Simulation driven robotic control of real robot(s)

US11938638B2 · GDM Holding LLC

Patent number
US11938638B2
Applicant
GDM Holding LLC
Filing date
2021-06-03
CPC class
B25J9/16
Pillar
Brain

Methods and systems for active utilization of a robotic simulator in the control of one or more real-world robots. The simulator is configured with a simulated environment reflecting the real environment based on sensor observations, and used to determine sequences of robotic actions via machine lea

Claims

Independent claims cover a method and system wherein a robotic simulator generates a simulated environment from real environment state data, applies simulated state data to ML models/rules to determine action sequences for real robots, and controls the real robots accordingly. Dependent claims detail sensor integration, task types (e.g., grasping), and synthetic data generation.

Technical analysis

The patent describes real-time integration of simulation with physical robot control, using perception data to update the sim environment and leveraging synthetic viewpoints or data not available in the real world for improved decision-making. CPC classifications focus on program-controlled manipulators and simulation aspects.

Commercial impact

Enables more reliable and adaptable robotic systems for complex tasks in dynamic environments, with potential applications in manufacturing, logistics, and service robotics. Assigned through Google/X Development lineage, indicating strong tech transfer potential in AI-driven automation.

Related patents

All patents · Source filing