RoboSignal Daily · 2026-10-25 · Sunday
ROBOSIGNAL
Gaode Releases First Full-Body Base Model λ0 Based on Open-World Behavior
Gaode, under the Alibaba Group, officially released its first full-body base model λ0 based on open-world behavior. This model enables robots to learn full-body coordination, including movement, posture, and hand operations, solely through human action data, thus overcoming the limitations of 'local action learning.' In the SIMPLE simulation benchmark and multiple real-world evaluations, λ0 achieved state-of-the-art performance, with data collection efficiency approximately 3.4 times that of mainstream methods and a 22.5% increase in real-world task success rate compared to the strongest baseline.
Research
Gaode Releases First Full-Body Base Model λ0 Based on Open-World Behavior
Gaode, under the Alibaba Group, officially released its first full-body base model λ0 based on open-world behavior. This model enables robots to learn full-body coordination, including movement, posture, and hand operations, solely through human action data, thus overcoming the limitations of 'local action learning.' In the SIMPLE simulation benchmark and multiple real-world evaluations, λ0 achieved state-of-the-art performance, with data collection efficiency approximately 3.4 times that of mainstream methods and a 22.5% increase in real-world task success rate compared to the strongest baseline.
Changsha's Embodied Intelligence Push: Manufacturing as the New Frontier
Changsha is positioning itself as a hub for embodied intelligence by leveraging its manufacturing strengths. Companies are repurposing existing capabilities in materials, precision manufacturing, and industrial automation to build robotic hands and systems. The city is focusing on scalable production rather than algorithmic innovation, with local firms like Zhongke Huisil, Daili New Materials, and HuaShu High Tech contributing to the development of robotic components and systems. The article outlines how Changsha is transforming its industrial base into a platform for embodied intelligence, emphasizing the importance of physical system development over algorithmic breakthroughs.
Robotics API Service: Physical Intelligence Launches Pi API Beta
Physical Intelligence (PI) has launched a beta version of its Pi API, offering a service chain that includes data upload, model fine-tuning, cloud inference, and evaluation feedback. This marks a significant step in commercializing embodied AI by making robot 'brains' accessible as APIs, though challenges like data privacy, cost, and deployment readiness remain unresolved.
PrismWAM and RoboDoctor: Exploring the Scaling Path of Physical AI
Professor Li Yiming's team from ClearMind Intelligence has released two Physical AI research results: PrismWAM enhances robot motion generation capabilities by decoupling semantic and physical state modeling; RoboDoctor enables diagnosis and self-improvement through execution feedback. Both works explore Scaling Physics from the model and system levels. PrismWAM achieved first place in RoboTwin 2.0 evaluations, while RoboDoctor scored 57.1 in RoboDojo tests, ranking first among current comparisons.
Changsha's Embodied AI Push: Manufacturing Capabilities as the New Frontier
Changsha is positioning itself as a hub for embodied AI by converting its existing manufacturing strengths into robotic capabilities. Companies like Zhongke Huisil, Daile New Materials, and HuaShu High Tech are adapting their expertise in materials, 3D printing, and precision manufacturing to develop functional robotic hands and systems. The city is also creating dedicated training environments and data ecosystems to support the transition from prototyping to mass production.
110 Humanoid Robots Complete Complex Synchronized Actions, Break Guinness World Record
110 humanoid robots completed complex synchronized actions in 5 minutes and 43 seconds, successfully breaking the Guinness World Record. The system achieved high levels of autonomous operation, including autonomous positioning, dynamic formation, and task scheduling, marking an important step toward the practical application of humanoid robots.
Robots Learn Skills on Their Own. Fudan Proposes New Embodied Self-Evolution Framework
A team from the Fudan University Trustworthy Embodied Intelligence Research Institute has proposed the RoboSkill framework, enabling robots to complete tasks through on-site observation, reasoning, and code writing. The framework also enhances execution efficiency through skill accumulation and tactile feedback. The study shows that even without pre-collected data, a general agent can perform various operational tasks, although the execution time is longer. RoboSkill improves task completion efficiency by saving effective methods and code, reducing redundant exploration.
Physical Intelligence Launches Pi API Beta for Embodied Intelligence Deployment
Physical Intelligence (PI) has launched the Pi API beta, offering a service chain that includes data upload, hosted fine-tuning, cloud inference, and evaluation feedback. This marks a significant step toward commercializing embodied intelligence models, with potential impacts on both hardware companies and other model providers.
Annual Production Line for 100,000 Joint Modules Launched in Foshan; Listed Company Invests 1 Billion Yuan in Robot Parts Base in Suzhou | Market Observation
The article reports on several significant developments in China's robotics industry, including the launch of the production line in Foshan, the construction of the base in Suzhou, collaboration among multiple companies to advance embodied intelligent technology applications, and the release of several new robot products. These events highlight China's ongoing growth and technological breakthroughs in robotics manufacturing, R&D, and application.
CloudDeepTech Releases Five Key Achievements of Embodied Brain, Building a Complete Technical System
CloudDeepTech has released five research achievements around its self-developed embodied brain DEEPRobotics-Brain, covering world models, foundational models, agents, strategy optimization frameworks, and data closed-loop platforms. These achievements cover the full technical chain of embodied intelligence, aiming to address issues such as perception, prediction, navigation, and autonomous iteration of robots in real physical environments. The research collaborates with multiple top international universities, introducing physical constraints, unlabeled video pre-training, self-evolving agents, and unified data benchmarks to advance embodied intelligence into real-world scenarios.
Robots & Hardware
9-minute Fast Charge, 3-minute Battery Swap, 1-hour Wireless Charge: How Much Can a Humanoid Robot Do in a Day?
The article explores three energy replenishment strategies for humanoid robots: fast charging (9 minutes), battery swapping (3 minutes), and wireless charging (1 hour). Each method is tailored to different use cases, such as industrial assembly lines, warehouse logistics, and human-robot coexistence environments. The focus is on improving operational uptime and reducing downtime through advanced battery technology and energy management systems.
Deployments
Shanghai Hosts Real-World Robot Testing Event with 800 Teams
In Shanghai, over 800 teams are participating in a real-world robot testing event, simulating industrial and service scenarios to evaluate robot performance under unpredictable conditions. The event aims to bridge the gap between lab testing and real-world deployment, with a focus on skill modularity and industry collaboration.
Briefs
- Full-stack Autonomous Controllable Large Model Enters Factory, Lingdong General Accelerates Embodied Intelligence DeploymentChinese robotics — Robot learning research
- Neura Robotics Launches Training Infrastructure for Humanoid RobotsRoboSpeak — WeChat
- RoboSkill: A Skill Acquisition and Reuse Loop for Embodied AgentsChinese robotics — Robot learning research
- 110 Humanoid Robots Synchronize Dance to Set Guinness World Record, Zhiyuan Demonstrates Scalable Deployment CapabilityChinese robotics — Hardware and sensing
- Shanghai Drives Robotics Deployment Through Real-World TestingChinese robotics — Deployment discovery
- AI WAVE｜Zhiyuan Releases Genie Studio 2.0, Bridging the Final Mile for Robot DeploymentChinese robotics — Dataset and collection discovery
- Without Leaving Pudong, You Can 'Assemble a Robot'. Supply Chain Evolves from 'Enterprise Clustering' to 'Collaborative Supply'Chinese robotics — Company news discovery
- Robotics' OpenAI, Here?Chinese robotics — Research news discovery
- 9-minute Fast Charge, 3-minute Battery Swap, 1-hour Wireless Charge: How Much Can a Humanoid Robot Do in a Day?RoboSpeak — WeChat
- Automotive Companies Developing Humanoid Robots for Production and Customer InteractionChinese robotics — Hardware and sensing