A single destination for timely, editor-curated robotics news from around the world.
Robotera's open-source VPP2 has surpassed RoboDojo's dual-arm benchmark by achieving a 58.5% zero-shot success rate on Real ALOHA Arms. This success was made possible by training the model to predict video sequences before executing robot actions, demonstrating a significant advancement in robotic c...
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-...
Figure has unveiled Helix 2.5, a neural network that allows its humanoid robot to perform three household tasks across 30 unfamiliar homes without prior data collection or model adaptation. This innovation marks a significant leap in robotic generalization, increasing zero-shot task success from 9%...
Figure has unveiled Helix 2.5, a groundbreaking neural network designed for humanoid robots. Unlike previous models, Helix 2.5 can autonomously perform tasks in unfamiliar environments without prior data collection. This innovation was tested in 30 homes in the Bay Area, showcasing the robot's abili...
Reward AI has officially launched its foundation model, OM-1, which aims to revolutionize robot manipulation by learning directly from human demonstrations. This approach seeks to eliminate the need for teleoperation and on-robot training, addressing the limitations of traditional methods that can d...
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...
The research published in Science Robotics introduces ZEST, a novel approach for zero-shot embodied skill transfer in athletic robots. This method allows robots to learn and execute complex athletic skills without prior specific training on those tasks. ZEST is significant as it enhances the adapta...
Dyna Robotics has introduced Dyna-2, a world-action model (WAM) pre-trained on one million hours of human video data. This development provides significant evidence for a human-to-robot transfer scaling law, demonstrating that increased exposure to human video enhances a robot's ability to perform t...
Sunday Robotics has introduced ACT-2, a foundation model that claims a 99.1% zero-shot success rate in laundry folding across diverse environments. This achievement is significant as it required no per-home data collection or expert demonstrations, showcasing the model's ability to generalize effect...
Jiying Technology has unveiled its Jiying 2.0 physics foundation model, which is capable of zero-shot generalization across various geometries, materials, and boundary conditions. This model represents a significant advancement in physics AI, particularly for engineering simulations, and was announc...
Lumos Robotics says its Prime R0 industrial embodied AI model has achieved the highest overall score on the latest MolmoSpaces leaderboard, outperforming larger models from competitors including Nvidia and research teams from the United States. The Chinese robotics company said its 2.8-billion-param...
A new framework named XRZero-G0 has been introduced to enhance the quality of data collection and training for embodied artificial intelligence, eliminating the need for robotic assistance. This innovative approach aims to streamline the process of gathering high-quality data, which is crucial for d...
A research team at ETH Zurich has unveiled a groundbreaking neural network simulator designed to enhance robotic capabilities by enabling robots to visualize actions internally before executing them on physical hardware. This innovative framework, known as the Robotic World Model (RWM), facilitates...
Sanctuary AI has achieved a significant breakthrough in robotics by successfully demonstrating zero-shot transfer for in-hand manipulation. This innovative technique enables a policy that was entirely trained in a simulated environment to effectively reorient physical objects using actual hardware,...
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.
