Method for detecting safety of driving behavior, apparatus, device and storage medium

Patent metadata

Field	Value
Publication number	US-11084494-B2
Application number	US-201916508667-A
Country	US
Kind code	B2
Filing date	Jul 11, 2019
Priority date	Aug 3, 2018
Publication date	Aug 10, 2021
Grant date	Aug 10, 2021

Abstract

A method for detecting safety of a driving behavior, an apparatus, a device and a storage medium are provided. The method includes acquiring current driving data of a vehicle during a driving process of the vehicle; determining current driving behavior feature data of the vehicle according to the current driving data of the vehicle; inputting the current driving behavior feature data of the vehicle into a real-time safety detection model and calculating a security score corresponding to current driving behavior of the vehicle; and determining whether the current driving behavior of the vehicle is safe according to the security score corresponding to the current driving behavior of the vehicle. The method can assist an optimization of a vehicle driving system, reduce a safety risk of vehicle driving, and improve a riding experience of the user.

First claim

What is claimed is:

1. A method for detecting safety of a driving behavior, comprising:
acquiring current driving data of a vehicle during a driving process of the vehicle;
determining, according to the current driving data of the vehicle, current driving behavior feature data of the vehicle, wherein the driving behavior feature data includes a driving scene and driving data, and the driving data includes pose data, speed data and distance data of the vehicle;
inputting the current driving behavior feature data of the vehicle into a real-time safety detection model, and calculating a security score corresponding to current driving behavior of the vehicle, wherein the real-time safety detection model is obtained by training a neural network model via driving behavior feature data and a security marking score in a first training set; and
determining whether the current driving behavior of the vehicle is safe according to the security score corresponding to the current driving behavior of the vehicle.

2. The method of claim 1 , wherein the driving scene at least comprises: vehicle following, straight driving, turning, lane changing, making a U-turn, starting, parking and loop driving;
the speed data comprises a lateral driving speed, a longitudinal driving speed, and speeds of the vehicle relative to a road element, a traffic element and an obstacle element; and
the distance data comprises distances of the vehicle relative to the road element, the traffic element and the obstacle element.

3. The method of claim 1 , wherein the determining whether the current driving behavior of the vehicle is safe according to the security score corresponding to the current driving behavior of the vehicle, comprises:
determining whether the security score corresponding to the current driving behavior of the vehicle is lower than a preset early warning threshold;
determining the current driving behavior of the vehicle to be unsafe when the security score corresponding to the current driving behavior of the vehicle is lower than the preset early warning threshold; and
determining the current driving behavior of the vehicle to be safe when the security score corresponding to the current driving behavior of the vehicle is not lower than the preset early warning threshold.

4. The method of claim 3 , wherein after the determining the current driving behavior of the vehicle to be unsafe when the security score corresponding to the current driving behavior of the vehicle is lower than the preset early warning threshold, the method further comprises:
sending security low-score early warning information.

5. The method of claim 1 , further comprising:
acquiring a moment when an insecurity is generated during a historical driving process of the vehicle, and the security marking score and historical driving data corresponding to the moment when the insecurity is generated;
determining the driving behavior feature data corresponding to the moment when the insecurity is generated according to the historical driving data corresponding to the moment when the insecurity is generated; and
taking the driving behavior feature data and the security marking score corresponding to the moment when the insecurity is generated as a piece of training data, and adding it to the first training set.

6. The method of claim 5 , wherein the acquiring a moment when an insecurity is generated during a historical driving process of the vehicle, and the security marking score and historical driving data corresponding to the moment when the insecurity is generated, comprises:
acquiring overall driving process data and user insecurity description information corresponding to a plurality of historical driving processes of the vehicle in different driving environments, wherein the user insecurity description information is information recorded during the driving process of the vehicle to describe the insecurity of a user riding the vehicle;
determining, for any one historical driving process, a moment when preset description information appears in the user insecurity description information and the security marking score corresponding to the moment when the preset description information appears, according to the user insecurity description information corresponding to the historical driving process; and
determining the historical driving data corresponding to the moment when the preset description information appears according to the overall driving process data corresponding to the driving process;
wherein the moment when the preset description information appears in the user insecurity description information is a time point at which the insecurity is generated during the historical driving process.

7. The method of claim 1 , wherein after the determining whether the current driving behavior of the vehicle is safe according to the security score corresponding to the current driving behavior of the vehicle, the method further comprises:
determining a current moment as an unsafe moment during the driving process of the vehicle, when the current driving behavior of the vehicle is determined to be unsafe; and
determining an unsafe level corresponding to the unsafe moment according to the current driving behavior feature data of the vehicle and boundary information of a preset unsafe level.

8. The method of claim 7 , wherein after the determining whether the current driving behavior of the vehicle is safe according to the security score corresponding to the current driving behavior of the vehicle, the method further comprises:
determining the number of the unsafe moment corresponding to each driving scene during a current driving process after the current driving process is ended.

9. The method of claim 8 , wherein after the determining the number of the unsafe moment corresponding to each driving scene during a current driving process after the current driving process is ended, the method further comprises:
inputting the number of the unsafe moment corresponding to each driving scene during the current driving process and the unsafe level corresponding to the unsafe moment into a driving process security detection model, and calculating a security overall score corresponding to the current driving process;
wherein the driving process safety detection model is obtained by training the neural network model by the number of the unsafe moment corresponding to each driving scene during the historical driving process in a second training set, the unsafe level corresponding to the unsafe moment, and an overall marking score corresponding to the historical driving process.

10. An apparatus for detecting safety of a driving behavior, comprising: a memory, a processor, and a computer program stored on the memory and runnable on the processor, wherein the computer program, when executed by the processor, causes the processor to:
acquire current driving data of a vehicle during a driving process of the vehicle;
determine current driving behavior feature data of the vehicle according to the current driving data of the vehicle, wherein the driving behavior feature data includes a driving scene and driving data, and the driving data includes pose data, speed data and distance data of the vehicle;
input the current driving behavior feature data of the vehicle into a real-time safety detection model, and calculate a security score corresponding to current driving behavior of the vehicle, wherein the real-time safety detection model is obtained by training a neural network model via driving behavior feature data and a security marking score in a first training set; and
determine whether the current driving behavior of the vehicle is safe according to the security score c

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Hu Yunyan
• Wang Ya
• Pan Yifeng
• Hao Dayang
• Liu Rui
• Zhang Lizhi
• Luo Xiaojun
• Zhang Yaling
• Tao Ji
• Shen Ruixiang
• Huang Qihao

Classifications

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G05B13/027

  using neural networks only · CPC title

• B60W2050/0075

  Automatic parameter input, automatic initialising or calibrating means · CPC title

• G06Q10/0639 Primary

  Performance analysis of employees; Performance analysis of enterprise or organisation operations · CPC title