Articles tagged with "robot-control"
NVIDIA's Cosmos Policy helps robots predict what happens next
NVIDIA has introduced Cosmos Policy, a novel robot control framework that leverages large pretrained video prediction models to simplify decision-making in robotics. Unlike traditional robot policies that rely on separate perception, planning, and control modules and require extensive task-specific data, Cosmos Policy post-trains a pretrained video world model (Cosmos Predict) on robot demonstration data. This approach integrates robot actions, physical states, and task outcomes into a unified temporal representation, enabling the model to jointly predict the robot’s next actions, future states, and task success within a single architecture. This reduces architectural complexity and the need for large amounts of robot-specific training data. Benchmark tests demonstrate that Cosmos Policy achieves high success rates on multi-step robotic manipulation tasks, often matching or surpassing existing methods while using significantly fewer training demonstrations. A key advantage is its planning capability at inference time, allowing the model to generate and evaluate multiple candidate action sequences and select those with the best predicted outcomes over longer horizons. This strategic planning enables robots to perform complex tasks
roboticsrobot-controlAI-in-roboticsNVIDIA-Cosmos-Policyrobot-planningvideo-prediction-modelsrobotic-manipulationVideo: Humanoid robot knocked down playfully by US basketball star
During a Dallas Mavericks team event, NBA star Kyrie Irving playfully knocked down a 4’2” humanoid robot, the G1 model from Unitree Robotics, which had been demonstrating karate-style moves. The robot, dressed in a Mavericks jersey, attempted martial arts maneuvers before Irving gently shoved it, causing it to stumble and fall, creating a viral and humorous moment shared widely on social media. The incident highlighted both the entertainment value and current limitations of humanoid robots, with fans joking about the robot’s durability and praising Irving’s physical readiness post-injury. Unitree Robotics recently launched the world’s first humanoid robot app store, enabling users to control robots via phone and share actions like dance and martial arts. The G1 robot is designed for resilience in real-world environments, equipped with advanced sensor-driven perception and predictive motion control systems. It uses data from depth cameras, 3D LiDAR, and joint sensors to anticipate impacts and adjust posture before falling, allowing it to
robothumanoid-robotUnitree-Roboticsrobot-controlrobot-resiliencerobotics-technologyrobot-app-storeChina's Unitree launches world's first humanoid robot app store
China’s Unitree Robotics has launched what it calls the world’s first humanoid robot app store, enabling users to control and customize humanoid robots via smartphones. Announced on December 13, 2025, the Unitree Robotics Developer Platform serves as a centralized hub where users can access, share, and download action routines—ranging from martial arts and dance to custom training datasets—directly from their phones. This phone-first interface allows operators to guide robot movements remotely using the phone camera and trigger preloaded demonstrations, making humanoid robots more accessible and customizable much like smartphones and their app ecosystems. The platform encourages community participation by allowing users and developers to upload and adapt software and datasets, fostering rapid innovation and skill sharing among robots. This open, app-based approach could significantly accelerate the transition of robotics research into practical, everyday applications. Unitree, a prominent player in China’s advanced robotics sector, has gained national recognition, with its founder Wang Xingxing appointed to a key government robotics committee.
roboticshumanoid-robotsrobot-app-storeUnitree-Roboticsrobot-controlrobot-software-platformrobotics-innovationRobot Talk Episode 137 – Getting two-legged robots moving, with Oluwami Dosunmu-Ogunbi - Robohub
In episode 137 of the Robot Talk podcast, Claire interviews Oluwami Dosunmu-Ogunbi, an Assistant Professor of Mechanical Engineering at Ohio Northern University, about advances in bipedal robots capable of walking and climbing stairs. Dosunmu-Ogunbi’s research centers on control systems for bipedal locomotion and engineering education. Notably, she is the first Black woman to earn a PhD in Robotics from the University of Michigan. During her doctoral studies, she developed the "Biped Bootcamp," a technical document designed to introduce students to bipedal robotics. She is now adapting this material into an undergraduate curriculum aimed at providing both introductory and advanced coursework for junior and senior engineering students. The episode highlights her contributions to robotics education and the challenges involved in getting two-legged robots to move effectively.
roboticsbipedal-robotsrobot-controlrobot-locomotionrobot-educationautonomous-machinesrobotics-researchHumanoid robot uses human data to master cartwheels and sprints
Researchers at Cornell University have developed BeyondMimic, a novel framework enabling humanoid robots to perform complex, fluid human-like motions such as cartwheels, sprints, dance moves, and even Cristiano Ronaldo’s “Siu” celebration. Unlike traditional programming methods that require task-specific coding, BeyondMimic uses human motion capture data to train robots through a unified policy, allowing them to generalize and execute new tasks without prior training. This system leverages Markov Decision Processes and hyperparameters to seamlessly transition between diverse movements while preserving the style, timing, and expression of the original human actions. A key innovation in BeyondMimic is the use of loss-guided diffusion, which guides the robot’s real-time movements via differentiable cost functions, ensuring accuracy, flexibility, balance, and stability. The framework supports various real-world robotic controls such as path following, joystick operation, and obstacle avoidance, making it highly adaptable. The entire training pipeline is open-source and reproducible, providing a
roboticshumanoid-robotmotion-trackingmachine-learningrobot-controlartificial-intelligencerobotics-researchSkild AI is giving robots a brain - The Robot Report
Skild AI has introduced its vision for a generalized "Skild Brain," a versatile AI system designed to control a wide range of robots across different environments and tasks. This development represents a significant step in Physical AI, which integrates artificial intelligence with physical robotic systems capable of sensing, acting, and learning in real-world settings. Skild AI’s approach addresses Moravec’s paradox by enabling robots not only to perform traditionally "easy" tasks (like dancing or kung-fu) but also to tackle complex, everyday challenges such as climbing stairs under difficult conditions or assembling intricate items, tasks that require advanced vision and reasoning about physical interactions. Since closing a $300 million Series A funding round just over a year ago, Skild AI has expanded its team to over 25 employees and raised a total of $435 million. Physical AI is gaining momentum across the robotics industry, with other companies like Physical Intelligence pursuing similar goals of creating a universal robotic brain. This topic will be a major focus at RoboBusiness 202
robotroboticsartificial-intelligencephysical-AIrobot-controlmachine-learningautomationTackling the 3D Simulation League: an interview with Klaus Dorer and Stefan Glaser - Robohub
The RoboCup Soccer 3D Simulation League is a competition where teams control simulated Nao robots in an 11 versus 11 soccer match, with detailed motor-level control mimicking real robots. Unlike the 2D Simulation League, which focuses on simplified physics and team strategy, the 3D League aims for a more realistic robotic simulation. Currently, the league uses SimSpark, a simulator developed in the early 2000s that balances physical realism with the computational limitations of its time. However, SimSpark has limitations such as complexity, custom robot models, and communication protocols that hinder wider adoption and make it difficult to translate simulations to real robots. To address these issues, Stefan Glaser has been developing a new simulator based on the MuJoCo physics engine, which has recently become popular in machine learning communities due to its open-source availability and standardized model specifications. MuJoCo supports dynamic manipulation of the simulation environment, a key feature needed for RoboCup’s setup where agents join and form teams
roboticsrobot-simulationRoboCup3D-simulationNao-robotsrobot-controlrobotics-competitionGoogle rolls out new Gemini model that can run on robots locally
Google DeepMind has introduced Gemini Robotics On-Device, a new language model designed to run locally on robots without needing an internet connection. This model builds on the previous Gemini Robotics version by enabling direct control of robot movements through natural language prompts, allowing developers to fine-tune it for various applications. According to Google, Gemini Robotics On-Device performs nearly as well as its cloud-based counterpart and surpasses other unnamed on-device models in general benchmarks. In demonstrations, robots equipped with this local model successfully performed tasks such as unzipping bags and folding clothes. Although initially trained for specific tasks, the model was later adapted to work on different robot platforms, including the bi-arm Franka FR3, which managed to handle new scenarios and objects it had not encountered before. Additionally, Google DeepMind is releasing tools that allow developers to train robots on new tasks by providing 50 to 100 demonstrations using the MuJoCo physics simulator. This development aligns with broader industry trends, as companies like Nvidia, Hug
robotroboticsAIon-device-AIGoogle-DeepMindGemini-Roboticsrobot-control