Predicting an interface control action of a user with an in-vehicle user interface

• US 10,535,004 B2
• Filed: 12/06/2014
• Issued: 01/14/2020
• Est. Priority Date: 12/19/2013
• Status: Active Grant

• Current Assignee: Daimler AG (Mercedes-Benz Group AG)
• Original Assignee: Daimler AG (Mercedes-Benz Group AG)
• Inventors: Kreis, Marius, Kruzic, David, Mosalem, Khaled, Smiroldo, Rigel
• Primary Examiner(s): Woolwine, Shane D
• Application Number:
  US15/105,796
• Publication Number:
  US 20160321545A1
• Time in Patent Office: 1,865 Days
• Field of Search: 706 21, 706 23, 706 46
• CPC Class Codes: G05B 13/02 electric
• CPC Class Codes: G05B 13/048 using a predictor
• CPC Class Codes: G06F 3/04842 Selection of displayed objects or displayed text elements G06F3/0482 takes precedence
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06N 20/20 Ensemble learning
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition

Abstract

A method for predicting an interface control action of a user with an in-vehicle user interface involves collecting and storing data. The data can be vehicle data about the vehicle and its environment collected from at least one sensor of the vehicle and user data about user interactions with the user interface and/or different applications inside the vehicle. Likelihoods are assigned to at least two possible interface control actions by the user based on the collected and stored data. At least one most likely interface control action is determined from the likelihoods and the user is presented with the at least one most likely interface control action so that it is selectable and performable with one single user interaction with the user interface.

Claims

1. A method for predicting an interface control action of a user with an in-vehicle user interface, the method comprising:

collecting and storing data, wherein the data includes at least

vehicle data about the vehicle and the vehicle'"'"'s environment collected from at least one sensor of the vehicle, and

user data about user interactions with the user interface and/or different applications inside the vehicle;

assigning likelihoods to at least two possible interface control actions by the user based on the collected and stored data;

determining at least one most likely interface control action from the likelihoods; and

providing the user with the at least one most likely interface control action in a manner so that the at least one most likely interface control action is selectable and performable with one single user interaction with the user interface,

wherein a classifier calculates likelihoods for the at least two possible interface control actions based on the collected and stored data and these calculated likelihoods are then assigned to the at least two possible interface control actions,

wherein the classifier is an ensemble classifier with several individual classifiers, where each individual classifier is built on a data chunk of the collected and stored data,

wherein each data chunk contains all data for a time interval specific for each data chunk and a time interval assigned to at least two data chunks overlap with each other, and

wherein the overlap and a data chunk width controls when new data chunks are created.

2. The method of claim 1, wherein a larger likelihood is assigned to frequently used interface control actions compared to less frequently used interface control actions when assigning the likelihoods.

3. The method of claim 1, wherein a larger likelihood is assigned to recently used interface control actions compared to less recently used interface control actions when assigning the likelihoods.

4. The method of claim 1, wherein only data younger than a given age are taken into account when assigning the likelihoods.

5. The method of claim 1, wherein the step of collecting and storing data comprises collecting and storing of data about the user from and to a network external to the vehicle.

6. The method of claim 1, wherein the data chunks are weighted with a bias depending on an age of the collected data contained so that more weight is assigned to data chunks with recent data compared to less recent data.

7. The method of claim 6, wherein each data chunk contains all data stemming from a time interval specific for each data chunk, where time intervals assigned to one data chunk can overlap with time intervals assigned to other data chunks, so that specific data can be part of more than one data chunk.

9. The method of claim 1, wherein the several individual classifiers include a primary classifier that generates a first estimate of likelihoods and a further classifier to supplement the first estimate of likelihoods.

10. The method of claim 9, wherein the primary classifier is a naï

ve Bayes classifier and the further classifier is a random forest classifier.

8. An in-vehicle user interface, comprising:

a memory configured to collect and store data, wherein the data includes at least

vehicle data about the vehicle and the vehicle'"'"'s environment collected from at least one sensor of the vehicle, and

user data about user interactions with the user interface and/or different applications inside the vehicle;

a processor coupled to the memory and configured to

assign likelihoods to at least two possible interface control actions based on the collected and stored data, and

determine at least one most likely interface control action from the likelihoods; and

a display unit configured to provide the user with a set of the at least one most likely interface control action so that it is selectable and performable with one single user interaction with the user interface,

wherein a classifier calculates likelihoods for the at least two possible interface control actions based on the collected and stored data and these calculated likelihoods are then assigned to the at least two possible interface control actions,

wherein the classifier is an ensemble classifier with several individual classifiers, where each individual classifier is built on a data chunk of the collected and stored data,

wherein each data chunk contains all data for a time interval specific for each data chunk and a time interval assigned to at least two data chunks overlap with each other, and

wherein the overlap and a data chunk width controls when new data chunks are created.

12. The in-vehicle user interface of claim 8, wherein the several individual classifiers include a primary classifier that generates a first estimate of likelihoods and a further classifier to supplement the first estimate of likelihoods.

13. The in-vehicle user interface of claim 12, wherein the primary classifier is a naï

ve Bayes classifier and the further classifier is a random forest classifier.

14. The in-vehicle user interface of claim 8, wherein the overlap is based on overlap percentage, which is a percentage of data samples existing in multiple data chunks, and stack depth, which is a maximum number of data chunks a single data sample can be present within.

11. A method for predicting an interface control action of a user with an in-vehicle user interface, the method comprising:

collecting and storing data, wherein the data includes at least

vehicle data about the vehicle and the vehicle'"'"'s environment collected from at least one sensor of the vehicle, and

user data about user interactions with the user interface and/or different applications inside the vehicle;

assigning likelihoods to at least two possible interface control actions by the user based on the collected and stored data;

determining at least one most likely interface control action from the likelihoods; and

providing the user with the at least one most likely interface control action in a manner so that the at least one most likely interface control action is selectable and performable with one single user interaction with the user interface,

wherein a classifier calculates likelihoods for the at least two possible interface control actions based on the collected and stored data and these calculated likelihoods are then assigned to the at least two possible interface control actions,

wherein the classifier is an ensemble classifier with several individual classifiers, where each individual classifier is built on a data chunk of the collected and stored data,

wherein each data chunk contains all data for a time interval specific for each data chunk and a time interval assigned to at least two data chunks overlap with each other,

wherein the overlap and a data chunk width controls when new data chunks are created, and

wherein the overlap is based on overlap percentage, which is a percentage of data samples existing in multiple data chunks, and stack depth, which is a maximum number of data chunks a single data sample can he present within.