en.Wedoany.com Reported - Yaskawa Electric Corporation has developed an agent robot system that integrates the autonomous artificial intelligence (AI) robot MOTOMAN NEXT with Google DeepMind's generative AI model Gemini Robotics ER 1.6. This framework enables the automation system to independently analyze on-site environments, formulate work processes, and execute operational tasks without step-by-step manual programming. Upon receiving high-level instructions (such as sorting unclassified parts), the system assesses the work area and performs material handling procedures to address industrial labor shortages.
The architecture establishes a clear division of tasks between the "brain" (cloud or edge computing) and the robot's physical body. The generative AI model handles semantic reasoning and workflow synthesis, determining which tasks need to be performed; the MOTOMAN NEXT hardware platform translates logical steps into precise movements in the workplace. To bridge the gap between high-level decision-making and real-world execution, the robot is equipped with three core service layers: machine vision services identify workspace conditions and object positions, path planning services calculate collision-free motion paths, and force feedback services monitor contact forces in real-time to verify part gripping status.
The agent robot platform integrates software designed to stabilize manufacturing uptime and simplify enterprise control network integration. Traditional industrial robots require strict sequential programming, and programs are interrupted when part anomalies occur. The integrated generative AI layer continuously monitors execution status, enabling automated error recovery: if a part is dropped or misplaced, the robot independently detects and recalculates the recovery sequence, restarting the task without human intervention. The control platform can directly connect to internal production management networks, allowing the robot to query inventory databases or automatically issue alerts when parts are in short supply.
Unlike traditional industrial controllers that rely on external inference PCs, the MOTOMAN NEXT is equipped with an integrated edge computing graphics module that processes vision and path algorithms locally inside the control cabinet. Standard vision-guided configurations use static programming templates, where component displacement triggers fault shutdowns, while the integrated foundation model establishes continuous spatial reasoning for autonomous error recovery and can transfer learned operation procedures to different arm configurations without manual code retraining.
