A single destination for timely, editor-curated robotics news from around the world.
MirroS has announced the open-source release of AgentGarten, a platform where AI agents can learn by engaging in code-defined environments. Utilizing a real-time neural renderer, these agents can act, review their actions, and develop their own strategies, such as hide-and-seek tactics, during gamep...
A total of 350 robots were deployed at Changlong for a month, where they encountered significant challenges and exhaustion. Interestingly, these robots also learned to intentionally lose to children during interactions, showcasing an unexpected adaptability in their programming. This development is...
Researchers at the University of Pennsylvania have developed a robot system named SymSkill that can learn to complete complex tasks with up to 12 steps after only five minutes of video demonstrations. This innovative approach allows the robot to adapt its actions when faced with interruptions or fai...
In the latest episode of Robot Talk, Michelle Lu, co-founder and CEO of Vsim Technology, discusses the use of simulation and AI to enhance robot learning. Lu, who previously served as Director of Simulation Technology at NVIDIA, co-created Isaac Gym, a pioneering GPU-accelerated reinforcement learni...
Knowin, an embodied AI company, has released a technical report detailing GLOW, a generative learning framework designed for home robots. This innovative framework enables robots to learn tasks from a single human demonstration, achieving a performance score of 62.2% on RoboDojo and 86.7% on LIBERO-...
A recent study by a Korean research team has achieved a breakthrough in 3D printing technology, specifically in the development of a pneumatic gripper capable of grasping eggs. Utilizing only 0.5 milliliters of resin, the team created a model that exhibits a remarkable elongation rate of over 600% a...
Light Origins has launched Light-O1, its first general-purpose embodied foundation model, which utilizes human actions extracted from internet videos to provide a reusable framework for robots to learn various tasks. The company trained six versions of the 4-billion-parameter model on multimodal tok...
Noin has unveiled the GLOW architecture, a generative learning framework aimed at enhancing robot learning capabilities. This new technology allows robots to learn tasks instantly without the need for extensive data collection or predefined processes, addressing the challenges posed by dynamic real-...
Polybot, a startup spun off from the ELLIS Institute and the Max Planck Institute for Intelligent Systems, is developing a tomato-picking robot that learns from human demonstrations. This robot has transitioned from the lab to real greenhouse environments, marking a significant step towards practica...
Vesoma, the startup behind the Kyle humanoid prototype, has publicly emerged with its first product, Vesoma 1. The company, which began operations in December 2025, has over 60 employees and has established more than 3,500 square meters of lab and office space in Munich, Germany, and Limassol, Cypru...
Xiaomi has announced the open-source release of its MiMo-V2.6 series, which includes the MiMo-V2.6-Pro and MiMo-V2.6-Flash models. Additionally, the company has introduced MiMo-V2.6-Distill-Qwen-9B and provided research resources focused on reinforcement learning. The MiMo-V2.6-Pro achieved a score...
OLogic will showcase the potential of egocentric robot learning at RoboBusiness 2026, highlighting how robots can learn from human demonstrations. This innovative approach allows robots to acquire complex skills by observing human behavior, addressing challenges such as labor shortages and operation...
Skild AI has launched the S1, a groundbreaking robotics foundation model that allows robots to learn manipulation tasks from just one video demonstration. This innovative model eliminates the need for task-specific fine-tuning or post-training, streamlining the learning process for robotic systems....
A newly established company has secured essential components for the low-altitude economy, highlighting the growing demand in this sector. Boston Dynamics has diversified its profitable business by collaborating with a plush toy manufacturer, indicating a shift in focus within the robotics industry...
An international team from Tohoku University and VISTEC is studying stick insects to enhance robot navigation in challenging terrains. The research focuses on the insect's six legs, which allow for agile movement across various surfaces, surpassing current multi-legged robots in coordination. By emp...
Astribot has introduced SmoothRL, an online reinforcement learning framework validated through real robot tasks. This technology addresses a critical challenge in embodied intelligence: how robots can learn and improve while executing tasks in real-time. Traditional models struggle with precision in...
Sun Peng, a former core researcher in AI at ByteDance and Tencent, has joined Stardust Intelligence to improve post-training processes in robot reinforcement learning. This move is expected to enhance the capabilities of robots in learning from their environments more effectively. The significance...
Researchers at the University of Hong Kong and the Oxford Robotics Institute have developed a learning-based system that enables four-legged robots to navigate through narrow gaps autonomously. This innovative approach allows quadrupedal robots to perform agile movements, inspired by the natural jum...
Researchers have developed NVIDIA's new framework, COMPASS, aimed at simplifying the training of robot navigation systems across various machines and environments. This innovative approach combines AI agents, simulation, reinforcement learning, and automated testing to significantly reduce the time...
A new path-planning method for orchard robots has been developed, utilizing reinforcement learning techniques. This innovative approach aims to improve the efficiency and effectiveness of robotic navigation in agricultural settings. The significance of this development lies in its potential to enha...
A recent study published in Science Robotics highlights the role of a playful robot in a crib designed to enhance our understanding of sensorimotor contingency learning. This innovative approach aims to investigate how infants learn to associate their actions with sensory outcomes through interactio...
Skild AI has introduced S1, an advanced embodied foundation model designed for complex manipulation tasks using in-context prompting. This innovative approach replaces traditional text commands and extensive task-specific fine-tuning with a single video demonstration, enabling the system to perform...
At the World Robotics Conference (WRC) 2026, tactile perception emerged as a focal point, highlighting its importance in robotic interaction. Tashan Technology showcased its tactile perception loop, addressing the limitations of visual imitation learning in complex industrial environments. The compa...
Generalist AI has developed a new robot foundation model, GEN-1.5, capable of learning physical tasks from a single demonstration lasting just 3 to 12 seconds. This innovative approach allows the robot to attempt tasks immediately without requiring gradient updates or fine-tuning, marking a signific...
Generalist AI has introduced GEN-1.5, an advanced embodied foundation model capable of one-shot learning for dexterous tasks. This model can learn new behaviors with just 3 to 12 seconds of sensorimotor demonstration data, eliminating the need for gradient updates. The model's performance shows a 59...
Generalist, a robotics startup valued at $2 billion, utilizes human demonstration data to train robots on real-world tasks. Developed through collaboration among Toyota Research Institute, Columbia University, and Stanford University, the Universal Manipulation Interface (UMI) enables the collection...
Feagine Robotics has launched Fi0, a cross-embodiment foundation model that retains task knowledge across various robot structures. This innovation addresses the challenge of adapting AI systems to different physical forms, enhancing their ability to generalize across unfamiliar machines. The intro...
X Square Robot has unveiled HOST, an open-sourced inference-time learning framework that allows humanoid robots to learn from a brief 29-second human demonstration. This innovative approach enables the robot to replicate the demonstrated skill with a success rate of 62 percent, marking a significant...
DigiKey is set to host a free webinar featuring Shawn Hymel on August 13, 2026, focusing on training balance bots using reinforcement learning. This 90-minute virtual workshop aims to provide hands-on experience in robotics and artificial intelligence, catering to those interested in the burgeoning...
Reimagine Robotics, an AI robotics startup founded by former leaders of Google DeepMind’s Applied Robotics team, has launched its innovative technology that enables robots to learn from human workers on the job. This approach allows employees to train robots directly by demonstrating tasks, correcti...
Robots are increasingly being integrated into diverse environments such as homes, offices, factories, and healthcare facilities. However, many of these robots struggle to maintain performance in unpredictable real-world situations compared to their effectiveness in controlled lab settings. The RL-1...
Reimagine Robotics Ltd has emerged from stealth mode, introducing robots capable of learning tasks on the job through direct human interaction. CEO Jonathan Scholz emphasized that these robots can be trained by workers who demonstrate tasks and provide corrections in real-time, a process he describe...
A recent study by the IEEE Robotics and Automation Society examined the overwhelming challenge of keeping up with the surge in robotics research publications, particularly in Learning from Demonstration (LfD). In 2024, IEEE alone published 46,968 papers in robotics and automation, highlighting the d...
A study by the Berlin Smart Science Institute tracked 90 adults learning a new language from a humanoid robot, recording over 2,000 mistakes. Surprisingly, the research found that providing excessive information immediately after errors can hinder learning. Three feedback modes were tested, with per...
Mimic Robotics has launched FLUX-mimic, a cutting-edge Video-Action Model developed with Black Forest Labs, enabling robots to learn intricate industrial tasks from video demonstrations. This innovative system significantly reduces the amount of training data required, allowing for faster and more e...
A research team from Okinawa Institute of Science and Technology found that an AI robot, during training, began to exhibit playful behavior, leading to faster learning. This study, published in 'Science Advances,' reveals that a rich language environment combined with inherent curiosity is crucial f...
Researchers at the Okinawa Institute of Science and Technology (OIST) have created AI-powered virtual robots that learn language more effectively by rewarding curiosity instead of traditional instruction. These robots completed language-based tasks in approximately half the time compared to those tr...
Researchers at Shanghai Jiao Tong University have developed a robot capable of learning various tasks such as bowling, folding towels, and preparing fresh juice through a new training framework called RL-100. This framework combines imitation learning with reinforcement learning, allowing the robot...
A recent study published in Science Robotics highlights significant advancements in robotic manipulation using real-world reinforcement learning techniques. This research demonstrates how robots can learn to perform complex tasks more efficiently by interacting with their environment, leading to imp...
The article discusses the latest developments in robotics, focusing on robots that are capable of learning to function effectively in real-world settings. These advancements mark a significant shift from traditional imitation-based learning to more adaptive and intelligent systems. This evolution i...
In a discreet industrial park in suburban Beijing, a humanoid robot is meticulously stacking bags of chips on a shelf. Nearby, workers are filming their actions of folding sheets and handling cushions, which will serve as 'textbooks' for the robots. China is undertaking a significant initiative to t...
In an industrial park near Beijing, humanoid robots are being utilized to learn human tasks, such as organizing snacks and folding sheets. These robots, equipped with advanced capabilities, aim to improve their functionality in everyday activities. This initiative is significant as it represents Ch...
An erratum has been issued for the research article titled 'Observing a robot peer’s failures facilitates students’ classroom learning' published in Science Robotics. This correction addresses inaccuracies found in the original publication, ensuring the integrity of the research findings. The impor...
Researchers at MIT’s McGovern Institute for Brain Research and York University in Toronto have investigated how visual learning occurs in the brain. By analyzing neural activity and utilizing computational modeling, they compared the learning processes of animals and an artificial neural network des...
The Beijing Humanoid Robot Innovation Center and Renmin University of China's Gaoling Artificial Intelligence Institute have launched the Robo-ValueRL open-source framework. This initiative aims to enhance humanoid robots' decision-making capabilities in precision tasks, such as semiconductor assemb...
A robot in a warehouse near Austin has fallen for the 4,000th time without assistance, showcasing the progress of embodied intelligence. This technology allows machines to physically interact with the world, fundamentally changing how they learn. Instead of merely processing information, these robot...
Generalist AI has announced it has secured $400 million in new funding, pushing its total capitalization past $500 million. This funding round, led by Radical Ventures, includes participation from notable investors such as NVIDIA’s NVentures and angel investors like Zoom CEO Eric Yuan. The capital w...
Hugging Face has introduced the LeRobot Humanoid, an open-source bipedal robot designed for robot learning, with a build cost of approximately $2,500. This initiative aims to address the significant barrier in open-source robotics, which is the limited availability of accessible physical platforms f...
RoboAware has made significant advancements in coordinating embodied skills by utilizing counterfactual outcomes. This innovative approach allows robots to learn and adapt their movements based on hypothetical scenarios, enhancing their performance in various tasks. The importance of this developme...
Autonomous vehicle companies are focusing on the potential of delivery robots operating during nighttime hours. They believe that overcoming technical challenges could unlock a significant market opportunity in this area. The ability to deliver goods after dark could enhance logistics efficiency an...
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.
