RoboSignal Daily · 2026-10-27 · Tuesday
ROBOSIGNAL
Paimon-0 Leads in Multiple Embodied Benchmark Tests, Shenzhen Startup Primotion Makes Debut
Primotion, a Shenzhen-based company established just three months ago, has launched its embodied base model Paimon-0, which has achieved first place in several benchmarks including RoboColiseum, RoboCasa365, and VLA-Arena. The model employs a World Action Model (WAM) and Inverse Dynamics Model (IDM) architecture, combined with FPGA hardware control to achieve low latency. The company enhances the model's generalization capabilities through data closed-loop and self-improvement systems (RSI), and plans to open-source the Progress Model Paimon-Eval.
Research
Paimon-0 Leads in Multiple Embodied Benchmark Tests, Shenzhen Startup Primotion Makes Debut
Primotion, a Shenzhen-based company established just three months ago, has launched its embodied base model Paimon-0, which has achieved first place in several benchmarks including RoboColiseum, RoboCasa365, and VLA-Arena. The model employs a World Action Model (WAM) and Inverse Dynamics Model (IDM) architecture, combined with FPGA hardware control to achieve low latency. The company enhances the model's generalization capabilities through data closed-loop and self-improvement systems (RSI), and plans to open-source the Progress Model Paimon-Eval.
RoboCoach: World Models as Active Coaches for Compositional Robot Skills
RoboCoach is a system that uses world models to enable efficient active correction in embodied reinforcement learning. By decomposing long tasks into subtasks and directing human demonstrations to specific experts, RoboCoach achieves significant performance improvements on real robots like Franka and AgileX, with success rates increasing from 13.3% to 75.0% and 40.0% to 83.8%, respectively. The system's RIDI framework (Route—Imagine—Diagnose—Improve) enables targeted skill refinement, reducing the need for extensive human intervention.
Xpeng IRON Second Generation VLA Completes Edge Deployment: 0.2 Seconds per Red Inference, Predicts 6 Seconds of Actions
Xpeng Auto announced that its second-generation IRON humanoid robot VLA model has completed edge deployment, enabling red inference every 0.2 seconds and the ability to predict human actions for the next 6 seconds. The model enhances the smoothness and safety of human interaction through real-time perception and dynamic prediction, suitable for scenarios such as stores and factories.
arXiv Robotics — research abstractsEmergent Cooperative Thermoregulation in Lunar Swarm Robotics
This research explores how a swarm of 20 lunar robots can survive the extreme cold of the lunar night by collectively regulating temperature through emergent behavior, inspired by penguin huddling. A single reinforcement learning policy enables the swarm to aggregate and circulate between sheltered and exposed areas, achieving survival with 37% less energy than a motionless huddle.
RoboParty Releases Fully Open-Source Humanoid Robot RP1
RoboParty officially launched the high-performance, fully open-source humanoid robot RP1 at the IROS 2026 exhibition. This platform not only opens up code but also covers mechanical, electrical, joint, low-level control, simulation, training, and deployment development stages, providing a complete toolchain for embodied intelligence research. RP1 features 24 to 30 degrees of freedom, supporting bipedal movement, balance control, and complex task research, aiming to drive humanoid robot development toward a more open and collaborative direction.
Underwater Robotics Orders Surge, Why This Business Still Only Has Half the Accounting Done?
The article discusses the recent surge in underwater robotics orders and investments, focusing on how these systems are valued for their ability to replace expensive marine operations rather than individual labor. It outlines the industry's progress in securing contracts and funding, while also noting ongoing challenges in data scarcity, profitability, and talent shortages.
arXiv Robotics — research abstractsSee What I Need, Dream What I Want: A Navigation Foundation Model on a Self-Constructed Representation Space
This paper presents RepresentNav, a navigation foundation model that constructs a self-constructed representation space using a compact three-layer transformer. It learns to see what it needs for navigation and predicts future scene representations through a joint-embedding predictive architecture, achieving state-of-the-art results on multiple benchmarks and physical robot navigation tasks.
Industrial robotics — News discoveryHumanoid Robots Collaborate with Machine Tools Using Physical AI: Mitsubishi Electric's Demonstration at JIMTOF
Mitsubishi Electric has developed a machining system where humanoid robots and CNC machine tools collaborate using physical AI, allowing workers to issue natural language instructions. The system includes an AI Machining Task Planner, a CNC machine tool with physical AI, and a general-purpose humanoid robot. While the system demonstrates potential for reducing labor dependency, challenges such as safety and on-site verification remain before practical deployment.
Data & Training
Husai Releases Kosmo: A Data-Driven Approach to Embodied AI
Husai, a LiDAR company, has launched Kosmo, a 3D world data engine that captures real environments and converts them into editable, reusable data for robot training, world modeling, and content creation. The system includes the Kosmo One AI space camera and the Kosmo World cloud platform, enabling efficient spatial reconstruction and scene modification for diverse applications.
Deployments
Robotics & Automation NewsAdvaSolutions Automates Logistics Center for Drugstore Chain Hebe in Poland
AdvaSolutions is implementing a modular warehouse automation system for Hebe, a Polish drugstore chain, in an existing facility. The system includes 66 advaShuttles and 20,000 tote storage locations, integrated with SAP EWM. The project is designed to support Hebe's omnichannel logistics needs without disrupting ongoing operations.
Zhuyuan Releases G2 Max, a Safety-Grade Heavy-Lift Embodied Robot for
Zhuyuan has released the G2 Max, a safety-grade heavy-lift embodied robot with 40 kg load capacity and sub-millimeter-level repeat positioning accuracy, solving the challenge of precise operation in heavy-duty robotic arms. The robot features 7-degree-of-freedom force-controlled dual arms, capable of handling various material handling tasks and performing efficient unloading and stacking operations in confined spaces. Additionally, the G2 Max is tasked with
JD Logistics × Zhiyuan Robotics: Embodied Robot G2 Max Applied in Zhi Wang Warehouse
Zhiyuan Robotics has deployed the G2 Max, an industrial-grade embodied robot, in JD Logistics' Zhi Wang warehouse for inbound handling tasks. The robot can carry up to 50kg with high precision, achieving over 99% operational success rate. This marks a real-world application of embodied intelligence, with JD Logistics emphasizing the importance of industrial scenarios for testing and refining such technologies.
Briefs
- RoboSpeak — WeChat（mp_account，tier=T2）RoboSpeak — WeChat
- 50kg Load Capacity + 0.5mm Precision. Zhiyuan Unveils Heavy-Duty Embodied Robot Ready for Factory UseRoboSpeak — WeChat
- Sci. Robot.| Robots 'Hear' the Shape of the Field: No Localization, No Map, Just Neighbor-to-Neighbor CommunicationRoboSpeak — WeChat
- BTick — Behavior Tree Monitoring & Debugging for ROS 2ROS Discourse — Community updates
- Robotic Robot Launched: Zh iyuan Releases Heavy-Load Embodied Robot精灵G2 MaxChinese robotics — Deployment discovery
- GPT-6Astra Embodied Intelligence Breakthrough and Physical AI Data Flywheel InitiationChinese robotics — Dataset and collection discovery
- Elon Musk Built a School for Robots. Repurposing Million-unit Vehicle Line for Mass ProductionQbitAI — Robotics
- Full-Weight VLA Fine-Tuning Without an fp32 Master Copy Silently Erases Most UpdatesarXiv Robotics — research abstracts
- From Huawei to Entrepreneurship: 23 AI Startups and Their Journey in RoboticsChinese robotics — Company news discovery
- 110 Humanoid Robots Perform Synchronized Dance, Marking a Commercialization MilestoneChinese robotics — Deployment discovery