Humanoid Robotics and Embodied AI: Daily Wrap
Tesla’s Optimus Progress: AI-Driven Martial Arts and 2026 Mass Production Goal
Tesla’s humanoid robot, Optimus, has showcased new capabilities, with CEO Elon Musk confirming that recent “sparring” demonstrations were driven by artificial intelligence, not remote control. The video footage, which circulated on social media, showed the robot practicing martial arts with human staff members, responding to movements in real-time while tethered for safety.
This development aligns with the company’s aggressive production timeline. Tesla’s official AI account recently announced plans to launch the third-generation Optimus robot by the end of 2025, with mass production slated to begin in 2026. Musk has previously estimated that annual production could reach one million units before 2030, underscoring the company’s elevated strategic focus on the robotics division.
The new iteration, Optimus V3, is expected to feature hands with dexterity comparable to a human’s. The company’s renewed focus on robotics follows a period of perceived slowdown earlier in the year, with internal documents and analyst reports suggesting a continuous increase in investment to meet the ambitious production and delivery goals.
NVIDIA Unveils Open Foundation Models and Physics Engine to Accelerate Humanoid Development
NVIDIA has made a significant move to accelerate the research and development of humanoid robots by unveiling a suite of open-source AI models and simulation tools. This release aims to provide developers with a cohesive platform for creating more adaptable and intelligent robotic systems.
Key components of the announcement include:
- Newton Physics Engine: Now open-source and GPU-accelerated, this engine is designed to handle the complex physics required for realistic humanoid simulation, such as maintaining balance and performing intricate manipulation tasks. The engine was co-developed with Google DeepMind and Disney Research and is managed by the Linux Foundation.
- Isaac GR00T N1.6: An updated open robot foundation model that integrates with NVIDIA Cosmos™ Reason, a reasoning vision-language model built for physical AI. This model acts as the robot’s “deep-thinking brain,” enabling it to break down vague instructions into step-by-step plans, utilizing common sense and prior knowledge to generalize across various tasks.
- Cosmos World Foundation Models (WFMs): These models generate diverse, photorealistic synthetic datasets at scale, which is crucial for training physical AI models faster and safer in a simulated environment before deployment in the real world.
The company views these tools—Isaac GR00T as the robot’s brains, Newton as the simulator for its body, and NVIDIA Omniverse as the training ground—as the “three computers” necessary to transition robots from research into everyday life. Early adopters of the new platform include major research institutions like ETH Zurich and the Technical University of Munich, alongside leading robotics firms.
Seven-Eleven Japan Partners with Telexistence to Develop Humanoid Store Workers
In a major commercial application of humanoid technology, Seven-Eleven Japan, the Japanese arm of the global convenience store chain, has announced a partnership with robotics company Telexistence to develop a humanoid robot for in-store operations. The initiative is a direct response to ongoing staffing challenges and aims to significantly boost store efficiency.
The new humanoid robots, named “Astra,” will be powered by generative AI, utilizing a vision-language-action foundation model. The companies project that Astra robots will begin working in stores by 2029. This is an expansion of an existing relationship, as Seven-Eleven Japan is already piloting Telexistence’s “Ghost” beverage restocking robot in select Tokyo locations.
The partnership’s goals extend beyond simple task automation. By having robots handle routine operations, the companies hope to allow human employees to focus on customer services and other tasks that require a human touch, thereby strengthening the store’s appeal. The sheer scale of Seven-Eleven Japan’s network, which includes approximately 20,000 stores, is expected to provide an unparalleled volume of training data, accelerating the practical implementation of AI-powered humanoid robots at scale.
Humanoid Global Invests in Apptronik to Boost Apollo Commercialization
The financial backing for the humanoid robotics sector continues, as Humanoid Global Holdings Corp. announced a strategic investment in Apptronik, a prominent developer of AI-powered humanoid robots. The investment, made through a private fund, is intended to enhance Apptronik’s market position and accelerate the commercialization of its flagship robot, Apollo.
Apptronik’s Apollo robot, which is designed with a human-centered approach, is currently being expanded into multiple commercial sectors. Target applications include manufacturing, logistics, retail, healthcare, and home environments. This financial injection is expected to further the development and deployment of Apollo, aiming to transform collaborative workspaces where humans and humanoids work side-by-side.
Japan Sets Ambitious 2030 Moonshot Goal for Multipurpose Humanoid Robots
The Japanese government has signaled a national commitment to the humanoid robotics sector by establishing a new, ambitious goal under its Moonshot Research and Development Program. The revised goal focuses exclusively on developing multipurpose humanoid robots with physical and decision-making capabilities equal to or greater than those of humans.
The government’s aim is to have a prototype capable of handling complex tasks with human assistance by 2030. This initiative is viewed as a crucial step for laying the technological foundation that will spur private-sector investment and further AI research. The policy is motivated by a desire for robots to function across diverse environments, including households, manufacturing plants, and disaster-hit areas, and is seen as a strategic response to future labor shortages.
“As progress is happening faster than our initial assumption, we revised the goal to a more ambitious one,” said an official of the Cabinet Office, highlighting the accelerating pace of technological development in the field.