www.magazine-industry-usa.com
16
'26
Written on Modified on
Yaskawa Electric Develops Agentic Robot System with Gemini Integration
Integration with Google DeepMind 's advanced generative AI enables the robot to assess on-site conditions, assemble work procedures, and execute them independently.
www.yaskawa.eu.com
Yaskawa Electric Corporation has developed an agentic robot system that integrates its autonomous artificial intelligence (AI) robot, MOTOMAN NEXT, with Google DeepMind’s advanced generative AI model, Gemini Robotics ER 1.6. The integrated framework enables the automation system to independently analyze on-site environmental conditions, compile structural work procedures, and execute operational tasks without manual, step-by-step programming. By providing high-level commands, such as requesting the sorting of unsorted parts, the system evaluates the target workspace and carries out the necessary material manipulation routines to handle industrial labor shortages.
Functional Separation of Decision-Making and Kinematic Execution
The architecture establishes a clear division of operational tasks between the cloud or edge computing brain and the physical robot body. The generative AI model is responsible for semantic reasoning and workflow synthesis—determining what tasks must be performed—while the MOTOMAN NEXT hardware platform translates those logical steps into precise on-site movements.
To bridge the gap between high-level generative decisions and real-world execution, the MOTOMAN NEXT robot comes equipped with three core factory automation service layers:
- Machine Vision Service: This sensing layer identifies workspace conditions, structural shapes, and the exact physical positions of target objects. It converts visual observations into raw spatial data to feed the generative AI's decision-making loops.
- Path Planning Service: Operating in response to high-level objective vectors received from the generative AI, this service calculates safe, collision-free movement paths through dense or changing factory environments.
- Force Feedback Service: This physical tracking loop monitors real-time contact forces and tactile gripping pressure to verify whether parts are securely held, preventing object damage or structural slips.
Autonomous Error Recovery and Enterprise Connectivity
The agentic robot platform incorporates software features designed to stabilize manufacturing up-time and simplify integration with existing corporate control networks. Traditional industrial robotics require rigid, continuous sequential programming that breaks if an unexpected part anomaly occurs.
The integrated generative AI layer monitors execution states continuously, enabling automated error recovery; if a part is dropped or misplaced during transport, the robot detects the situational change independently and recalculates a recovery sequence to restart the task without human intervention. Furthermore, the control platform can link directly with internal production management networks and enterprise software systems, allowing the robot to programmatically query inventory databases or automatically push parts-shortage alerts and order updates when it detects an insufficient component supply on the line.
Additional Context
This section details technical specifications and competitive benchmarking not included in the original news release.
Unlike traditional industrial controllers that rely on external inference PCs—introducing communication latency along the interface —the MOTOMAN NEXT features an embedded edge-computing graphics module that handles vision and path algorithms locally within the cabinet. Furthermore, standard vision-guided setups employ static programming templates that trigger fault stops if components become displaced. The integrated foundation model establishes continuous spatial reasoning for autonomous error recovery and transfers learned manipulation procedures across varied arm shapes without manual code retraining.
Edited by Romila DSilva, Induportals Editor, with AI assistance.
The agentic robot platform incorporates software features designed to stabilize manufacturing up-time and simplify integration with existing corporate control networks. Traditional industrial robotics require rigid, continuous sequential programming that breaks if an unexpected part anomaly occurs.
The integrated generative AI layer monitors execution states continuously, enabling automated error recovery; if a part is dropped or misplaced during transport, the robot detects the situational change independently and recalculates a recovery sequence to restart the task without human intervention. Furthermore, the control platform can link directly with internal production management networks and enterprise software systems, allowing the robot to programmatically query inventory databases or automatically push parts-shortage alerts and order updates when it detects an insufficient component supply on the line.
Additional Context
This section details technical specifications and competitive benchmarking not included in the original news release.
Unlike traditional industrial controllers that rely on external inference PCs—introducing communication latency along the interface —the MOTOMAN NEXT features an embedded edge-computing graphics module that handles vision and path algorithms locally within the cabinet. Furthermore, standard vision-guided setups employ static programming templates that trigger fault stops if components become displaced. The integrated foundation model establishes continuous spatial reasoning for autonomous error recovery and transfers learned manipulation procedures across varied arm shapes without manual code retraining.
Edited by Romila DSilva, Induportals Editor, with AI assistance.
