RoboSignal Daily · 2026-10-21 · Wednesday
ROBOSIGNAL
Isaac Lab 3.0 Released: Multi-Backend Architecture for Robot Learning
Isaac Lab 3.0 is now generally available, offering a unified framework for robot learning with support for multiple physics, rendering, and visualization backends. Key features include Newton physics, OV libraries, and Isaac Sim integration, along with improved simulation tools and reinforcement learning compatibility.
Research
Isaac Lab — ReleasesFirst-partyIsaac Lab 3.0 Released: Multi-Backend Architecture for Robot Learning
Isaac Lab 3.0 is now generally available, offering a unified framework for robot learning with support for multiple physics, rendering, and visualization backends. Key features include Newton physics, OV libraries, and Isaac Sim integration, along with improved simulation tools and reinforcement learning compatibility.
RAX: Open-Source Framework for Autonomous Robot Exploration and Skill Acquisition
Researchers from Harvard, NVIDIA, UC Berkeley, MIT, and Stanford introduce RAX, an open-source framework that allows robot agents to autonomously explore and acquire skills through online experience. RAX enables robots to write and execute code in real or simulated environments, improving generalization and task performance across various robotic platforms. The framework shows significant improvements in benchmark tests, but real-world deployment and skill reuse still require further validation.
arXiv Robotics — research abstractsEndoVLA-OOV: Language-Conditioned Out-of-View Target Acquisition in Robotic Endoscopy
This research presents EndoVLA-OOV, a vision-language-action policy for robotic endoscopy that translates anatomical descriptions into camera motions. The policy uses a pretrained backbone adapted to continuous distal bending commands and combines images to guide search. An open-source dataset with 1,050 episodes and 262k observation-action pairs supports policy learning. Physical evaluations show 93.3% success in cross-embodiment transfer, with success rates improving from 40.0% to 90.0% with more training data.
arXiv Robotics — research abstractsRefuse the Harm, Complete the Task via Graph-of-Rules Planning for LLM-Enabled Robots
This research presents RuleGraph, a multi-agent Graph-of-Rules planning framework that allows LLM-enabled robots to reject harmful parts of instructions while completing legitimate tasks. The method distributes safety enforcement across multiple reasoning stages, achieving 0% attack success rate and high task success rates on benchmark environments.
GreenWAM: A World Foundation Model for Robotics
A research paper introduces GreenWAM, a world foundation model for robotics that unifies world modeling and action control. The model is trained on a unified action space covering 20 robot types and demonstrates strong performance in trajectory prediction, depth estimation, and zero-shot task success across multiple benchmarks and real-world robot tasks. The model's architecture integrates world modeling with action generation through a shared transformer-based framework.
RAX: Open-source framework for robot agents that autonomously explore and acquire skills
Researchers from Harvard, NVIDIA, UC Berkeley, MIT, and Stanford introduce RAX, an open-source framework that allows robot agents to autonomously explore and acquire skills through online experience. RAX enables agents to write and execute code in real or simulated environments, improving generalization and performance across tasks. The framework supports multiple robot types and demonstrates strong results in benchmark tests.
World Models × Embodied Intelligence | 2026 Five Representative Companies' Technical Paths and Scene Deployment Observations
This article observes the technical paths and scene deployment progress of five Chinese companies in the field of world models and embodied intelligence. It evaluates their core competencies and development prospects, emphasizing the transition from parameter scale to system-level performance and the importance of real-world data feedback in model iteration.
TechCrunch — RoboticsEka Robotics' $1.45B Valuation Driven by Physical AI Model for Dexterous Manipulation
Eka Robotics, founded by MIT professor Pulkit Agrawal and former Google DeepMind researcher Tuomas Haarnoja, has raised $150 million in Series A funding, valuing the company at $1.45 billion. The startup's physical AI model, which uses a vision-force-action architecture, enables a robot to manipulate objects with high reliability and speed, surpassing human performance in some metrics. The model leverages simulation and reinforcement learning to scale training, though real-world deployment remains a challenge.
Embodied Intelligence's ROI in Agriculture: A Case Study in Fruit Harvesting
The article explores the economic viability of deploying embodied intelligence in agricultural settings, using fruit harvesting as a case study. It details how robotic systems can reduce labor costs, with ROI calculations showing potential for cost recovery within a year. Challenges such as seasonal constraints, specialization, reliability, and service networks are also discussed, emphasizing the need for policy support and scalable solutions to achieve widespread adoption.
Dream API, the Open-Source Robot Brain API from Dream Labs
Dream Labs, founded by the NVIDIA DreamZero team, has launched Dream API, the first publicly available robot brain API. Dream-1, the model behind the API, enables zero-shot control of robots using natural language commands, with support for over 20 hardware partners. It outperforms several large models in benchmark tests, though it is still in beta with known limitations.
538 Hours of 'Extremely Dirty' Human Data: Hong Kong University Teaches Robots to Unbox, Use Hot Glue Guns, and Make Cold Drinks
A team from the MMLab at the University of Hong Kong has introduced EgoWild2Dex, a dataset and framework that uses raw, uncurated first-person human operation data to train robots on complex tasks such as unboxing, using a hot glue gun to fix figurines, and making cold drinks. The method achieves an average success rate of 96.7% by employing GeoFormer perspective alignment and action retargeting to transfer natural human data into the robot's action space, reducing reliance on laboratory environments.
Deployments
Climbing Robot Market Analysis: Technical Evolution and Player Landscape in Extreme Throughput Scenarios
Interact Analysis reports that the global mobile robot market in 2025 showed a 'slow growth, price increase' trend, with a 17.5% shipment growth and 24.6% revenue increase. The article outlines the technical advancements and market strategies of leading climbing robot solutions such as Exotec Skypod, HaiPick Climb, ZeeBot, and RS Hyper, focusing on metrics like climbing speed, system throughput, and safety mechanisms. It also discusses the growing adoption of climbing robots in high-density, high-throughput warehouse environments.
Briefs
- RAX: A Framework for Autonomous Robot Exploration and Skill AcquisitionChinese robotics — Research news discovery
- T(R,O) Grasp++: Cross-Embodiment Dexterous Grasping in Cluttered Environments at ScalearXiv Robotics — research abstracts
- Online World Modeling Enables Real-World Inverse Reinforcement Learning from ObservationarXiv Robotics — research abstracts
- Giving Robots Physical Intuition: A 47-Page Review Explains What AI Robots Really LearnChinese robotics — Research news discovery
- Robot Wars – How China Has the Robotics Industry in Its GripRobotics — Web news discovery
- Nuwa Robotics' Yang Ruiqiang: Why Delivery Robots Still Can't Reach Your Doorstep After a Decade of DevelopmentChinese robotics — Company news discovery
- TRACT: Temporally Routed Action Chunks with Chronological Phase Authority for Contact-Rich ManipulationarXiv Robotics — research abstracts
- Industrial Embodied Intelligence Takes Two Paths at 2026 China Robot ExpoHigh-tech Robotics — WeChat
- D-Robotics Open Sources Robot World Model Uranus: Letting AI Practice in a Virtual WorldChinese robotics — Dataset and collection discovery
- OccSim: Multi-kilometer Simulation with Long-horizon Occupancy World ModelsarXiv Robotics — research abstracts