The author uses EvergreenSLAM, a self-developed 2D lidar SLAM project, as Claude Code's spatial memory. Through the egs command line, each location corresponds to a folder and notes.md on the lidar map, with the agent writing notes itself, and a two-day real-world demonstration was completed on an XLeRobot.
Based on a survey of more than 20 mainstream quadruped robot projects on GitHub, the article maps the open-source quadruped technology landscape across five schools: kinematic control, full-stack engineering, optimization algorithms, reinforcement learning, and hardware platforms.
The ICLR 2026 paper SAGE-3D proposes decoupling and assembling 3DGS rendering, source-mesh collision bodies, and manual semantic annotations into an executable navigation environment, modeled as a semantics- and physics-enhanced POMDP. Its object-level semantic anchoring provides category, instance ID, and bounding box, while physics-aware execution assembles collision bodies extracted from source meshes via CoACD convex decomposition and embeds them into the 3DGS scene.
A team from Tianjin University and Hefei University of Technology proposes FAME, which uses Conditional Flow Matching to generate multiple exploration anchors simultaneously, then determines exploration routes through A* path planning, path clustering, and hierarchical selection. The paper was accepted to NeurIPS 2026 and selected as a Spotlight.
In each round, Dijkstra selects the node with the smallest cumulative cost from the source among the nodes not yet finalized and formally finalizes it, then performs relaxation updates on its neighbors; under the premise of non-negative edge weights, this guarantees the globally shortest path. If negative-weight edges exist, a finalized node may be overturned by a cheaper route, so standard Dijkstra no longer guarantees correctness, and algorithms such as Bellman-Ford must be used instead.
REEMAN outlines an approach for retrofitting existing stretch wrappers using Ironhide autonomous forklifts: the forklift handles pallet pickup, placement, and warehouse transport, while the stretch wrapper continues to perform packaging, requiring a machine control interface, status feedback, and safety interlocks. In its wireless docking demonstration, pickup alignment deviation was within 2 cm and turntable placement deviation was within 5 cm, with the demonstration process taking about 6 minutes. There is also an Ironhide 3.0 deployment at a factory in Jingmen, Hubei; the two applications are independent of each other, and the data comes from REEMAN application materials and has not been independently verified.
At an electronic materials factory in Jingmen, Hubei, REEMAN Ironhide 3.0 autonomous forklifts complete an integrated process: picking up goods in the production area, passing through two electric doors, handing off to a stretch wrapper, and delivering to the warehouse. The actual load is about 1,300 kg, with a rated load of 1,500 kg. The customer reports about 75 transport cycles per day, about 6.5 minutes per task, and a loaded route of about 270 meters, with support for automatic return to a mobile charging station. The forklift's reference price is about USD 9,900, and integration with doors and the stretch wrapper requires customization based on the site's controller interfaces.
REEMAN demonstrated a library inventory robot built on the Hot Wheels robot chassis, completing a mobility demo in indoor bookshelf aisles, but did not provide OCR recognition accuracy or inventory results. The chassis uses Laser SLAM and V-SLAM for mapping and localization, combined with lidar, 3D vision, and multi-sensor fusion for obstacle avoidance; published specifications include a 270° laser scanning angle and a maximum detection range of 25 meters.
The inaugural Global Ocean Robotics Day was held on October 21, with Constructor University as one of 10 in-person hubs worldwide. The event focused on how artificial intelligence and autonomous robots are transforming ocean research and exploration. Topics covered marine AI, autonomous underwater vehicles, and ocean observation robotic systems, including mapping ocean currents using marine vehicle trajectories, simulating autonomous docking with digital twins, and underwater robot swarms. The university's Ocean Systems and Robotics research group also demonstrated several autonomous marine robots developed by the team on site.
Maximo Robotics announced that its US-made robotic systems have completed more than 300,000 construction tasks under task-based robotics-as-a-service contracts, and that workers collaborating with the Max 4.0 system achieved installation efficiency 2.5 times that of manual labor per person per shift hour.
The study proposes REA (Routed Embodied Agent), a unified runtime that lets a reasoning-focused System 2, an embodied routing System 1, and reactive System 0 policies collaborate within the same framework, with the core rule that a running policy does not wait on the brain.
Daxiao Robotics, in collaboration with Nanyang Technological University's S-Lab, has released Mundo, a world model that reconstructs 1 to 12 real-scene images into an explorable, physically interactive 3D simulation environment. The model has built-in physical properties, with MuJoCo computing object motion under gravity and contact constraints, and it supports training-data generation and policy testing for tasks such as navigation and pick-and-place. The team has reconstructed a hotel floor and laundry area for training service robots on navigation and interaction tasks.
Hangcha Group is positioning itself in physical AI with its AI-Fork full-series intelligent forklift robots and the LogiMind industrial embodied intelligence large model, and plans to launch new products at the fourth Global Dealer Conference on November 10. Global forklift stock stands at about 12 million units, while unmanned forklift penetration in 2025 is only around 2%; Hangcha aims to lower deployment barriers through environment understanding and task planning capabilities. Its technology portfolio also extends to Hangcha Guazi's latent transport robots, bin robots, and the X1 wheeled humanoid robot.
BlueBotics CEO Nicola Tomatis points out that key decisions in mobile robot projects—such as fleet size, routes, charging points, and pick/drop locations—are already fixed before deployment, and simulation can verify these assumptions in advance. In simulation tests, his team found that after accurately modeling pick/drop behavior, similar vehicle configurations required about 30% different numbers of vehicles to complete the same task; adding just 1 second of delay at each intersection also significantly affected system performance. Once a project reaches about 5 to 10 vehicles, the value of simulation rises rapidly.
Smart wheelchair makers are redefining their products as 'travel robots' or home entry points for embodied AI, seeking to attract younger users with 'light autonomous driving.' Strutt's EV1 carries 20 sensors—2 LiDARs, 10 ToF sensors, 6 ultrasonic radars, and 2 cameras—enabling 360° environmental perception and 3D mapping; Xcan Mate X from Xingcan Intelligence uses a LiDAR, ultrasonic radar, and camera setup. By 2026, the sector has seen at least 9 publicly disclosed funding rounds, but buyers remain concentrated among people aged 30-50 purchasing for their parents.
At the 2026 China Marine Economy Expo, Lead Robot released "Dingfengbo," a world model for shipbuilding. It takes design drawings, process documents, and bills of materials as unified inputs, parses the engineering logic among them, breaks an entire ship down into blocks, sub-assemblies, sub-assemblies and parts, and reconstructs the 3D structure. This frees robots from reliance on manual teaching, enabling them to perceive autonomously, decompose tasks, and decide how to work.
A team from Fudan University and Nanyang Technological University, Singapore, proposed SimpleVLN, which performs vision-language navigation without changing the core structure of the multimodal large model or adding dedicated navigation modules, relying only on historical frame selection, action-chunk decision-making, and three-stage training with SFT, DAgger, and reinforcement learning.
A team from Fudan University and Nanyang Technological University, among others, proposed SimpleVLN, which performs vision-language navigation without changing the core structure of a multimodal large model or adding dedicated navigation modules, relying only on observation filtering, action-chunk decision-making, and three-stage training with SFT, DAgger, and reinforcement learning.
Daxiao Robotics, in collaboration with Nanyang Technological University's S-Lab, has released Mundo, an interactive world model built for specific environments that integrates 3D scene reconstruction, physics simulation, and video generation. The system uses scene code to record object weight, center of mass, and friction coefficients, computes motion and collisions with a 960Hz MuJoCo physics engine, and then renders visuals combined with on-site photos; inputting 1 to 12 phone photos is enough to reconstruct a 3D space that supports pushing, pulling, and collisions.
Following the thread of "how to solve problems in different environments," the article reviews five categories of path planning paradigms in embodied intelligence: graph search (Dijkstra, A*, Hybrid A*), sampling methods (RRT, PRM), artificial potential field (APF), optimal control (MPC), and reinforcement learning and data-driven methods.
AUBO collaborative robots cover six major industrial scenarios—palletizing, assembly, grinding, spraying, inspection, and sorting—through force control, vision, and flexible deployment. The palletizing workstation supports one-click pallet pattern generation, 2-hour deployment, a maximum pallet height of 2.5 m, and a cycle rate of 9 boxes/min; the assembly station has a cycle time of 15 seconds and a fastening success rate of 99.8%; the newly released G40 adaptive collaborative robot has been used for grinding automotive metal structural parts and features IP67 protection and an explosion-proof design.
Mino's companion behavior application is implemented on the Jetson Orin Nano based on ROS 2, with three states: finding a person, following, and accompanying; the video is its first outdoor test. The application uses open-source modules such as depth estimation, YOLO-World person detection and tracking, pose feedback, odometry, and motion control, and a coding agent assembled the behavior logic within a weekend. The robot perceives the person by 3D position and velocity, smooths the trajectory with a Kalman filter, and continuously sends head pose and chassis velocity commands to custom hardware.
The REEMAN YIMA YBOT20P outdoor delivery robot can automatically call an elevator, enter it, and deliver across floors through a compatible elevator control interface, without anyone needing to press the elevator buttons. The platform is designed for indoor and outdoor transport, has a payload of 100 kg, uses dual-lidar SLAM navigation, autonomous obstacle avoidance, and automatic charging, and can connect the delivery process from a community parcel station to a designated apartment entrance. Elevator integration requires confirming compatibility with the manufacturer's API, control module, or building system interface, and the elevator's existing safety controls must not be bypassed.
REEMAN has released the YIMA YBOT20P outdoor delivery robot, which carries parcels from courier stations to apartment doorsteps along mapped community routes, with a rated payload of 100 kg. The robot uses dual-laser SLAM navigation, with a 360° top laser and a 230° front laser, an adjustable speed of 0.3–1 m/s, maximum obstacle clearance of 3 cm, maximum climbing grade of 10%, a nominal battery life of 10–15 hours, and support for automatic return to charge on low battery. It enables cross-floor delivery through integration with compatible elevator and automatic door controllers, and outdoor operation is limited to dry paved roads under no-rain conditions.
OpenFlyV2 is a unified onboard vision-language-action architecture that uses a shared vision-language representation to support semantic understanding, language-grounded perception, spatial reasoning, and continuous flight control within a single model. The model generates action chunks of 12 waypoints each entirely onboard at 10 Hz on an NVIDIA Jetson Orin NX, and transfers zero-shot from simulation to a real quadrotor, achieving a 48% navigation success rate in four real environments without using real aerial training data.
On October 22, 2026, Signode announced the expansion of its autonomous mobile robot (AMR) portfolio, adding autonomous lift robots, pallet trucks, counterbalance forklifts, reach trucks, and stacker forklifts, covering everything from end-of-line packaging to warehouse transport.
The Mino robot completed its first outdoor autonomous stroll test. Its ROS 2-based following app runs on a Jetson Orin Nano and has three states: finding people, following people, and companionship. The app uses YOLO-World to detect pedestrians and stereo vision depth estimation to generate 3D bounding boxes, combined with Kalman filter tracking, and continuously recomputes trajectories to control the base and head. The app was completed over a weekend using open-source software modules and a coding agent, and will be released as open source with the robot if the Kickstarter crowdfunding campaign succeeds.
Ropedia held Ropedia Super Week from October 19 to 23, releasing six studies at once: Ropedia W0, HoliWorld, EgoXperience, DONIC, S-Agent 2.0, and Ropedia A0, covering human experience recording, world model prediction, and two-handed manipulation learning.
A research team from Aether AI and UC San Diego has proposed RobotAgentBench (RAB), a comprehensive benchmark for open-world robot autonomy, building 50 household layouts, 90 atomic tasks, 18 short-horizon tasks, 8 long-horizon tasks, and 4 types of open-ended scenarios.
A team from Tianjin University and Hefei University of Technology proposes FAME, which uses Conditional Flow Matching to simultaneously generate multiple exploration anchors, then determines routes through A* path planning, path clustering, and hierarchical selection. The paper was accepted to NeurIPS 2026 and selected for Spotlight.
A research team from Harvard, Nvidia, Berkeley, MIT and Stanford proposed RAX (Robot Agent eXploration), which connects a general coding agent to robot perception and control interfaces, uses online exploration to write successful decisions into executable programs and accumulate them as skills, while keeping model parameters frozen.
Why it matters: RAX uses a frozen model plus online exploration to generate reusable programs, offering a robot learning path that does not depend on action demonstrations.
Berkshire Grey releases BG Pilot software with enhanced AI capabilities, built on a decade of real-world warehouse operations data and billions of robotic picks, to address the diversity of packaging and mixed SKUs.
One robot moving a load is a demo. Two robots sharing one floor is a fleet. Two Sherpa 10K robots, two fully loaded racks, one clean handoff. Planning a multi-robot floor? Contact our team: https://www.atirobotics.ai/contact/ #AMR #Robotics
Galileo has released an upgraded petrochemical intelligent inspection solution, centered on the new-generation Ex2 intelligent bionic legged explosion-proof robot and paired with an embodied AI inspection platform. The Ex2 has obtained IIB+H₂ extended explosion-proof certification and reaches the T5 temperature class; it can climb 35° slopes and 30cm steps, with runtime upgraded to 7 hours and payload upgraded to 100 kilograms.
The Yufeng 250-L LiDAR Edition quadcopter drone kit is aimed at research and education. It consists of three core modules: a flight controller, an onboard computer, and LiDAR. It can deploy SLAM, planning, and control algorithms to achieve real-time navigation in unknown environments.
REEMAN Robotics is developing a commercial autonomous tennis ball-collecting robot for professional tennis academies, multi-court clubs, and commercial tennis facilities, with the goal of reducing repetitive ball-collecting work and allowing coaching teams to focus on training. The product is in the development and technical evaluation stage. Development priorities include a ball-storage target of about 200–300 balls, autonomous navigation, tennis ball detection and collection, automatic charging, and court surface compatibility; these are engineering goals rather than verified mass-production specifications. REEMAN has not yet announced a confirmed launch date, final price, mass-production specifications, or sales channels.
In a Gazebo mecanum-wheel robot simulation, A* and Dijkstra showed no measurable difference after 18 runs each, and the author wrote "no difference" into the conclusion. The article also discusses how to determine that an experiment showing no measurable difference is not simply the result of a flawed setup.
An analysis of the minimum-snap quadrotor trajectory pipeline: piecewise ninth-order polynomials (ten coefficients) are solved in closed form with snap as the cost and derivative-continuity constraints, while COBYLA trades off time against smoothness. The differential-flat mapping yields thrust and attitude, and geometric control is validated in closed loop in MuJoCo.
Pangqi Technology is using a RaaS (Robotics as a Service) leasing and operating model to bring embodied AI robots into factories. Customers pay based on actual work volume and avoid large one-time investments; projects already in operation can achieve a 30%–300% improvement in per-square-meter efficiency and rapid deployment in 1–3 months.
The authors have released LEAN, a reinforcement learning-based local energy-aware navigation planner. The code is open-sourced on GitHub, and the paper was published in IEEE Robotics and Automation Letters. The authors state that the method does not require using proximity to the goal as a reward; instead, it lets the agent learn from failed trajectories and provides an explicit energy signal. An accompanying video compares the model's inference performance in real environments against an ablation version without energy awareness and a classical NMPC planner.