CN107358678B - Driving event detection and identification method - Google Patents
Driving event detection and identification method Download PDFInfo
- Publication number
- CN107358678B CN107358678B CN201710425167.6A CN201710425167A CN107358678B CN 107358678 B CN107358678 B CN 107358678B CN 201710425167 A CN201710425167 A CN 201710425167A CN 107358678 B CN107358678 B CN 107358678B
- Authority
- CN
- China
- Prior art keywords
- acceleration
- event
- suspected
- distance
- automobile
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C5/00—Registering or indicating the working of vehicles
- G07C5/08—Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
- G07C5/0841—Registering performance data
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C5/00—Registering or indicating the working of vehicles
- G07C5/08—Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
- G07C5/0808—Diagnosing performance data
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides a driving event detection and identification method, which comprises the following steps: a suspected event judgment stage: continuously acquiring an acceleration value of the advancing direction of the automobile, comparing the acceleration value with a threshold value of a suspected event, and judging the severity level of the suspected sudden acceleration/deceleration event; template distance matching stage: and respectively matching the determined suspected rapid acceleration/deceleration event with a corresponding event template with a severity grade, comparing the distance between the determined suspected rapid acceleration/deceleration event and the template, and when the distance is smaller than a set template distance threshold value, determining the severity grade of the event template as the real dangerous driving grade of the vehicle. The driving event detection and identification method provided by the invention can effectively identify the dangerous driving event of sudden acceleration/deceleration, thereby reminding a driver to correct the driving habit of the driver and ensuring the driving safety.
Description
Technical Field
The invention belongs to the field of computer identification, and particularly relates to a driving event detection and identification method.
Background
Dangerous driving events are often generated by drivers due to road conditions or individual driving habits in the driving process, and the dangerous driving events not only threaten the life and property safety of the drivers, but also influence the traffic. Timely detection and identification of dangerous driving events is essential to safe driving by the driver.
The dangerous driving events are rapid acceleration/deceleration events, wherein the rapid acceleration events are rapid speed increase of the automobile in a short time due to the fact that a driver steps on the accelerator too hard in a short time during driving, and acceleration is increased to a very high value. The sudden deceleration event is caused by that the driver steps on the brake plate too violently in a very short time to cause the speed of the automobile to drop rapidly in a short time, and the acceleration reaches a very high value (negative direction). These two types of dangerous driving events are related to the acceleration of the vehicle in the forward direction, which can be obtained by some sensors.
The current methods for detecting and identifying events are mainly divided into three categories:
1. threshold-based decision methods are used because sharp acceleration/deceleration events are often accompanied by very high acceleration values, as identified using threshold methods. When acceleration exceeding a certain threshold (positive and negative for emergency acceleration and rapid deceleration, respectively) is detected, a rapid acceleration/deceleration event is determined to have occurred. Such recognition algorithms are efficient and easy to implement, but noise is often accompanied in raw data, and a large number of false judgments are generated when an event is recognized based on a simple threshold value.
2. And (3) identifying by using a classifier, extracting the characteristics of the average value, the extreme value and the like of the acceleration values in the sudden acceleration/deceleration event, putting the characteristics into an SVM classifier for training, and obtaining the classifier capable of identifying the sudden acceleration/deceleration event. The algorithm has high recognition accuracy, but a large amount of training sets are required to be used for training in order to ensure the accuracy of the classifier, so that the time cost and the energy consumption are high.
3. DTW is used for recognition, and a Dynamic Time Warping (DTW) algorithm is a template matching algorithm based on a Dynamic programming idea and is applied to the field of voice recognition at the earliest Time. Due to different individual driving habits, driving environments, road conditions and the like, the duration time and the force of different behaviors are different. Therefore, the event recognition algorithm based on DTW can recognize events with different data lengths by matching the data segment to be recognized with the template. DTW is used for event identification. The accuracy of identification by using a DTW algorithm is directly influenced by the selection of the DTW template. In practical situations, the influence degrees of the rapid acceleration and deceleration events are different due to different acceleration extreme values, so that it is unreasonable to generally identify the events under various complex conditions into a type of rapid acceleration/deceleration event, and a single template is easy to cause that real driving events slightly different from the template cannot be correctly identified.
Disclosure of Invention
The driving event detection and identification method disclosed by the invention can be used for carrying out intelligent analysis on the acceleration data through the detection system and effectively identifying the dangerous driving event of sudden acceleration/deceleration, so that a driver is reminded to assist in correcting the driving habit of the driver and the driving safety is ensured.
Drawings
FIG. 1 is a general flow diagram of a driving event monitoring and identification method of the present invention;
FIG. 2 is a step of mapping acceleration data collected in the present invention to a vehicle heading direction;
FIG. 3 is a diagram of a rapid acceleration event versus threshold in accordance with the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The method disclosed by the invention relates to technologies such as intelligent identification and analysis, and can be directly used for obtaining the forward acceleration value of the automobile according to the acceleration value obtained by the mobile equipment, and further used for identifying the running state of the automobile and dividing the danger level.
As shown in fig. 1, the method comprises the steps of:
and S1, continuously acquiring the acceleration value of the advancing direction of the automobile.
And S2, detecting the start point and the end point of the rapid acceleration event and recording the value of the period.
And S3, matching with the corresponding event template to obtain an event identification result.
Further, in the step S1, the method includes:
and S11, collecting acceleration data through a sensor built in the vehicle-mounted equipment or the mobile phone.
And S12, mapping the acceleration data collected from the equipment to the acceleration vector of the advancing direction of the automobile.
Further, mapping the acquired acceleration data to the vehicle heading in step 2 is performed by,
and S121, extracting the average value of the three-axis acceleration of the acceleration sensor of the first 1S as a mobile terminal coordinate system vector (x0, y0, z 0).
S122, judging the standard postures closest to the current mobile terminal position according to the absolute values of the three axes (x0, y0, z0), wherein the standard postures are divided into 6 types: horizontal up, horizontal down, vertical long side up, vertical long side down, vertical short side up, vertical short side down.
And S123, setting a basic vector (x1, y1, z1) of the automobile coordinate system, wherein the horizontal two axial accelerations are 0, and the third axial acceleration is the gravity acceleration.
And S124, determining a transformation matrix R from the mobile terminal coordinate system to the automobile coordinate system according to the mobile terminal coordinate system vector (x0, y0, z0) and the automobile coordinate system vector (x1, y1, z 1).
The transformation matrix R is derived from the vectors (x0, y0, z0) and (x1, y1, z1) by equations (1-1) and (1-2), where K and V correspond to (x0, y0, z0) and (x1, y1, z 1).
cosθ＝K·V/|K||V| (1-1)
And S125, starting the automobile, and judging that the current driving behavior of the automobile is forward.
S126, extracting an acceleration value of 0 by using the current acceleration sensor and right-multiplying the transformation matrix A0 to obtain an acceleration vector (x, y, z) of the automobile coordinate system.
And S127, determining the deflection angle between the acceleration of the remaining two shafts and the advancing direction of the automobile after the gravity acceleration of the vector is removed.
And S128, obtaining a forward acceleration acc _ brake and a turning acceleration right _ left in the final automobile running coordinate system according to the deflection angle and the transformation matrix.
And S13, removing the influence of gravity acceleration from the acceleration vector of the automobile forward direction, and finishing filtering and noise reduction.
As shown in fig. 2, the step S2 (for example, the rapid acceleration and the rapid deceleration are the same) includes:
s21, when the acceleration value exceeds a certain threshold (0.2g) and continues to exceed 1S, it means that from this 1S, the driver is performing a continuous acceleration action, i.e. determining a suspected rapid acceleration event, and recording the acceleration value from the first point exceeding the threshold.
S22, when the acceleration value is lower than a certain threshold (0.2g) and continuously exceeds 1S, it means that from the 1S, the driver is not continuously accelerating with high threshold, i.e. it is determined that a suspected rapid acceleration event is over, and the recording of the acceleration value is stopped.
And S23, storing the acceleration data recorded between the start point and the stop point.
And S24, comparing the acceleration data with the threshold values according to three preset threshold values, such as 0.2g, 0.3g and 0.4g, and dividing dangerous driving events exceeding the corresponding threshold values into suspected slight rapid acceleration events, suspected medium rapid acceleration events and suspected serious rapid acceleration events.
Further, the step S3 includes:
and S31, respectively matching the three suspected rapid acceleration/deceleration events extracted in the step S2 with corresponding event templates, wherein when the distance between the three suspected rapid acceleration/deceleration events and the template is less than a certain threshold (1.5), the similarity between the data segment and the template data segment is high, and the current suspected rapid acceleration/deceleration event is a real rapid acceleration/deceleration event.
The event template is sensor data corresponding to a pre-collected and marked real driving event, and the mapping from the original data to a coordinate system of the advancing direction of the automobile is completed.
As shown in fig. 3, after a suspected rapid acceleration event is intercepted by using a threshold of 0.2g, 1 suspected mild rapid acceleration event, 2 suspected moderate rapid acceleration events, and 1 suspected severe rapid acceleration event are separated from the whole rapid acceleration event according to the severity (0.3g, 0.4 g). When the distance comparison is performed with the corresponding event template data, a threshold value of the distance, for example, 1.5, is preset.
For two time sequence data segments, a data segment E of a suspected event and a data segment M corresponding to an event template, calculating the distance between the two data segments:
1. two sequences are defined, two points at random Ei，MjIs the difference of the real data values, i 1,2,3.. n, j 1,2,3.. m, n, m are positive integers.
2. Defining two time sequences E, wherein the distance between M is as follows: first data point from both (E)1，M1) Beginning, last data point (E)n，Mm) Ending, calculating the distance of the corresponding data points, and adding the distances between the points to obtain the E and M distances.
When the distance between two points is calculated, the rule corresponding to the data point is as follows:
a) boundary: the head and the tail of the two pieces of data are the first one corresponding to the first one, and the last one corresponding to the last one.
b) Continuity: for a certain point in E, it can only correspond to the first unmatched data point or the just matched data point in the M sequence (the same applies to the points in M), that is, it cannot cross a certain point to make a correspondence or match, thus ensuring that all points in the two data segments have corresponding points.
c) Monotonicity: the points in the two sequences are matched in time sequence and cannot cross one point.
The data segments E and M are not always the same length (n is not necessarily the same as M). At this point, distances may be calculated from a plurality of points in M for a point in E (it may also happen that a plurality of points in E and a point in M are calculated distances). The distance between E and M is the shortest distance among the many possible matching results
The method for calculating the shortest distance comprises the following steps:
a) the initialization system sets the initial distance of E and M to be D-0.
b) The detection system detects from the first point pair (E)1，M1) Initially, for each point pair (E)i，Mj) Calculating the distance D of the corresponding point according to the formula (1-3),
D＝min{D(Ei-1，Mj-1)+d(i，j),D(Ei-1，Mj)+d(i，j),D(Ei，Mj-1)+d(i，j)} (1-3)，
wherein d (i, j) is E in the templatei，MiThe euclidean distance of (D) is calculated as the sum of the piecewise path distances D as i, j increases.
Can cause the current to be matchedThe matched point pair is (E)i，Mj) There are three cases of (1):
the last matching point pair is (E)i-1，Mj) Then, the event section E moves forward by one point, and M is unchanged;
the last matching point pair is (E)i，Mj-1) Then, the template section M moves forward by one point, and E is unchanged;
the last matching point pair is (E)i-1，Mj-1) Then, the event segment E and the template segment M both move forward by one point;
the cumulative distance D to be finally calculated should be equal to the cumulative distance of the previous matching point plus the current matching point (E)i，Mj) Since the final goal is to obtain the minimum cumulative distance, D in each of the above three cases may be calculated and the minimum distance may be obtained.
c) Plus the last point pair (E)n，Mm) In the distance obtained after the distance, the path with the minimum value is the matching result of each point of the two sequences, and the value is the distance between E and M.
And S32, when two or three types of rapid acceleration/deceleration events are identified, judging the types of the whole acceleration records according to the sequence of the rapid acceleration/deceleration events with the severity degree from large to small, namely the severe rapid acceleration/deceleration event > medium rapid acceleration/deceleration event > slight rapid acceleration/deceleration event. For example, if a severe rapid acceleration event and a medium rapid acceleration event are identified in a certain data segment, the suspected rapid acceleration event corresponding to the data segment is marked as a real severe rapid acceleration event. Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (1)
1. A driving event detection and identification method, comprising:
(1) a suspected event judgment stage: continuously acquiring an acceleration value of the advancing direction of the automobile, comparing the acceleration value with a threshold value of a suspected event, and judging the severity level of the suspected sudden acceleration/deceleration event of the automobile;
(2) template distance matching stage: respectively matching the determined suspected rapid acceleration/deceleration event with an event template with a corresponding severity level, comparing the distance between the determined suspected rapid acceleration/deceleration event and the event template, and when the distance is smaller than a threshold value of the set event template distance, the severity level of the event template is the real dangerous driving level of the vehicle, and calculating the shortest distance between a data section E of the suspected event and a data section M corresponding to the event template to calculate the shortest distance between a plurality of possible matching results of the E and the M, wherein the specific steps are that a) an initialization system sets the initial distance between the E and the M to be D-0; b) the detection system detects from the first point pair (E)1，M1) Initially, for each point pair (E)i，Mj) Min { D (E) } according to the formula Di-1，Mj-1)+d(i，j),D(Ei-1，Mj)+d(i，j),D(Ei，Mj-1) + D (i, j) } calculating the distance D of the corresponding point, wherein D (i, j) is E in the templatei，MiThe Euclidean distance of (D), which calculates the sum of the piecewise path distances D as i, j increases; c) adding the sum of the piecewise-path distances D calculated in step b) to the last point pair (E)n，Mm) In the distances obtained after the distance, the path with the smallest value is the result of matching of each point of the two sequences, namely the distance between E and M, wherein i is 1,2,3. When E and M calculate the distance, the rule corresponding to the data point is as follows: the head and the tail of the two sections of data are that the first strip corresponds to the first strip and the last strip corresponds to the last strip; the points in the two sequences need to be correspondingly matched according to the time sequence, and can not be correspondingly or matched across a certain point, the certain point in the E can only be matched with the first unmatched data point or the data point which is just matched in the M sequence, and the point in the M can only be matched with the first unmatched data point or the data point which is just matched in the E sequenceData points, which ensure that all points in the two data segments have corresponding points and cannot cross one point; in a suspected event judgment stage, acquiring acceleration data through a built-in sensor of vehicle-mounted equipment or a mobile phone, mapping the acquired acceleration data to an acceleration vector of an automobile advancing direction, removing the influence of gravity acceleration in the acceleration vector of the automobile advancing direction, and then obtaining an acceleration value of the automobile advancing direction through filtering and noise reduction; comparing the acceleration value of the automobile in the advancing direction with three preset threshold values, and setting the suspected rapid acceleration/deceleration event of the automobile exceeding the three threshold values as a suspected slight rapid acceleration/deceleration event, a suspected medium rapid acceleration/deceleration event and a suspected serious rapid acceleration/deceleration event; by severity varying over a period of time, i.e. severe jerk/deceleration events>Moderate jerk acceleration/deceleration events>The sequence of slight jerk/deceleration events defines a category of severity of the overall acceleration record; the step of mapping the acquired acceleration data to the advancing direction of the automobile comprises the following steps of extracting an average value of three-axis accelerations of the acceleration sensor as a mobile terminal coordinate system vector, setting an automobile coordinate system base system vector, determining a transformation matrix from a mobile terminal coordinate system to an automobile coordinate system according to the mobile terminal coordinate system vector and the automobile coordinate system vector, and removing the influence of the gravity acceleration according to the current acceleration value and the transformation matrix relation of the automobile to obtain the advancing acceleration and the turning acceleration of the automobile.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710425167.6A CN107358678B (en) | 2017-06-07 | 2017-06-07 | Driving event detection and identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710425167.6A CN107358678B (en) | 2017-06-07 | 2017-06-07 | Driving event detection and identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107358678A CN107358678A (en) | 2017-11-17 |
| CN107358678B true CN107358678B (en) | 2020-11-27 |
Family
ID=60273537
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710425167.6A Active CN107358678B (en) | 2017-06-07 | 2017-06-07 | Driving event detection and identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107358678B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AT521724A1 (en) * | 2018-09-24 | 2020-04-15 | Avl List Gmbh | Method and device for analyzing a sensor data stream and method for driving a vehicle |
| CN111951550B (en) * | 2020-08-06 | 2021-10-29 | 华南理工大学 | Traffic safety risk monitoring method and device, storage medium and computer equipment |
| CN112950900A (en) * | 2021-01-27 | 2021-06-11 | 吉林云帆智能工程有限公司 | Driver behavior monitoring and detecting method |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9846174B2 (en) * | 2013-07-12 | 2017-12-19 | Roy Schwartz | Computer-implemented methods and computer systems/machines for identifying dependent and vehicle independent states |
| CN105718710B (en) * | 2014-12-02 | 2019-01-11 | 高德软件有限公司 | A kind of driving behavior analysis method and apparatus |
| CN104786967A (en) * | 2015-04-22 | 2015-07-22 | 北京九五智驾信息技术股份有限公司 | Driving behavior analysis system |
| CN105096543B (en) * | 2015-08-24 | 2017-10-03 | 中国电子科技集团公司第二十九研究所 | A kind of vehicular rescue method and system based on drive recorder |
| CN105389984A (en) * | 2015-11-16 | 2016-03-09 | 北京智视信息科技有限公司 | Driving behavior identification method based on mobile terminal sensing information fusion |
| CN106218641B (en) * | 2016-08-04 | 2018-10-09 | 武汉理工大学 | A kind of vehicle new hand driver automatic identification and safe early warning method |
-
2017
- 2017-06-07 CN CN201710425167.6A patent/CN107358678B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN107358678A (en) | 2017-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112389448B (en) | Abnormal driving behavior identification method based on vehicle state and driver state | |
| CN107848478B (en) | Automatic identification and evaluation of low speed collisions | |
| CN108469806B (en) | Driving right transfer method in alternating type man-machine common driving | |
| CN107358678B (en) | Driving event detection and identification method | |
| CN112053589B (en) | Target vehicle lane changing behavior adaptive identification model construction method | |
| JP2016146162A (en) | Driving determination device, driving determination program, calculation system, detection device, detection system, and detection method and program | |
| CN108682119B (en) | Driver fatigue state detection method based on smart phone and smart watch | |
| CN108229304B (en) | Systematic driving behavior identification method based on clustering idea | |
| CN106956680B (en) | Electric automobile driving behavior recognition analysis method | |
| JP2009157606A (en) | Driver status estimation device and program | |
| US11417108B2 (en) | Two-wheel vehicle riding person number determination method, two-wheel vehicle riding person number determination system, two-wheel vehicle riding person number determination apparatus, and program | |
| JP2014089686A (en) | Vehicle speed estimation apparatus and method | |
| JP2017073021A (en) | Driving support device | |
| JP2009175929A (en) | Driver condition estimating device and program | |
| CN114037015A (en) | Driving behavior safety analysis method based on driving mode primitive | |
| CN107644475B (en) | Vehicle collision detection method and device and OBD box | |
| CN117235656B (en) | Urban traffic management system and method based on big data and cloud computing | |
| JP2012128739A (en) | Collision risk determination device and program | |
| CN113657265A (en) | Vehicle distance detection method, system, device and medium | |
| CN113838277A (en) | Method, device and equipment for determining abnormal occurrence time point of vehicle and storage medium | |
| CN112132861A (en) | Vehicle accident monitoring method and system and computer equipment | |
| JP2009154675A (en) | Driver state estimation device and program | |
| CN107622667B (en) | Method and system for detecting altered license plate number | |
| CN113291311B (en) | Method for detecting abnormal behavior of driver under emergency collision avoidance working condition and storage medium | |
| Gao et al. | Driver identification based on stop-and-go events using naturalistic driving data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |