Granted Patent B2

US 12,246,739 · App. 17/883,540 · Granted Mar 11, 2025

Inventors: Zhaobo K. Zheng (San Jose, CA); Teruhisa Misu (San Jose, CA); Kumar Akash (Milpitas, CA)

Assignee: Honda Motor Co., Ltd.

B60W50/10G05B13/0265G05B13/04B60W2050/0075B60W2420/403B60W2540/221B60W2540/225

Abstract

Claims (38)

1. An adaptive driving style system, comprising:

a set of two or more sensors receiving two or more sensor signals grouped into one or more training portions and a validation portion, wherein one or more of the training portions is grouped into one or more training sub-portions and classification sub-portion;

a memory storing one or more instructions;

a processor executing one or more of the instructions stored on the memory to perform:

training a trust model using one or more of the training sub-portions as input, wherein the trust model is utilized to classify or label the classification sub-portion based on the one or more training sub-portions, wherein after classification of the classification sub-portion, the processor reassigns one or more of the training sub-portions as the classification sub-portion and trains the trust model using one or more of the training sub-portions until each of the sub-portions has been designated as the classification sub-portion at least once;

training a preference model using the trust model and two or more of the sensor signals as input; and

generating a driving style preference based on an adaptive driving style model including the trust model and the preference model; and

a controller controlling one or more of throttle, braking and steering of the vehicle in accordance with the driving style preference while the vehicle is controlled in an autonomous driving mode.

2. The adaptive driving style system of claim 1 , wherein a sensor of two or more of the sensors is an eye gaze sensor, a physiological sensor, a grip sensor, or an image capture device.

3. The adaptive driving style system of claim 1 , wherein the two or more sensor signals are received from a set of users subjected to a driving simulation or a controlled driving environment.

4. The adaptive driving style system of claim 3 , wherein the processor trains the trust model using two or more of the sensor signals and a set of ground-truth preferences from the set of users as input.

5. The adaptive driving style system of claim 3 , wherein the processor trains the preference model using two or more of the sensor signals, the trust model, and a set of ground-truth preferences from the set of users as input.

6. The adaptive driving style system of claim 1 , wherein the validation portion is utilized to validate the preference model.

7. A computer-implemented adaptive driving style method, comprising:

receiving two or more sensor signals from a set of two or more sensors, wherein the two or more sensor signals are grouped into one or more training portions and a validation portion, wherein one or more of the training portions is grouped into one or more training sub-portions and a classification sub-portion;

training a trust model using one or more of the training sub-portions as input, wherein the trust model is utilized to classify or label the classification sub-portion based on the one or more training sub-portions, and after classification of the classification sub-portion, reassigning one or more of the training sub-portions as the classification sub-portion and training the trust model using one or more of the training sub-portions until each of the sub-portions has been designated as the classification sub-portion at least once;

training a preference model using the trust model and two or more of the sensor signals as input;

generating a driving style preference based on an adaptive driving style model including the trust model and the preference model; and

controlling one or more of throttle, braking and steering of the vehicle in accordance with the driving style preference while the vehicle is controlled in an autonomous driving mode.

8. The computer-implemented adaptive driving style method of claim 7 , wherein a sensor of two or more of the sensors is an eye gaze sensor, a physiological sensor, a grip sensor, or an image capture device.

9. The computer-implemented adaptive driving style method of claim 7 , wherein the two or more sensor signals are received from a set of users subjected to a driving simulation or a controlled driving environment.

10. The computer-implemented adaptive driving style method of claim 9 , comprising training the trust model using two or more of the sensor signals and a set of ground-truth preferences from the set of users as input.

11. The computer-implemented adaptive driving style method of claim 9 , comprising training the preference model using two or more of the sensor signals, the trust model, and a set of ground-truth preferences from the set of users as input.

12. The computer-implemented adaptive driving style method of claim 7 ,

wherein the two or more sensor signals are grouped into one or more training portions and a validation portion; and

comprising training the trust model using one or more of the training portions.

13. An adaptive driving style system, comprising:

a set of two or more runtime sensors receiving two or more runtime sensor signals grouped into one or more training portions and a validation portion, wherein one or more of the training portions is grouped into one or more training sub-portions and a classification sub-portion;

a storage drive storing an adaptive driving style model;

a memory storing one or more instructions;

a processor executing one or more of the instructions stored on the memory to perform generating a runtime driving style preference using the adaptive driving style model and two or more of the runtime sensor signals;

wherein the adaptive driving style model is trained by:

training a trust model using one or more of the training sub-portions as input, wherein the trust model is utilized to classify or label the classification sub-portion based on the one or more training sub-portions, wherein after classification of the classification sub-portion, the processor reassigns one or more of the training sub-portions as the classification sub-portion and trains the trust model using one or more of the training sub-portions until each of the sub-portions has been designated as the classification sub-portion at least once;

training a preference model using the trust model and two or more of the runtime sensor signals as input; and

generating a driving style preference based on an adaptive driving style model including the trust model and the preference model; and

a controller controlling one or more of throttle, braking and steering of the vehicle in accordance with the driving style preference while the vehicle is controlled in an autonomous driving mode.

14. The adaptive driving style system of claim 13 , wherein a runtime sensor of the two or more of the runtime sensors is an eye gaze sensor, a physiological sensor, a grip sensor, or an image capture device.

15. The adaptive driving style system of claim 13 , wherein the two or more runtime sensor signals are received from a set of users subjected to a driving simulation or a controlled driving environment.

Assignments (1)

Continuity (1)

References Cited (129)