Artificial intelligence system for learning robotic control policies

This system teaches robots how to perform tasks by first letting them practice in a computer simulation. The simulation provides feedback, which an artificial intelligence system uses to improve the robot's control. After the robot learns in the simulation, its skills are fine-tuned using real-world trials. The claims specifically narrow the artificial intelligence to either an evolutionary strategy or a reinforcement learning system.

Why it matters: Since 2017, reinforcement learning and evolutionary algorithms have become significantly more robust and accessible, making the training and refinement of robotic control policies described here much more efficient and effective than when filed.

Status: Not in forceListed as no longer active. The specific reason is not in the record we hold.
How hard to build: SpecializedRobotic system, physics simulation, machine learning software

3 AI reinvention directions for this patent

See all three

AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.

Reinvent this with AI

Where this one came from

Every morning we go through millions of abandoned patents and pick out 15 worth your time. This was one of today’s.

This one

+10