clipped into the base and routed through the head PCB; after checking rotation, the Tricap and head shell are secured. The head section covers installing the fisheye lens, glasses assembly, and antenna motor, then connecting the speaker, motors, power, and USB extension cable to complete the build.
Addverb Technologies proposes five core indicators—order volume and speed, return on investment, technological progress, competitive landscape, and seasonal fluctuations—to help companies judge the right timing for warehouse automation investment. The article notes that rising labor costs, surging e-commerce orders, and the need for 24/7 operations are pushing companies to adopt automation, and that AI can support real-time data analysis and predictive maintenance. It mentions that DHL's collaboration with Addverb is transforming warehouse operations by improving speed, accuracy, and scalability.
Tianlian Robotics has published a video imitation learning training pipeline: extract reference motions with GVHMR/TRAM (or use motion capture datasets such as AMASS), then after human-to-robot joint mapping and training a motion orientation model, use PPO reinforcement learning in simulation with curriculum learning to progress from static poses to long-sequence complex motions, then optimize through domain randomization before deploying to the real robot, with fine-tuning via residual reinforcement learning or adaptive gains. The company says it has achieved key breakthroughs in dynamic force-control precision and stability, and that motions can also be reproduced in real robotic arm operation.
Want to learn more about programming social robots? 👩💻🤖🐍 On the Misty Lessons platform, you can now learn how to program Misty II with Python and use all of her features 🚀 Visit: https://buff.ly/3vIoYwA #computerscience #robotics #humanrobotinteraction
One of Misty's most outstanding talents is display customizability, which allows you to design your very own robot characters and build unique robot interactions for your projects 🔬💫 🎉 Check out our short video tutorial to learn how to make this happen behind the scenes using Misty's Blockly 😉
Misty was delighted to see the positive impact her robotics course had on all students at the local school "Vaxmoraskolan," located in Sollentuna, Stockholm. Last week, she decided to let us share all the fun and excitement they had with Blockly programming. 🤩💫👩💻
Want to know how to get started with Misty II? We've got you covered 😉 Watch our video tutorial on Misty Studio, Misty's graphical user interface 👉 https://buff.ly/3PzcKfV #edutech #hri #humanrobotinteraction #education #innovation #robotics #ai #socialrobotics
Under computer and wireless LAN control, medical AGVs travel along set routes, connecting departments such as outpatient clinics, pharmacies, operating rooms, and blood banks through an information network and send/receive workstations, automatically completing the delivery of items such as meals, sterilized supplies, and medical waste. They can travel at speeds of up to 1.5 m/s, feature automatic obstacle avoidance and path planning, can operate continuously for 24 hours, and recharge automatically when battery power is low. Compared with traditional logistics systems, they require no pre-laid pipes or tracks, reduce the risk of infection from personnel contact, and save manpower.
Internal material transport in manufacturing and warehouse environments relies heavily on manual forklifts and carts, which brings high labor costs and is prone to errors. Autonomous mobile robots (AMRs) require no additional infrastructure and rely on built-in sensors, cameras, and intelligent software to navigate safely around workers and equipment. They can integrate with ERP, MES, and WMS to automate processes such as raw material storage, line-side delivery, and finished goods transfer, allowing workers to focus on high-value work.