Humanized steering model for automated vehicles

Patent metadata

Field	Value
Publication number	US-9914475-B2
Application number	US-201514874760-A
Country	US
Kind code	B2
Filing date	Oct 5, 2015
Priority date	Oct 5, 2015
Publication date	Mar 13, 2018
Grant date	Mar 13, 2018

Abstract

A humanized steering system for an automated vehicle includes one or more steering-wheels operable to steer a vehicle, an angle-sensor configured to determine a steering-angle of the steering-wheels, a hand-wheel used by an operator of the vehicle to influence the steering-angle and thereby manually steer the vehicle, a steering-actuator operable to influence the steering-angle thereby steer the vehicle when the operator does not manually steer the vehicle, a position-sensor operable to indicate a relative-position an object proximate to the vehicle, and a controller. The controller is configured to receive the steering-angle and the relative-position, determine, using deep-learning techniques, a steering-model based on the steering-angle and the relative-position, and operate the steering-actuator when the operator does not manually steer the vehicle to steer the vehicle in accordance with the steering-model, whereby the vehicle is steered in a manner similar to how the operator manually steers the vehicle.

First claim

We claim:

1. A humanized steering system for an automated vehicle, said system comprising:
one or more steering-wheels operable to steer a vehicle;
an angle-sensor configured to determine a steering-angle of the steering-wheels;
a hand-wheel used by an operator of the vehicle to influence the steering-angle and thereby manually steer the vehicle;
a steering-actuator operable to influence the steering-angle thereby steer the vehicle when the operator does not manually steer the vehicle;
a position-sensor operable to indicate a relative-position an object proximate to the vehicle; and
a controller configured to
receive the steering-angle and the relative-position,
determine, using deep-learning techniques, a steering-model based on the steering-angle and the relative-position while the operator is using the hand-wheel to manually steer the vehicle, and
operate the steering-actuator when the operator does not manually steer the vehicle to steer the vehicle in accordance with the steering-model, whereby the vehicle is steered in a manner similar to how the operator manually steers the vehicle.

2. The system in accordance with claim 1 , wherein the steering-model outputs a desired-angle of steering-wheels to steer the vehicle when the operator does not manually steer the vehicle.

3. The system in accordance with claim 1 , wherein the steering-model is characterized by a binary-vector composed of a plurality of bits indicative of parameters of the steering-model.

4. The system in accordance with claim 3 , wherein the binary-vector includes a parameter indicative of a weather-condition.

5. The system in accordance with claim 4 , wherein the weather-condition includes sunny, cloudy, foggy, rainy, snowy, windy, and temperature.

6. The system in accordance with claim 3 , wherein the binary-vector includes a parameter indicative of a vehicle-state.

7. The system in accordance with claim 6 , wherein the vehicle-state includes lane-position, lateral-acceleration, longitudinal-acceleration, vehicle-speed, and the steering-angle.

8. The system in accordance with claim 3 , wherein the binary-vector includes a parameter indicative of a vehicle-specification.

9. The system in accordance with claim 8 , wherein the vehicle-specification includes vehicle-size, vehicle-mass, steering-response-time, and engine-power.

10. The system in accordance with claim 3 , wherein the binary-vector includes a parameter indicative of an object-characteristic.

11. The system in accordance with claim 10 , wherein the object-characteristic includes object-location, object-vector, object-classification, and lane-position.

12. The system in accordance with claim 3 , wherein the binary-vector includes parameters indicative of a weather-condition, a vehicle-state, a vehicle-specification, and an object-characteristic.

Assignees

• Delphi Tech Inc

Inventors

• Sun Ludong
• Laur Michael H
• Wieskamp Jonathan L
• Yan Miao

Classifications

• G06N3/048

  Activation functions · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G01S2013/9323

  Alternative operation using light waves · CPC title

• G01S13/931

  of land vehicles · CPC title

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title