Method and apparatus for generating information

• US 10,809,723 B2
• Filed: 09/18/2018
• Issued: 10/20/2020
• Est. Priority Date: 12/11/2017
• Status: Active Grant

• Current Assignee: Apollo Intelligent Driving Technology Company Limited
• Original Assignee: Baidu Online Network Technology (Beijing) Co., Ltd (Baidu Incorporated)
• Inventors: Kong, Fanduo, Yan, Jing, Tao, Yang, Hu, Yunyan
• Primary Examiner(s): Butler, Rodney A
• Application Number:
  US16/134,850
• Publication Number:
  US 20190179312A1
• Time in Patent Office: 763 Days
• CPC Class Codes: B60W 40/02 related to ambient conditions
• CPC Class Codes: G05D 1/0088 characterized by the autonomous decision making process, e.g. artificial intelligence, predefined behaviours using knowledge based models G06N5/00
• CPC Class Codes: G05D 1/0221 involving a learning process
• CPC Class Codes: G05D 1/0246 using a video camera in combination with image processing means
• CPC Class Codes: G05D 1/0276 using signals provided by a source external to the vehicle involving a plurality of vehicles G05D1/0287; automatically controlling vehicle speed responsive to externally generated signals B60K31/0058
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06N 3/044 Recurrent networks, e.g. Hopfield networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G08G 1/0112 from the vehicle, e.g. floating car data [FCD]
• CPC Class Codes: G08G 1/0116 from roadside infrastructure, e.g. beacons
• CPC Class Codes: G08G 1/012 from other sources than vehicle or roadside beacons, e.g. mobile networks
• CPC Class Codes: G08G 1/0129 for creating historical data or processing based on historical data
• CPC Class Codes: G08G 1/0133 for classifying traffic situation
• CPC Class Codes: G08G 1/0145 for active traffic flow control
• CPC Class Codes: G08G 1/096811 where the route is computed offboard

Abstract

Embodiments of the present disclosure disclose a method and apparatus for generating information. A specific embodiment of the method comprises: acquiring associated information of a road to be evaluated for driving difficulty; generating traffic environment information of the road based on the associated information; and inputting the traffic environment information of the road into a driving difficulty evaluation model to obtain a driving difficulty level of the road for an autonomous driving vehicle. The driving difficulty of the road for the autonomous driving vehicle is evaluated, to obtain the driving difficulty level of the road for the autonomous driving vehicle. Thus, based on a driving difficulty level of each road for the autonomous driving vehicle, the autonomous driving vehicle may select a road having a low driving difficulty level to drive.

Claims

1. A method for generating information, comprising:

acquiring associated information of a road to be evaluated for driving difficulty, the associated information including road condition images and road attributes;

generating traffic environment information of the road based on the associated information; and

inputting the traffic environment information of the road into a driving difficulty evaluation model to obtain a driving difficulty level of the road for an autonomous driving vehicle, the driving difficulty evaluation model being used to represent a corresponding relationship between the traffic environment information and the driving difficulty level, each information item in the traffic environment information corresponding to a model parameter of the driving difficulty evaluation model, and a parameter value of each model parameter indicating an importance degree when an information item corresponding to the model parameter is used to define the driving difficulty level.

2. The method according to claim 1, wherein the traffic environment information includes at least one of:

an average speed of vehicles on the road, a saturation of the road indicating a congestion degree of the road, an average number of objects of a preset type at intersections on the road, an average number of conflict points of the intersections on the road, a number of obstacles on the road, or a number of road facilities on the road.

3. The method according to claim 2, wherein the driving difficulty evaluation model is obtained by:

creating a neural network for generating the driving difficulty evaluation model, wherein network parameters of the neural network include weight parameters, each weight parameter corresponds to an information item in the traffic environment information, and a parameter value of the each weight parameter indicates an importance degree when the information item corresponding to the weight parameter is used to define the driving difficulty level;

acquiring a plurality of training samples and annotation information of each training sample, wherein the training samples include the traffic environment information of the road, and the annotation information of the training sample includes an annotated driving difficulty level of the road;

training the neural network using the plurality of training samples and the annotation information of the each training sample in a supervised learning mode; and

defining the trained neural network as the driving difficulty evaluation model.

4. The method according to claim 3, wherein the training the neural network using the plurality of training samples and the annotation information of the each training sample in a supervised learning mode comprises:

performing, for the each training sample of the plurality of training samples, following training;

inputting the training sample into the neural network to obtain a predicted driving difficulty level;

determining a difference between the predicted driving difficulty level and the driving difficulty level in the annotation information of the training sample, and calculating gradient information based on the difference, the gradient information indicating a correlation between a parameter value of each current weight parameter of the neural network and the difference; and

adjusting the parameter value of the weight parameter of the neural network based on the gradient information.

5. The method according to claim 4, wherein the road attributes include a length of the road;

and

the generating traffic environment information of the road based on the associated information comprises;

finding, from the acquired road condition images, road condition images corresponding to a collection location being a starting point or an endpoint of the road, wherein a collection location corresponding to a road condition image is a location of a camera collecting the road condition image;

determining vehicles passing through the road based on the found road condition images;

calculating, for each determined vehicle passing through the road, a time difference between time collecting the road condition image in which a location of the vehicle is the starting point of the road and time collecting the road condition image in which a location of the vehicle is the end point of the road, and dividing the length of the road by the time difference to obtain a speed of the vehicle; and

calculating an average value of speeds of the determined vehicles passing through the road, and using the average value as the average speed of the vehicles on the road.

6. The method according to claim 4, wherein the road attributes include a type of the road, a number of lanes of the road, and widths of the lanes of the road;

and

the generating traffic environment information of the road based on the associated information comprises;

determining vehicles passing through the road in a preset time length based on road condition images corresponding to collection locations being a starting point and an end point of the road in the acquired road condition images, wherein a collection location corresponding to a road condition image is a location of a camera collecting the road condition image;

calculating a number of vehicles passing through the road in unit time based on a number of the vehicles passing through the road in the preset time length and the preset time length;

calculating a maximum traffic capacity of the road based on the road attributes, wherein the maximum traffic capacity of the road is a maximum number of the vehicles passing through the road in the unit time; and

dividing the number of the vehicles passing through the road in the unit time by the maximum traffic capacity to obtain the saturation of the road.

7. The method according to claim 4, wherein the generating traffic environment information of the road based on the associated information comprises:

finding, from the acquired road condition images, road condition images corresponding to collection locations being the intersections of the road, wherein a collection location corresponding to a road condition image is a location of a camera collecting the road condition image;

determining, for each intersection of the road, a plurality of road condition images corresponding to a collection location being the intersection in the found road condition images, wherein a time difference between time collecting adjacent road condition images in the plurality of road condition images is a preset time length;

performing image recognition on the plurality of road condition images;

calculating an average value of numbers of the recognized objects of the preset type; and

using the average value as a number of the objects of the preset type at the intersection; and

calculating an average value of numbers of the objects of the preset type at the intersections of the road; and

using the average value of the numbers of the objects of the preset type at the intersections of the road as the average number of the objects of the preset type at the intersections of the road.

8. The method according to claim 4, wherein the generating traffic environment information of the road based on the associated information comprises:

finding, from the acquired road condition images, road condition images corresponding to collection locations being the intersections of the road, wherein a collection location corresponding to a road condition image is a location of a camera collecting the road condition image;

determining, for each intersection of the road, a plurality of road condition images corresponding to a collection location being the intersection in the found road condition images;

determining changing situations of driving directions of vehicles at the intersection and changing situations of states of traffic signal lights based on the plurality of road condition images; and

determining conflict points of the intersection based on the changing situations of the driving directions of the vehicles at the intersection and the changing situations of the states of the traffic signal lights; and

calculating an average value of numbers of the conflict points of the intersections on the road, and using the average value as the average number of the conflict points of the intersections on the road.

9. An apparatus for generating information, comprising:

at least one processor; and

a memory storing instructions, the instructions when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising;

acquiring associated information of a road to be evaluated for driving difficulty, the associated information including road condition images and road attributes;

generating traffic environment information of the road based on the associated information; and

inputting the traffic environment information of the road into a driving difficulty evaluation model to obtain a driving difficulty level of the road for an autonomous driving vehicle, the driving difficulty evaluation model being used to represent a corresponding relationship between the traffic environment information and the driving difficulty level, each information item in the traffic environment information corresponding to a model parameter of the driving difficulty evaluation model, and a parameter value of each model parameter indicating an importance degree when an information item corresponding to the model parameter is used to define the driving difficulty level.

10. The apparatus according to claim 9, wherein the traffic environment information includes at least one of:

an average speed of vehicles on the road, a saturation of the road indicating a congestion degree of the road, an average number of objects of a preset type at intersections on the road, an average number of conflict points of the intersections on the road, a number of obstacles on the road, or a number of road facilities on the road.

11. The apparatus according to claim 10, the driving difficulty evaluation model is obtained by:

creating a neural network for generating the driving difficulty evaluation model, wherein network parameters of the neural network include weight parameters, each weight parameter corresponds to an information item in the traffic environment information, and a parameter value of the each weight parameter indicates an importance degree when the information item corresponding to the weight parameter is used to define the driving difficulty level;

acquiring a plurality of training samples and annotation information of each training sample, wherein the training samples include the traffic environment information of the road, and the annotation information of the training sample includes an annotated driving difficulty level of the road;

training the neural network using the plurality of training samples and the annotation information of the each training sample in a supervised learning mode; and

defining the trained neural network as the driving difficulty evaluation model.

12. The apparatus according to claim 11, wherein the training the neural network using the plurality of training samples and the annotation information of the each training sample in a supervised learning mode comprises:

performing, for the each training sample of the plurality of training samples, following training;

inputting the training sample into the neural network to obtain a predicted driving difficulty level;

determining a difference between the predicted driving difficulty level and the driving difficulty level in the annotation information of the training sample, and calculating gradient information based on the difference, the gradient information indicating a correlation between a parameter value of each current weight parameter of the neural network and the difference; and

adjusting the parameter value of the weight parameter of the neural network based on the gradient information.

13. The apparatus according to claim 12, wherein the road attributes include a length of the road;

and

the generating traffic environment information of the road based on the associated information comprises;

finding, from the acquired road condition images, road condition images corresponding to a collection location being a starting point or an endpoint of the road, wherein a collection location corresponding to a road condition image is a location of a camera collecting the one road condition image;

determining vehicles passing through the road based on the found road condition images;

calculating, for each determined vehicle passing through the road, a time difference between time collecting the road condition image in which a location of the vehicle is the starting point of the road and time collecting the road condition image in which a location of the vehicle is the end point of the road, and dividing the length of the road by the time difference to obtain a speed of the vehicle; and

calculating an average value of speeds of the determined vehicles passing through the road and using the average value as the average speed of the vehicles on the road.

14. The apparatus according to claim 12, wherein the road attributes include a type of the road, a number of lanes of the road, and widths of the lanes of the road, and

the generating traffic environment information of the road based on the associated information comprises:

determining vehicles passing through the road in a preset time length based on road condition images corresponding to collection locations being a starting point and an end point of the road in the acquired road condition images, wherein a collection location corresponding to a road condition image is a location of a camera collecting the road condition image;

calculating a number of vehicles passing through the road in unit time based on a number of the vehicles passing through the road in the preset time length and the preset time length;

calculating a maximum traffic capacity of the road based on the road attributes, wherein the maximum traffic capacity of the road is a maximum number of the vehicles passing through the road in the unit time; and

dividing the number of the vehicles passing through the road in the unit time by the maximum traffic capacity to obtain the saturation of the road.

15. The apparatus according to claim 12, wherein the generating traffic environment information of the road based on the associated information comprises:

finding, from the acquired road condition images, road condition images corresponding to collection locations being the intersections of the road, wherein a collection location corresponding to a road condition image is a location of a camera collecting the road condition image;

determining, for each intersection of the road, a plurality of road condition images corresponding to a collection location being the intersection in the found road condition images, wherein a time difference between time collecting adjacent road condition images in the plurality of road condition images is a preset time length;

performing image recognition on the plurality of road condition images;

calculate an average value of numbers of the recognized objects of the preset type;

using the average value as a number of the objects of the preset type at the intersection;

calculating an average value of numbers of the objects of the preset type at the intersections of the road; and

using the average value of the numbers of the objects of the preset type at the intersections of the road as the average number of the objects of the preset type at the intersections of the road.

16. The apparatus according to claim 12, wherein the generating traffic environment information of the road based on the associated information comprises:

finding, from the acquired road condition images, road condition images corresponding to collection locations being the intersections of the road, wherein a collection location corresponding to a road condition image is a location of a camera collecting the road condition image;

determining, for each intersection of the road, a plurality of road condition images corresponding to a collection location being the intersection in the found road condition images;

determining changing situations of driving directions of vehicles at the intersection and changing situations of states of traffic signal lights based on the plurality of road condition images;

determining conflict points of the intersection based on the changing situations of the driving directions of the vehicles at the intersection and the changing situations of the states of the traffic signal lights; and

calculating an average value of numbers of the conflict points of the intersections on the road, and using the average value as the average number of the conflict points of the intersections on the road.

17. A non-transitory computer storage medium storing a computer program, the computer program when executed by one or more processors, causes the one or more processors to perform operations, the operations comprising:

acquiring associated information of a road to be evaluated for driving difficulty, the associated information including road condition images and road attributes;

generating traffic environment information of the road based on the associated information; and

inputting the traffic environment information of the road into a driving difficulty evaluation model to obtain a driving difficulty level of the road for an autonomous driving vehicle, the driving difficulty evaluation model being used to represent a corresponding relationship between the traffic environment information and the driving difficulty level, each information item in the traffic environment information corresponding to a model parameter of the driving difficulty evaluation model, and a parameter value of each model parameter indicating an importance degree when an information item corresponding to the model parameter is used to define the driving difficulty level.