Back to results
US 2021/0080966 A1Published application
Smart vehicle
OwnerHa Q. Traninventor
Life of this patent
Estimating the term…
Abstract
Smart car operations are detailed. The system uses neural network and population statistics to determine whether a vehicle operation is reasonably safe operation of the autonomous vehicle to guide the operation of cars as a group.
Claims20
A method for navigating a vehicle, comprising:
- using a trained neural network receiving data from lidar or long-range radar in a front of the car;
- a multi-focal camera in the front;
- one or more thermal imager(s) in the front or rear of the car;
- a processor coupled to the lidar, radar, multi-focal camera and thermal imagers, the neural network trained to make a driving decision for the vehicle;
- conforming to one or more traffic rules on driving speed or lane changing, and
- reasonably ignoring the one or more traffic rules during an imminent accident, wherein a reasonableness of the overruling is determined by featurizing a plurality of vehicle sensor parameters including vehicle speed or acceleration, vehicle location, wireless device use, turn signal use, time, traffic signal, lane departure, collision detection, pedestrian detection and traffic and weather conditions, and
- further training the neural network on traffic case law history by comparing the vehicle operation data and one or more traffic case law historical outcomes to make the driving decision.
4 more elements
(canceled)
A method for navigating a vehicle, comprising:
- using a trained neural network to make driving decisions for an autonomous street car;
- conforming to one or more traffic rules on driving speed or lane changing, and applying the neural network to park the vehicle by extending one or more extendable rolling wheels at a vehicle bottom to facilitate lateral parking movement into a parking spot as guided by side sensors including sonar and camera sensors.
Dependencies are inferred from the claim text.