YASKAWA Electric Corporation (Head Office: Kitakyushu, Fukuoka, Japan; Representative Director, Chairman of the Board and President: Hiroshi Ogasawara) is pleased to announce the development of an “Agentic Robot System.” This system integrates the AI robot “MOTOMAN NEXT” with Google Deepmind’s Gemini Robotics ER 1.6 model Google DeepMind’s embodied reasoning model enables our robots to assess on-site conditions, devise its own work procedures, and execute them autonomously.

　“Agentic” refers to a robot that does not wait for detailed instructions from humans, but rather thinks, judges, and acts on its own to achieve a given objective. With this system, even without humans creating detailed programs, simply giving an instruction such as “Please sort these parts” allows the robot to recognize the situation and autonomously carry out a series of tasks. Against the backdrop of a worsening labor shortage, Yaskawa has been working to pioneer areas where automation has traditionally been difficult, and this system represents a major step forward in realizing that goal.　

　In recent years, efforts to utilize generative AI as the “brain” of robots have been gaining momentum. However, no matter how sophisticated the decisions made by generative AI may be, true agentic autonomy cannot be achieved unless those decisions can be accurately executed as actual tasks.

　A key feature of the AI robot “MOTOMAN NEXT” is that it comes standard with functions essential for autonomous operation built directly into the robot itself. Specifically, services such as machine vision, path planning, and force sensing bridge the gap between generative AI decisions and on-site operations.

• Seeing (Machine Vision Service): It accurately recognizes the on-site conditions as well as the position and shape of objects, translating the generative AI’s decisions into the information necessary for the robot’s tasks.

• Planning its own movements (Path Planning Service): In response to objectives received from generative AI—such as “pick up this object”—the robot itself meticulously plans and accurately executes a safe movement path that avoids surrounding obstacles.

• Touch and sense (Force Sensing Service): It confirms the sensation of contact and the grip status to determine whether the object is being held securely.

　These capabilities are the result of Yaskawa’s long-standing expertise in robot control technology. By integrating Google DeepMind’s Gemini Robotics ER 1.6” model with the AI robot “MOTOMAN NEXT,” which is equipped with these autonomous functions, this system consistently delivers everything from advanced decision-making by generative AI (the “brain”) to precise physical movements on the shop floor (the “body”).

　In other words, generative AI determines “what needs to be done,” while the AI robot “MOTOMAN NEXT” is responsible for “how to move to achieve it”—this clear division of roles establishes an agentic robot system. Yaskawa believe that the sophistication of the “body,” which reliably translates the generative AI’s decisions into action, is of paramount importance.

• No need for detailed programming

  Based on general instructions, the generative AI independently constructs the work procedures. No prior teaching is required.

• Automatic recovery from errors during operation

  Even if an unexpected situation occurs—such as dropping an object during transport—the robot independently recognizes the situation and repeats the task.

• Integration with production management systems and other systems

  Operations can be conducted by linking on-site work with internal systems, such as sharing order information when a parts shortage is detected.

　Yaskawa will further accelerate its efforts toward the practical application and societal implementation of this system. Yaskawa will provide an update regarding specific application areas and availability as soon as preparations are complete.

　Through this technology, Yaskawa will contribute to the further automation and autonomy of manufacturing and logistics sites, as well as the societal implementation of agent-based robot systems.

For inquires, contact:

Corporate Communications Dept.

YASKAWA Electric Corporation

E-mail: pr@yaskawa.co.jp