Adaptive, self-evolving learning and testing platform for self-driving and real-time map construction
US10803324B1 · kind B1 · utility
Assignee
Inventors
Key dates
| Filing date | Jan 3, 2018 |
| Grant date | Oct 13, 2020 |
| Priority date | — |
| Expiry date | Jan 3, 2038 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06V20/588
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
Computer systems and methods for training and/or testing a self-driving vehicle. The systems and methods, via in-car cameras and sensors, capture video of conditions surrounding the vehicle, driving data related to control of the vehicle, and driver behavior data representing actions by a human controlling the vehicle. The systems and methods analyze the captured video data and driving data to derive decisions a computer would make to control the vehicle. The systems and methods analyze the captured driver behavior data to determine decisions made by the human to control the vehicle. The systems and methods evaluate the derived computer decisions in comparison to the determined human decisions to detect data indicating discrepancies between the derived computer decisions and the determined human decisions for controlling the vehicle. The systems and methods provide the detected data as input to a model for training and/or testing a self-driving vehicle.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.