Humanoid Robotics and Physical AI: The Daily Wrap
The field of humanoid robotics saw significant movement across policy, corporate strategy, and safety this week, underscoring the rapid transition from lab prototypes to commercial deployment. Key developments include a major government funding shift in Japan toward multipurpose humanoids, a significant acquisition in the mobile robotics space by a humanoid developer, and critical new research highlighting cybersecurity risks in commercially available bots.
Japan’s Government Shifts Major R&D Fund Focus to Multipurpose Humanoid Robots
The Japanese government has announced a strategic refocus of its Moonshot Research and Development Program, setting a new goal to develop multipurpose humanoid robots by 2030. This initiative, backed by a substantial ¥400 billion fund, previously supported general robotics but will now specifically target humanoids capable of coexisting with people in everyday environments, including homes, factories, and disaster zones.
The Cabinet Office has established this new goal amid rapid global advances in AI and robotics, particularly in the U.S. and China. The program aims to boost private investment, accelerate AI research, and strengthen Japan’s global competitiveness in the burgeoning field of physical AI.
The focus is on creating robots with human-like capabilities and autonomy, enabling them to perform a wide range of tasks and seamlessly integrate into existing human infrastructure. This policy shift reflects a growing national commitment to leveraging humanoid technology as a solution for demographic challenges, such as labor shortages.
- New Goal: Develop multipurpose humanoid robots by 2030.
- Funding: Part of the government’s ¥400 billion Moonshot R&D Program.
- Target Environments: Homes, factories, and disaster zones, focusing on seamless human coexistence.
NEURA Robotics Acquires ek robotics to Scale Mobile and Cognitive Capabilities
NEURA Robotics, a German manufacturer specializing in cognitive robots, including its 4NE1 humanoid, announced the acquisition of ek robotics, a long-standing German specialist in driverless transport systems (DTS), such as Automated Guided Vehicles (AGVs) and Autonomous Mobile Robots (AMRs). The terms of the acquisition were not disclosed, but the move is described by NEURA’s CEO, David Reger, as “the dawn of a new dimension in mobile robotics”.
The acquisition significantly expands NEURA’s expertise in mobile robotics by incorporating ek robotics’ 60 years of experience, a team of over 300 employees, and a strong customer base. The strategic goal is to combine NEURA’s innovative cognitive technologies and its proprietary cloud platform, Neuraverse, with ek robotics’ proven intralogistics solutions. This integration is intended to speed up the market introduction of new innovations and create a global offering that sets new industry standards.
While ek robotics will continue to operate its core business and established brand, NEURA plans to integrate cognitive and AI technologies into the AGV portfolio, allowing all systems to become part of the central platform for continuous learning and improvement. The move also leverages NEURA’s global sales network to accelerate ek robotics’ growth beyond its primary European market.
Google DeepMind Unveils Gemini Robotics 1.5, Enabling Cross-Embodiment Skill Transfer
In a major AI breakthrough for physical agents, Google DeepMind introduced Gemini Robotics 1.5, its most capable vision-language-action (VLA) model to date, which unlocks “agentic experiences with advanced thinking” for robots. The model is designed to turn visual information and natural language instructions into motor commands, allowing robots to perform complex, multi-step tasks.
A key advancement of Gemini Robotics 1.5 is its ability to learn across different embodiments. The model can transfer motions and skills learned on one type of robot to a completely different one, eliminating the need to specialize the model for each new hardware design. For example, a task presented only to the ALOHA 2 robot during training can now be performed by Apptronik’s humanoid robot, Apollo, and vice versa. This cross-embodiment capability is expected to dramatically accelerate the development and deployment of new robot behaviors.
The system also features an embodied reasoning model, Gemini Robotics-ER 1.5, which acts as a high-level brain for orchestrating activities, excelling at planning, logical decision-making, and state-of-the-art spatial understanding. This model can even natively call tools like Google Search to gather information, adding a layer of transparency by showing its thinking process before taking action.
Cybersecurity Researchers Expose Alarming Vulnerabilities in Unitree G1 Humanoid
A new research paper has brought a critical safety concern to the forefront of commercial robotics, revealing alarming cybersecurity vulnerabilities in the Unitree G1 humanoid robot. Researchers from Alias Robotics, along with independent security experts, conducted a systematic security assessment and found that the robot, despite having what was described as a “most mature security architecture,” can be compromised.
The findings indicate that a compromised robot can be transformed into a “bidirectional attack vector” and an “Autonomous Active Threat Platform.” The research demonstrated the feasibility of deploying a Cybersecurity AI (CAI) agent onto the robot, which, powered by large language models (LLMs), could systematically execute a four-phase penetration test without human intervention. This included reconnaissance, identifying network connections, analyzing vulnerabilities, and preparing for exploitation, such as command injection attacks.
The study highlights a growing and tangible threat as sophisticated, AI-enabled humanoid robots begin to enter workplaces, hospitals, and factories, underscoring the urgent need for robust security protocols to prevent these helpful assistants from becoming covert surveillance nodes or active cyber weapons.
Diligent Robotics Expands AI Advisory Board to Guide Hospital Robot Deployment
Diligent Robotics, the company behind the Moxi mobile manipulator robot currently deployed in U.S. hospitals, announced the addition of two prominent robotics experts to its AI advisory board. The new members, a professor of robotics and a distinguished engineer specializing in robotic perception and autonomy, will provide strategic counsel to guide the company’s next generation of AI development.
The advisory board’s mission is to ensure that Diligent’s robots, which assist nurses with routine tasks like delivering medications and lab samples in over 25 hospitals, are grounded in responsible practices as the company continues to scale its fleet. This focus on governance and ethical AI development is crucial for robots operating in complex, human-facing environments like healthcare, where real-time adaptation and reliable, socially intelligent interaction are paramount.
The company emphasized that the guidance from the board will help their robots continue to lead the field in how they interact, adapt, and learn from the people they support. The goal is to improve workflow efficiency and help prevent burnout among hospital staff by automating rote tasks.