Training, testing, and verifying autonomous machines using simulated environments

US11436484B2 · kind B2 · utility

Assignee

• NVIDIA Corporation · US

Inventors

• Clement Farabet · Mill Valley, US
• John Zedlewski · San Francisco, US
• Zachary Taylor · Moraga, US
• Greg Heinrich · Nice, FR
• Claire Delaunay · Menlo Park, US
• Mark Thomas Daly · San Francisco, US
• Matthew Campbell · Charlotte, US
• Curtis Beeson · Omaha, US
• Gary D. Hicok · Mesa, US
• Michael Cox · Menlo Park, US
• Rev Lebaredian · Austin, US
• Tony Tamasi · Los Gatos, US
• David Auld · San Jose, US

Key dates

• Filing date: Mar 27, 2019
• Grant date: Sep 6, 2022
• Expiry date: Jan 9, 2041

Classification

• Technology area (CPC G): Physics
• CPC primary: G06V20/56
• WIPO field: Computer technology
• WIPO sector: Electrical engineering

Abstract

In various examples, physical sensor data may be generated by a vehicle in a real-world environment. The physical sensor data may be used to train deep neural networks (DNNs). The DNNs may then be tested in a simulated environment—in some examples using hardware configured for installation in a vehicle to execute an autonomous driving software stack—to control a virtual vehicle in the simulated environment or to otherwise test, verify, or validate the outputs of the DNNs. Prior to use by the DNNs, virtual sensor data generated by virtual sensors within the simulated environment may be encoded to a format consistent with the format of the physical sensor data generated by the vehicle.