CN105574537B - Dangerous driving behavior testing and evaluation method based on multisensor - Google Patents
Dangerous driving behavior testing and evaluation method based on multisensor Download PDFInfo
- Publication number
- CN105574537B CN105574537B CN201510813444.1A CN201510813444A CN105574537B CN 105574537 B CN105574537 B CN 105574537B CN 201510813444 A CN201510813444 A CN 201510813444A CN 105574537 B CN105574537 B CN 105574537B
- Authority
- CN
- China
- Prior art keywords
- driving behavior
- wave
- vehicle
- waveform
- sensor
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
Abstract
The dangerous driving behavior testing and evaluation method based on multisensor that the invention proposes a kind of, comprising: using multiple sensors on vehicle, acquire multiple sensor feedback data in the vehicle operation；The multiple sensor feedback data are analyzed with default driving behavior distinguished number, the driving mode of the vehicle is identified based on the analysis results, generates driving behavior analysis result；The driving behavior analysis result is compared with default normal driving model, judges that the driving behavior whether there is danger, and the driving behavior assessment report for each driving behavior and the whole generation vehicle of driving.The present invention identifies and judges the driving mode that steering, lane change and the vehicle to turn around just occur under steam, and using differentiation as a result, in conjunction with real time running speed threshold value corresponding with safety traffic model into comparison, to judge the safety of current driving behavior.
Description
Technical field
The present invention relates to multisensor Data Fusion technology field, in particular to a kind of dangerous driving based on multisensor
Behavioral value and appraisal procedure.
Background technique
With the fast lifting of living standard, self-driving automobile has become the main traffic mode of people's trip.It is same with this
When, also along with the appearance of dangerous driving behavior.
From angle is driven, the threat for constituting people's trip includes two aspects: can all emerge in large numbers large quantities of driving new hands daily；
It is consummate but to the unfamiliar personnel of current friendship rule that there is also many driving simultaneously.Therefore driving user is required to drive itself
Technology carries out the standard and guidance of quantization judge.In addition, for car insurance industry, it can according to user behavior adjustment business demand
Targetedly release more reasonable insurance business.Quantization modulation is carried out by the driving procedure to user, on the one hand can be closed
The sales mode of physical and chemical vehicle insurance increases the profit of vehicle insurance business；On the other hand the raising that can also promote user's driving ability, is formed
Good driving attainment.
Further investigation reveals that in addition to drunk driving and force majeure factor, the cause of most of traffic accident be driving user not
Good driving habit.Undesirable driving habit mainly includes improperly steering, lane change and turns around.The driving status that auxiliary drives at present
Detection method mainly has three classes:
(1) it is based on the detection method of OBD (On-Board Diagnostic): there is highest Stability and veracity, but
The OBD interface and coding of different automobile production companies are different, and interface reads cost of implementation height, while may cause user's driving
Security control, this makes the technology based on OBD be difficult to popularize；
(2) based on the detection method of video monitoring: the Technical comparing is accurate, but is constrained to ambient light and weather shadow
It rings, while the discriminance analysis of great amount of images calculates, calculates cost and carrying cost is very high；
(3) detection method based on vehicle-mounted built-in sensors: in some high-end vehicles, it is loaded with expensive sensing
Device, electronic radar and monitoring device.These equipment can provide convenience to a certain extent to drive, but not have to driving generally
The ability of behavioural analysis.
In addition, to reduce cost and convenient for universal, prestige exists in the prior art by smart phone, tablet computer, vehicle-mounted
The smart machines such as vehicle-mounted auxiliary driving are navigated and customized, carry out the mode of vehicle detection, but be respectively present different degrees of defect.
(1) detection using the rear camera of smart phone according to the information progress driving condition of acquisition road, but by
It is limited in hardware environment, real-time and precision are difficult to be satisfied.
(2) driving behavior is detected using acceleration transducer, gyroscope and the electronic compass in smart phone, energy
The Space-time Complexity of calculating is enough effectively reduced, and is not readily susceptible to the influence of ambient lighting and weather.But electronic compass holds
Vulnerable to the magnetic interference in environment, judging result is influenced.In addition, the process that acceleration transducer and gyroscope are acquired in data
In be easy noise jamming by sensor, it is difficult to the driving mode that identification wherein captures.
The prior art proposes a kind of detection pattern for being based on Inertial Measurement Units (IMUs), Neng Gou
Sensor noise is separated to a certain extent, obtains sensor of interest numerical parameter.But this method divides driving behavior
The logic of class is fairly simple, can not analyze relative complex driving behavior.
As shown in Figure 1, the prior art also proposes a kind of utilization Hall magnetic effect sensor, that is, electronic compass is to vehicle
Steering judged, come measure drive in steering behavior.But it when nearby having high-intensity magnetic field or electromagnetic interference, will lead to
Electronic compass disturbance is even failed, therefore detection method stability is weaker.
Summary of the invention
The purpose of the present invention aims to solve at least one of described technological deficiency.
For this purpose, it is an object of the invention to propose a kind of dangerous driving behavior testing and evaluation side based on multisensor
Method identifies and judges the driving mode that steering, lane change and the vehicle to turn around just occur under steam, then by differentiating knot
Fruit judges the safety of current driving behavior in conjunction with real time running speed threshold value corresponding with safety traffic model into comparison.
To achieve the goals above, implementation of the invention is quasi- provides a kind of dangerous driving behavior detection based on multisensor
And appraisal procedure, include the following steps:
Step S1 acquires the multiple sensors feedback in the vehicle operation using multiple sensors on vehicle
Data；
Step S2 analyzes the multiple sensors feedback data with default driving behavior distinguished number, according to point
Analysis result identifies the driving mode of the vehicle, generates driving behavior analysis result.Wherein, the driving behavior analysis result
Including each driving behavior and corresponding sensor feedback data；
The driving behavior analysis result is compared with default normal driving model, judges the driving by step S3
Behavior is assessed with the presence or absence of danger, and for each driving behavior and the whole driving behavior for generating the vehicle of driving
Report.
Further, the multiple sensor includes: angular-rate sensor and GPS velocity sensor.
Further, the multiple sensor is using one of following two form:
(1) angular-rate sensor and GPS velocity sensor being fixed in the vehicle；
(2) angular-rate sensor and GPS velocity sensor of the gyroscope in smart machine are integrated in, wherein the intelligence
Energy equipment is fixed on the front window of the vehicle.
Further, multiple sensor feedback data include vehicle speed data and the institute of the GPS velocity sensor acquisition
The vehicle of angular-rate sensor acquisition is stated in steering, lane change and waveform sensor data when turning around.
Further, in the step S2, the default driving behavior distinguished number uses MultiWave algorithm.
Further, in the step S2,
Firstly, carrying out waved filter to the waveform sensor data, Wave data after filtering is stored in MultiWave chain
Table；
Then, the Wave data in the MultiWave chained list is analyzed, calculate wave quantity and corresponding spy
Value indicative is identified according to the quantity of the wave and characteristic value, is identified to the driving mode of the vehicle, wherein the feature
Value include: waveform the positive negativity of PN and maximum angular rate value, the angular speed mean value of single wave and single wave duration and turn
To the absolute value of angle.
Further, the waveform sensor data for meeting the following conditions are filtered:
(1) lower than the Wave data of default angular-rate sensor lower threshold；
(2) maximum value of single wave sensor numerical value is lower than the Wave data for presetting effective waveform peak threshold value；
(3) waveform duration is lower than the Wave data of default basic waveform duration；
When being more than the default detection waiting time new Wave data to be still not detected, this detection will be terminated, start to identify
Driving behavior.
Further, it is identified in the quantity according to wave and characteristic value, the driving mode of the vehicle is identified, wrapped
It includes::
(1) single wave: waveform is forward wave, then the vehicle is turned left out；Waveform is negative wave, then out the vehicle to
It turns right；The absolute value of steering angle is mild lane change when being located at section [0,20]；The absolute value of steering angle be located at section [65,
It 115] is steering；It is to turn around that the absolute value of steering angle, which is located at section [155,205],；
(2) two wave situations:
When two waveforms are forward wave, continue to judge steering angle, if the absolute value of steering angle is located at section
It [65,115] is left steering；It is to turn around to the left that the absolute value of steering angle, which is located at section [155,205],；
When two waveforms are negative wave, continue to judge steering angle, if the absolute value of steering angle is located at section
It [65,115] is right turn；It is to turn around to the right that the absolute value of steering angle, which is located at section [155,205],；
First waveform is forward wave, and when second waveform is negative wave, the vehicle is lane change to the left；
First waveform is negative wave, and when second waveform is forward wave, the vehicle is lane change to the right；
(3) three waves: if the absolute value of steering angle is mild lane change when being located at section [0,20]；Steering angle it is exhausted
Being located at section [65,115] to value is to turn to；It is nearby to turn around that the absolute value of steering angle, which is located at section [155,205],；
(4) more than three waves: the vehicle short time occurs acutely to swing.
Further, in the step S3, normal driving model is preset by the normal vehicle operation under different speed per hours
Sensor feedback data are calculated, wherein the default normal driving model is the driving model being classified by speed per hour.
Further, in the step S3, according to the driving behavior judged, by the vehicle speed data with it is described pre-
If the security model for corresponding to the speed per hour section of driving behavior in normal driving model is compared, the peace in corresponding speed per hour section is found
Full model threshold value, and each data of current form sensor are compared whether in secure threshold section, if sensor values are right
Within the scope of the security value (s) for answering speed per hour section, then determines the driving behavior safety, be otherwise judged to the driving behavior and there is danger
Danger.
Dangerous driving behavior testing and evaluation method according to an embodiment of the present invention based on multisensor, to travelling
The middle driving mode that steering, lane change and the vehicle to turn around occurs is identified and judgeed, and using differentiation as a result, in conjunction with real-time row
Speed threshold value corresponding with safety traffic model is sailed into comparison, to judge the safety of current driving behavior.
The present invention only needs to call gyroscope angular-rate sensor and GPS velocity sensor in common smart machine, no
Need additional hardware device, installation cost is almost nil, and never calls camera, and required memory space is low, not by weather,
The influence of the links factor such as light.The present invention does not also require the placement position of equipment specifically, it is only necessary to keep equipment with
Vehicle is horizontal, and furthermore the computation complexity of algorithm and power consumption are all relatively low, not high to the CPU occupancy of equipment, has general well
Adaptive can in real time judge driving behavior, while carry out Comprehensive Evaluation, tool to entire driving behavior by log recording
Standby preferable traceability.
The additional aspect of the present invention and advantage will be set forth in part in the description, and will partially become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
Above-mentioned and/or additional aspect of the invention and advantage will be apparent in the description from following accompanying drawings to embodiment
Be readily appreciated that, in which:
Fig. 1 is that the ZenDrive of the prior art turns to exemplary diagram using the detection of magnetic effect sensor；
Fig. 2 is the dangerous driving behavior testing and evaluation method based on multisensor according to one embodiment of the invention
Flow chart；
Fig. 3 is the dangerous driving behavior testing and evaluation method based on multisensor according to another embodiment of the present invention
Flow chart；
Fig. 4 is the preset equipment and vehicle relative position schematic diagram according to the embodiment of the present invention；
Fig. 5 is the acquisition according to the dangerous driving behavior testing and evaluation method based on multisensor of the embodiment of the present invention
The left steering waveform example arrived；
Fig. 6 is the collected right turn waveform example figure according to the embodiment of the present invention；
Fig. 7 is the collected primary left lane change waveform example figure according to the embodiment of the present invention；
Fig. 8 is the collected primary right lane change waveform example figure according to the embodiment of the present invention；
Fig. 9 is the collected waveform example figure that once turns around according to the embodiment of the present invention；
Figure 10 is according to the embodiment of the present invention to the threshold value schematic diagram for turning to waveform recognition use；
Figure 11 is the threshold value schematic diagram used to lane change waveform recognition according to the embodiment of the present invention；
Figure 12 is the collected more wave waveform diagrams of a complexity according to the embodiment of the present invention；
Figure 13 is the schematic diagram according to the MultiWave detection finite-state automata of the embodiment of the present invention；
Figure 14 is the direct and road and slow and road and comparison of wave shape figure according to the embodiment of the present invention；
Figure 15 is the driving behavior decision flowchart according to the embodiment of the present invention.
Specific embodiment
The embodiment of the present invention is described below in detail, the example of embodiment is shown in the accompanying drawings, wherein identical from beginning to end
Or similar label indicates same or similar element or element with the same or similar functions.It is retouched below with reference to attached drawing
The embodiment stated is exemplary, it is intended to is used to explain the present invention, and is not considered as limiting the invention.
The present invention proposes that a kind of dangerous driving behavior testing and evaluation method based on multisensor, this method belong to Internet of Things
Network technology field is acquired and is divided to the feedback data of sensor each in vehicle driving in real time by a variety of different sensors
Analysis, to judge that driver is assessed in driving procedure with the presence or absence of dangerous driving behavior, and for its driving procedure.
Fig. 2 is the dangerous driving behavior testing and evaluation method based on multisensor according to one embodiment of the invention
Flow chart.
Step S101 obtains the sensor values in vehicle driving.
Step S102 handles sensor values substitution MultiWave detection algorithm.
Step S103 determines that the driving behavior of vehicle belongs to left and right turn.
Step S104 determines that the driving behavior of vehicle belongs to left and right lane change.
Step S105 determines that the driving behavior of vehicle belongs to left and right and turns around.
Step S106, the vehicle speed data that input GPS velocity sensor detects.
Vehicle speed data is compared step S107 with default normal driving model.
Step S108 is generated and is driven as a result, the driving behavior to vehicle is scored.
As shown in figure 3, the dangerous driving behavior testing and evaluation method based on multisensor of the embodiment of the present invention, including
Following steps:
Step S1 acquires the multiple sensors feedback data in vehicle operation using multiple sensors on vehicle.
In one embodiment of the invention, multiple sensors include: angular-rate sensor and GPS velocity sensor.Its
In, multiple sensors are using one of following two form:
(1) angular-rate sensor and GPS velocity sensor being fixed in vehicle；
(2) angular-rate sensor and GPS velocity sensor of the gyroscope in smart machine are integrated in, wherein intelligence is set
It is standby to be fixed on the front window of vehicle, with reference to shown in Fig. 4.It should be noted that the position of smart machine is not limited to be mounted on vehicle
Front window on, other fixed positions that can also be installed in vehicle, as long as meet do not interfere drive visual angle, herein not
It repeats again.
Multiple sensor feedback data include that the vehicle speed data of GPS velocity sensor acquisition and angular-rate sensor are adopted
The vehicle of collection is in steering, lane change and waveform sensor data when turning around.Figure 12 is according to the collected of the embodiment of the present invention
The more wave waveform diagrams of one complexity.
Step S2 is analyzed multiple sensor feedback data with default driving behavior distinguished number, is tied according to analysis
Fruit identifies the driving mode of vehicle, generates driving behavior analysis result, wherein driving behavior analysis result includes each driving
Behavior and corresponding sensor feedback data.
In one embodiment of the invention, it presets driving behavior distinguished number and uses MultiWave algorithm.It utilizes
MultiWave algorithm detects the sensing data in vehicle traveling process in real time, is filtered to above-mentioned data and shaping, and
According to the Wave data detected, driving mode is identified, such as: left/right steering, left/right lane change, left/right turns around and it
Its behavior.
Sensor preheating is carried out first, obtains initial offset, the sensor values for correction tape noise.
Specifically, it is deviated for sensor initial value in equipment caused by different placement positions, by a pre- Warm status, to biography
Sensor numerical value carries out pre-read and sensor coordinate system remaps, and for statistical analysis to initial static number, obtains one
A sensor values offset as caused by the out-of-flatness of equipment placement position, and the biography that the later period is acquired in real time by offset
Sensor numerical value is modified, and obtains corresponding waveform sensor data.
Then, waved filter is carried out to waveform sensor data, Wave data after filtering is stored in MultiWave chained list.
Sensor values variation, angular velocity sensor when by the steering of vehicle travel process generation, lane change and turning around
The sensor values of acquisition are acquired, and pass through preset minimum sensor threshold value, effective waveform peak threshold, effective waveform
The parameters such as duration and detection waiting time, detect the waveform sensor data of formation.By filtered Wave data and wave
The parameters such as time started, wave end time, wave duration, maximum sensor numerical value, sensor values mean value, the positive negativity of wave with
Wave is unit, is recorded in MultiWave chained list together.
In one embodiment of the invention, the waveform sensor data for meeting the following conditions are carried out:
(1) lower than the Wave data of default angular-rate sensor lower threshold；
(2) maximum value of angular-rate sensor is lower than the waveform number for presetting effective waveform peak threshold value in current single wave
According to；
(3) waveform duration is lower than the Wave data of default basic waveform duration；
When being more than the default detection waiting time new Wave data to be still not detected, this detection will be terminated, start to identify
The driving mode of vehicle.
In one embodiment of the invention, as shown in figure 14, waveforms detection can be carried out in the following manner: being used
This four states of Silence (monitoring), Check (detection), Record (record) and Next (waiting potential wave), which constitute one, to be had
Automatic state machine is limited, sensor values are filtered and are handled in a manner of poll.Table 1 is shown in MultiWave algorithm
Relevant parameter, table 2 shows the data structure of single wave.Figure 10 and Figure 11 respectively illustrates steering and lane change waveform threshold value is retouched
It states.
Table 1
| Parameter name | Type | Description |
| wbegin | long | At the beginning of wave (ms) |
| wend | long | The end time (ms) of wave |
| wdur | long | The duration (ms) of wave |
| wmax | float | The maximum value (rad/s) of this wave |
| pn | boolean | The positive negativity of wave |
| speed | float | GPS velocity (km/h) when waveform formation |
| wavg | float | Angular-rate sensor mean value (rad/s) in wave duration |
Table 2
Separately below to Silence (monitoring), Check (detection), Record (record) and Next (waiting potential wave) four
A state is illustrated.
(1) Silence (monitoring): persistently monitoring the numerical value dz of angular-rate sensor, when dz is greater than threshold value Min_dz, jumps
Go to Check state；Otherwise continue to monitor.
(2) Check (detection):
2.1) when jumping to Check from Silence: persistently monitoring the numerical value dz of angular-rate sensor, work as sensor values
When dz is greater than Min_dz, start recording wave, and be updated to when prewave information (table 2)；When sensor number dz value is lower than Min_
When dz, when prewave terminates, the condition of a wave is constituted when whether prewave meets by the record information judgement of wave, if meeting
Jump to Record state；Otherwise Silence state is jumped to restart.
2.2) when being jumped from Next: the numerical value dz of angular-rate sensor is persistently monitored in Delay time range, if super
When jump then go to Next.If sensor values dz is greater than Min_dz, it is updated to when prewave information, and one week under poll
Phase；When sensor values dz is lower than Min_dz, when prewave terminates, by the record information judgement of wave when whether prewave meets structure
At the condition of a wave, Record state is jumped to if met, otherwise continues poll.
(3) Record (record): will detect chained list MultiWave when the data structure deposit wave of prewave, subsequently into
Next state.
(4) Next (next time): persistently monitoring angular-rate sensor numerical value dz in Delay time range, when dz is lower than
When Min_dz, it is directly entered next round；Otherwise enter Check state, and one new wave of start recording.It is if overtime, i.e., super
Delay time range is crossed, then illustrates that an event terminates, exports MultiWave chained list, jump to Silence and start next time
Monitoring.
Then, the Wave data in MultiWave chained list is analyzed, calculates the quantity and corresponding feature of wave
Value is identified according to the quantity of wave and characteristic value, is identified to the driving mode of vehicle, wherein characteristic value includes: the PN of waveform
Positive negativity, maximum angular rate value, the angular speed mean value of single wave, the duration of single wave and the absolute value of steering angle.
In one embodiment of the invention, the positive negativity of the PN of waveform is calculated by sensor mean value wavg, and wavg >
0, then PN is positive；Otherwise PN is negative.
Steering angle, which is first passed through, obtains single wave steering to the sensor degree mean value of each wave and wave duration quadrature
Angle, then all wave steering angles are summed to obtain total steering angle.
Reference table 2, N are the quantity of wave, and n is the serial number of wave, wavgnFor the angular speed sensing in n-th of wave duration
Device mean value, wdurnThe duration for being wave for n-th.
Specifically, by positive negativity to waveform quantity, each wave in MultiWave chained list and steering angle into
Row comprehensive descision identifies the driving mode of vehicle, comprising:
Single wave:
With reference to Fig. 5 to Fig. 9, when steering, lane change occurs and turns around, other than the difference on direction, if the driving
Behavior is disposably completed, and turns to and turn around that a wave can be only formed, and wherein left steering and left-hand rotation turn around to form a positive wave, right
It turns to and right-hand rotation turns around to form a negative wave.Lane change will form positive minus two waves, and when left lane change is first positive wave, after be negative
Wave；First it is negative wave when right lane change, is afterwards positive wave.
Based on the above principles, the driving mode of the affiliated vehicle of single wave is classified as follows:
Turn left: driving direction is judged by the positive negativity PN state of wave, PN is timing, when prewave is movement to the left.That is,
Waveform is forward wave, then vehicle is turned left.
It turns right: when judging that PN is negative, when prewave is movement to the right.That is, waveform is negative wave, then vehicle is turned right.
It turns around: calculating the total angle of rotation θ of this steering, right-angle steering is 90 degree under normal circumstances, is turned around as 180 degree, but one
As few perfect driving in real travel situations, set as follows to improve fault-tolerance and the stability of judgement: if |
θ | ∈ [65,115], this steering belong to right-angle steering；If | θ | ∈ [155,205], this steering, which belongs to, turns around.
There are a kind of special feelings when lane change, some relatively better drivers of driving habit, the road Shi Bing is slower driving, and makes
The deflection for obtaining second wave is less obvious, is specifically shown in Figure 14, and in this case, detection algorithm possibly can not be examined in given threshold
Measure second wave in lane change traveling.
The steering angle of first wave of left lane change and right lane change is calculated, practical | θ | belong to lane change when [0,20] ∈.Cause
This, also needs to be added in unicast detection logic | θ | ∈ [0,20] judges whether this belongs to lane change situation.
It to sum up, is mild lane change when the absolute value of steering angle is located at section [0,20]；The absolute value of steering angle is located at
Section [65,115] is to turn to；It is to turn around that the absolute value of steering angle, which is located at section [155,205],.
For the situation of two waves and three waves:
If turning to, lane change and turning around to fail disposably to complete, several waves will be formed, but will not be surpassed under normal circumstances
4 waves are crossed, will form waveform as shown in figure 12 under complex situations.At this moment waveform differentiates that situation is more complicated, needs to produce waveform
Multiple wave comprehensive descisions in raw time range, the sensor values to vehicle in entire MultiWave record section are segmented
It is accumulative, judged that calculation formula is as follows by last total steering angle θ:
For convenience of calculating, directly use the mean value of angular-rate sensor degree multiplied by generation duration of wave in the present invention
For formula (2), is calculated, repeated no more using above-mentioned formula (1).
(2) two waves: two continuous equidirectional waves may be because steering or have pause during turning around, so that angular speed becomes
Change, which has, first to increase, then reduces, then increase, and finally turns to and completes, reduces again.When lane change, vehicle can occur contrary twice
Slight steering, that is, formed two opposite waves of positive negativity.For turning left, vehicle can first turn to the left in motion,
It into left-hand lane, then turns to the right, headstock returns just, and sensor values form two opposite waves of wave crest just.
For these reasons, the driving mode of two affiliated vehicles of waveform is classified as follows:
When two waveforms are forward wave, continue to judge steering angle, if the absolute value of steering angle is located at section
It [65,115] is left steering；It is to turn around to the left that the absolute value of steering angle, which is located at section [155,205],；
When two waveforms are negative wave, continue to judge steering angle, if the absolute value of steering angle is located at section
It [65,115] is right turn；It is to turn around to the right that the absolute value of steering angle, which is located at section [155,205],；
First waveform is forward wave, and when second waveform is negative wave, vehicle is lane change to the left；
First waveform is negative wave, and when second waveform is forward wave, vehicle is lane change to the right.
Table 3 is the driving mode judging result of two waves.
Table 3
(3) three waves: when there are three wave, the angle in entire steering procedure is directly calculated, and just by steering angle
Negativity come judge turn to direction.
For these reasons, the driving mode of three affiliated vehicles of waveform is classified as follows: if steering angle is absolute
Value is mild lane change when being located at section [0,20]；The absolute value of steering angle is located at section [65,115] to turn to；Steering angle
Absolute value be located at section [155,205] be turn around.
(4) more than three waves: when the waveform quantity in MultiWave chained list is more than or equal to 4 at this time, the vehicle short time is indicated
Occur acutely to swing, driver is without stably driving vehicle straight trip or rationally being turned to, and the vehicle body of swing is to nearby vehicle shape
It is threatened at certain, belongs to dangerous driving.
Figure 15 is the driving behavior decision flowchart according to the embodiment of the present invention.
Step S1501 reads the data in MultiWave chained list.
Step S1502 judges the quantity of wave.
Step S1503 judges the positive negative shape PN of waveform when the quantity for judging wave is 1 or 2.
Step S1504 according to judging that vehicle turns left or turn right, and judges wave in the case where 1 or 2 wave
Quantity when being 3, further judge the absolute value of steering angle.
Step S1505 then determines that the driving behavior of vehicle belongs to change if the absolute value of steering angle is located at [0,20]
Road；
Step S1506 then determines that the driving behavior of vehicle belongs to and turns if the absolute value of steering angle is located at [65,115]
To；
Step S1507 then determines that the driving behavior of vehicle belongs to if the absolute value of steering angle is located at [155,205]
It turns around；
Step S1508 determines that the driving behavior of vehicle belongs to dangerous driving when the quantity for judging wave is more than 3.
Driving behavior analysis result is compared with default normal driving model, whether judges driving behavior by step S3
It is dangerous, and for each driving behavior and the whole driving behavior assessment report for generating vehicle is driven, utilize waveform
Various features identify vehicle running state and are driving procedure scoring according to code of points.
Normal driving model is preset first below to be illustrated.
In one embodiment of the invention, normal driving model is preset by the normal vehicle operation under different speed per hours
Sensor feedback data are calculated.Wherein, presetting normal driving model is the driving model being classified by speed per hour.
It should be noted that current normal driving model is driven at low speed situation modeling mainly for urban district, and normally drive
Sailing model can constantly be added by the later period, kept model more perfect, enabled to handle more travel situations.
Specifically, by being fixed on the smart machine of front windshield of vehicle, vehicle in the case of different preset vehicle speeds is recorded
It is turned in conventional city road, lane change and when turning around, waveform quantity that the angular-rate sensor numerical value in gyroscope is formed,
Waveform duration (wdur), the extreme value (wmax) of wave numerics, the positive negativity of angular-rate sensor mean value (wavg) and wave
Etc. data.Wherein wdur unit is millisecond (ms), wmax and wavg unit is degree per second (rad/s).Then by largely counting
According to acquisition and comparison, determines steering, lane change and the security model threshold value to turn around, corresponding security parameter section is generated, such as 4 institute of table
Show.
Table 4
It finds after tested, the angular-rate sensor numerical value under normal road condition when right-angle steering is most worth generally higher than 0.25rad/
S, section is at [0.29,0.76] (rad/s), and the wave duration is at [3600,7500] (ms), and turning velocity can not generally be greater than
30km/h, i.e. speed per hour are more than the steering behavior of 30km/h than relatively hazardous.
Meanwhile speed per hour is inversely proportional with the wave duration, is most worth to angular speed and mean value is directly proportional.Angular speed turns to linear velocity
Formula (3) are changed, wherein v is linear velocity, and w is angular speed, and r is turning radius.
It is consistent with test result by formula (3) it is found that theoretically real-time linear velocity is directly proportional to angular speed.Meanwhile speed per hour
Under the same conditions, angular-rate sensor numerical value when right turn, which is most worth, generally higher than turns left, while the wave duration is lower than a left side
Turn the wave duration, this also complies with the case where Ordinary Rd crossing left steering radius is greater than right turn radius, and specific value is shown in Table
4, Fig. 5 and Fig. 6 is respectively the waveform diagram that left and right turn is formed.
V=wr (3)
With reference to Fig. 4 and Fig. 5, for lane change, be generally made of under normal circumstances two continuous, opposite waves: left lane change by
One forward wave and a negative wave are constituted；Right lane change is made of a negative wave and a forward wave.Lane change waveform is by real-time
It is unobvious that speed influences variation, but maximum angular rate numerical value and wave duration are all smaller than turning to numerical value, and specific value is shown in
Table 5.In addition, the sum of the numerical value mean value and the product of wave duration of forward wave and negative wave is typically approximately zero, calculate as public
Formula (4), wherein wavg is to turn to waveform mean value, and wdur is to turn to waveform duration.
wavg+*wdur++wavg-*wdur-≈0 (4)
Table 5
For turning around, in addition to angular speed changes greatly during turning around, the mean value and waveform of angular-rate sensor continue
Total time it is equally all very big.Turn to more close under angular speed mean value and high-speed case when turning around, but the wave duration is most
It is long, it is 180 degree or so, reference table 6 in 10s or so, total steering angle θ.
| Wmax | Wavg | Wdur | θ | |
| It turns around on a left side | 0.5 | 0.31 | 9510 | 168 |
| It turns around on the right side | -0.56 | -0.33 | 10063 | -177 |
Table 6
According to the steering analyzed in step S2, lane change and turn around as a result, the waveform obtained in conjunction with GPS velocity sensor
The vehicle speed data of beginning is mapped to the driving behavior category of model in corresponding speed section, to when front steering data and default
Normal driving behavior model carries out data comparison, whether in the reasonable scope to judge each value, if current driving waveform is each
Item data is considered as safety in security value (s) interval range, otherwise it is assumed that dangerous.
Specifically, it according to the driving behavior judged, drives vehicle speed data is corresponding with default normal driving model
The security model for sailing the speed per hour section of behavior is compared, and finds the security model threshold interval in corresponding speed per hour section, and compares
Whether each data of current form sensor then determine driving behavior safety, otherwise determine driving behavior in secure threshold section
It is dangerous.
Safe scoring is carried out to single driving event in real time first, it is comprehensive then to drive whole overall behavior progress to certain
Close scoring.
(1) single driving behavior is scored, to the Quantitative marking of each driving behavior:
In single driving behavior scoring, using friction speed under secure threshold float up and down 20% as security interval,
The processing that rounds up is carried out to real-time speed position, re-maps security interval.
Code of points: each driving behavior is judged using 5 points of system scorings.Main factor of judgment includes: wave
Mean value (wavg), the most value (wmax) of wave and wave duration (wdur), code of points is shown in Table 7.Wherein turn in the case of unicast
To with turn around directly to use waveform mean value, waveform most value and waveform duration to calculate；In the case where greater than one wave, waveform is held
The continuous time adds up as last total duration, and waveform mean value and waveform are most worth the mean value computation using multiple waves.For becoming
Road, to averaging after the calculating separately of former and later two waves.The note that single parameter belongs to corresponding security parameter section is once effective.It is right
0 point is directly determined in the situation that wave number amount is more than 3.
| Effective degree | Score value |
| 3 | 5 |
| 2 | 4 |
| 1 | 2 |
| 0 | 1 |
| Wave number amount >=4 | 0 |
Table 7
(2) this drives comprehensive score, carries out comprehensive score to driving behaviors a series of in a driving procedure.
Single driving behavior and score value are recorded, the mean value that last total score value takes each driving to score, 3 points or more
It is unqualified lower than 3 points for qualification.Calculate the frequency of occurrence of each driving behavior and the qualified ratio of each behavior.And it is low to score value
In 3 points, and behavior accounts for driving behavior of total driving behavior quantity greater than 20% and is recorded and exported, for prompting driver to need
Carry out concern and improved driving behavior.
To sum up, the present invention can be analyzed potential using the multinomial sensor capture steering of vehicle, lane change and the events such as turn around
Dangerous driving behavior, and provide a set of driving scoring.
The present invention detects vehicle using a plurality of high-, middle- and low-end smart machines, and detected type of vehicle includes small-sized
Car three and SUV two, participating in tester includes special train driver, taxi driver and private car car owner, drives the testing time
It is about 40 hours total.Test SUV vehicle driving event 336 times；Special train car 462 times；Taxi car 420 times.
Test result is that left and right turn detects bat 100%, and lane change detects bat 89%, turns around to detect
Bat 93%.
According to test result analysis: low side smart machine discrimination is lower, and about 85%, the discrimination 90% that turns around is left
It is right；Middle and high end smart machine (Samsung s3, Samsung s4, association S850 and Nexus6) discrimination is each about 90%, and turn around discrimination
About 94%.Compared with middle and high ends type, the detection performance of low side smart phone by sensor and CPU computing capability influenced compared with
Greatly, every discrimination is lower.
Dangerous driving behavior testing and evaluation method according to an embodiment of the present invention based on multisensor, to travelling
The middle driving mode that steering, lane change and the vehicle to turn around occurs is identified and judgeed, and using differentiation as a result, in conjunction with real-time row
Speed threshold value corresponding with safety traffic model is sailed into comparison, to judge the safety of current driving behavior.
The present invention only needs to call gyroscope angular-rate sensor and GPS velocity sensor in common smart machine, no
Need additional hardware device, installation cost is almost nil, and never calls camera, and required memory space is low, not by weather,
The influence of the links factor such as light.The present invention does not also require the placement position of equipment specifically, it is only necessary to keep equipment with
Vehicle is horizontal, and furthermore the computation complexity of algorithm and power consumption are all relatively low, not high to the CPU occupancy of equipment, has general well
Adaptive can in real time judge driving behavior, while carry out Comprehensive Evaluation, tool to entire driving behavior by log recording
Standby preferable traceability.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.In the present specification, schematic expression of the above terms are not
Centainly refer to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be any
One or more embodiment or examples in can be combined in any suitable manner.
Although the embodiments of the present invention has been shown and described above, it is to be understood that above-described embodiment is example
Property, it is not considered as limiting the invention, those skilled in the art are not departing from the principle of the present invention and objective
In the case where can make changes, modifications, alterations, and variations to the above described embodiments within the scope of the invention.The scope of the present invention
It is extremely equally limited by appended claims.
Claims (7)
1. a kind of dangerous driving behavior testing and evaluation method based on multisensor, which comprises the steps of:
Step S1 acquires the multiple sensors feedback data in the vehicle operation using multiple sensors on vehicle；
Step S2 is analyzed the multiple sensors feedback data with default driving behavior distinguished number, is tied according to analysis
Fruit identifies the driving mode of the vehicle, generates driving behavior analysis result, wherein the driving behavior analysis result includes
Each driving behavior and corresponding sensor feedback data, wherein the multiple sensor include: angular-rate sensor and
GPS velocity sensor；The multiple sensor is using one of following two form:
(1) angular-rate sensor and GPS velocity sensor being fixed in the vehicle；
(2) angular-rate sensor and GPS velocity sensor of the gyroscope in smart machine are integrated in, wherein the intelligence is set
It is standby to be fixed on the vehicle front window；
The default driving behavior distinguished number uses MultiWave algorithm；
The driving behavior analysis result is compared with default normal driving model, judges the driving behavior by step S3
With the presence or absence of danger, and for each driving behavior and drive the whole driving behavior assessment report for generating the vehicle
It accuses.
2. the dangerous driving behavior testing and evaluation method based on multisensor as described in claim 1, which is characterized in that more
A sensor feedback data include the vehicle speed data and angular-rate sensor acquisition of the GPS velocity sensor acquisition
Vehicle in steering, lane change and waveform sensor data when turning around.
3. the dangerous driving behavior testing and evaluation method based on multisensor as described in claim 1, which is characterized in that
In the step S2,
Firstly, carrying out waved filter to the waveform sensor data, Wave data after filtering is stored in MultiWave chained list；
Then, the Wave data in the MultiWave chained list is analyzed, calculates the quantity and corresponding feature of wave
Value is identified according to the quantity of the wave and characteristic value, is identified to the driving mode of the vehicle, wherein the characteristic value
It include: the positive negativity of PN, maximum angular rate value, the angular speed mean value of single wave, the duration of single wave and the steering angle of waveform
The absolute value of degree.
4. the dangerous driving behavior testing and evaluation method based on multisensor as claimed in claim 3, which is characterized in that mistake
Filter the waveform sensor data for meeting the following conditions:
(1) lower than the Wave data of default angular-rate sensor lower threshold；
(2) maximum value of single wave sensor numerical value is lower than the Wave data for presetting effective waveform peak threshold value；
(3) waveform duration is lower than the Wave data of default basic waveform duration；
When being more than the default detection waiting time new Wave data to be still not detected, this detection will be terminated, start identification and drive
Behavior.
5. the dangerous driving behavior testing and evaluation method based on multisensor as claimed in claim 3, which is characterized in that in institute
It states and is identified according to the quantity and characteristic value of wave, the driving mode of the vehicle is identified, comprising:
(1) single wave: waveform is forward wave, then the vehicle is turned left out；Waveform is negative wave, then the vehicle is turned right；
The absolute value of steering angle is mild lane change when being located at section [0,20]；The absolute value of steering angle is located at section [65,115]
It turns to；It is to turn around that the absolute value of steering angle, which is located at section [155,205],；
(2) two waves:
When two waveforms are forward wave, continue to judge steering angle, if the absolute value of steering angle be located at section [65,
It 115] is left steering；It is to turn around to the left that the absolute value of steering angle, which is located at section [155,205],；
When two waveforms are negative wave, continue to judge steering angle, if the absolute value of steering angle be located at section [65,
It 115] is right turn；It is to turn around to the right that the absolute value of steering angle, which is located at section [155,205],；
First waveform is forward wave, and when second waveform is negative wave, the vehicle is lane change to the left；
First waveform is negative wave, and when second waveform is forward wave, the vehicle is lane change to the right；
(3) three waves: if the absolute value of steering angle is mild lane change when being located at section [0,20]；The absolute value of steering angle
It is to turn to positioned at section [65,115]；It is to turn around that the absolute value of steering angle, which is located at section [155,205],；
(4) more than three waves: the vehicle short time occurs acutely to swing.
6. the dangerous driving behavior testing and evaluation method based on multisensor as described in claim 1, which is characterized in that
In the step S3, presets normal driving model and calculated by the sensor feedback data in the normal vehicle operation under different speed per hours
It obtains, wherein the default normal driving model is the driving model being classified by speed per hour.
7. the dangerous driving behavior testing and evaluation method based on multisensor as claimed in claim 6, which is characterized in that
It, will be in the vehicle speed data and the default normal driving model according to the driving behavior judged in the step S3
The security model in the speed per hour section of corresponding driving behavior is compared, and finds the security model threshold value in corresponding speed per hour section, and right
Whether data more each than current form sensor are in secure threshold section, if sensor values are in the safety in corresponding speed per hour section
In numberical range, then determines the driving behavior safety, otherwise determine that the driving behavior is dangerous.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510813444.1A CN105574537B (en) | 2015-11-23 | 2015-11-23 | Dangerous driving behavior testing and evaluation method based on multisensor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510813444.1A CN105574537B (en) | 2015-11-23 | 2015-11-23 | Dangerous driving behavior testing and evaluation method based on multisensor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105574537A CN105574537A (en) | 2016-05-11 |
| CN105574537B true CN105574537B (en) | 2018-12-28 |
Family
ID=55884643
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510813444.1A Expired - Fee Related CN105574537B (en) | 2015-11-23 | 2015-11-23 | Dangerous driving behavior testing and evaluation method based on multisensor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105574537B (en) |
Families Citing this family (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108122298B (en) * | 2016-11-30 | 2021-06-29 | 厦门雅迅网络股份有限公司 | Method and system for improving accuracy of steering wheel angle data of vehicle |
| CN106740852A (en) * | 2016-12-15 | 2017-05-31 | 重庆典果成实业有限公司 | The method that safe auxiliary is carried out during for motorcycle turns |
| CN106627871A (en) * | 2016-12-15 | 2017-05-10 | 重庆典果成实业有限公司 | Safety auxiliary system for motorcycle |
| CN106682295A (en) * | 2016-12-16 | 2017-05-17 | 清华大学 | Analysis method for evaluating real-time safety characteristics of drivers |
| CN106882141B (en) * | 2017-02-16 | 2020-02-21 | 深圳市航天无线通信技术有限公司 | Driving behavior detection method and device |
| CN107024358B (en) * | 2017-05-23 | 2019-03-01 | 徐工集团工程机械股份有限公司 | A kind of operating loading analysis system based on multiple-sensor network |
| JP6792704B2 (en) | 2017-06-02 | 2020-11-25 | 本田技研工業株式会社 | Vehicle control devices and methods for controlling self-driving cars |
| JP6901555B2 (en) | 2017-06-02 | 2021-07-14 | 本田技研工業株式会社 | Vehicle control devices and methods for controlling self-driving cars |
| US20190009785A1 (en) * | 2017-07-05 | 2019-01-10 | Panasonic Intellectual Property Management Co., Ltd. | System and method for detecting bullying of autonomous vehicles while driving |
| CN107444404B (en) * | 2017-07-25 | 2019-10-25 | 武汉依迅北斗空间技术有限公司 | A kind of evaluation method and device of driving behavior standard degree |
| CN109754595B (en) * | 2017-11-01 | 2022-02-01 | 阿里巴巴集团控股有限公司 | Vehicle risk assessment method and device and interface equipment |
| CN110211401A (en) * | 2018-02-28 | 2019-09-06 | 长城汽车股份有限公司 | A kind of dangerous tip method and device of driving |
| CN109002797B (en) * | 2018-07-16 | 2023-07-25 | 腾讯科技（深圳）有限公司 | Vehicle lane change detection method, device, storage medium and computer equipment |
| CN111599164B (en) * | 2019-02-21 | 2021-09-24 | 北京嘀嘀无限科技发展有限公司 | Driving abnormity identification method and system |
| CN109878530B (en) * | 2019-02-28 | 2021-04-06 | 中国第一汽车股份有限公司 | Method and system for identifying lateral driving condition of vehicle |
| CN110706517B (en) * | 2019-11-07 | 2020-08-28 | 深圳鼎然信息科技有限公司 | Intelligent safety early warning method, device and system for vehicle and storage medium |
| CN111783536A (en) * | 2020-05-29 | 2020-10-16 | 北京百度网讯科技有限公司 | Method, device and equipment for evaluating lane change reasonability and storage medium |
| CN112287797A (en) * | 2020-10-23 | 2021-01-29 | 北京嘀嘀无限科技发展有限公司 | Data processing method and device, electronic equipment and readable storage medium |
| CN114172705B (en) * | 2021-11-29 | 2023-11-14 | 北京国瑞数智技术有限公司 | Network big data analysis method and system based on pattern recognition |
| CN114221794B (en) * | 2021-11-29 | 2024-02-02 | 北京国瑞数智技术有限公司 | User behavior analysis method, system and medium based on improved pattern recognition algorithm |
| CN115063055A (en) * | 2022-08-17 | 2022-09-16 | 山东宇翔车桥制造有限公司 | Semitrailer equipment operation performance supervisory systems based on data analysis |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104463244A (en) * | 2014-12-04 | 2015-03-25 | 上海交通大学 | Aberrant driving behavior monitoring and recognizing method and system based on smart mobile terminal |
| CN104527647A (en) * | 2014-12-15 | 2015-04-22 | 清华大学 | Monitoring and evaluation method of driving behavior risk degrees |
| CN104802737A (en) * | 2015-03-25 | 2015-07-29 | 清华大学 | Mobile phone based vehicle abnormality driving behavior detection method |
-
2015
- 2015-11-23 CN CN201510813444.1A patent/CN105574537B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104463244A (en) * | 2014-12-04 | 2015-03-25 | 上海交通大学 | Aberrant driving behavior monitoring and recognizing method and system based on smart mobile terminal |
| CN104527647A (en) * | 2014-12-15 | 2015-04-22 | 清华大学 | Monitoring and evaluation method of driving behavior risk degrees |
| CN104802737A (en) * | 2015-03-25 | 2015-07-29 | 清华大学 | Mobile phone based vehicle abnormality driving behavior detection method |
Non-Patent Citations (1)
| Title |
|---|
| 基于多传感器信息融合的危险行为检测系统;詹彤 等;《汽车零部件》;20110728;第54-57页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105574537A (en) | 2016-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105574537B (en) | Dangerous driving behavior testing and evaluation method based on multisensor | |
| US20220262239A1 (en) | Determining causation of traffic events and encouraging good driving behavior | |
| CN106934876B (en) | A kind of recognition methods and system of vehicle abnormality driving event | |
| CN103043057B (en) | Abnormal driving based on vehicle position information judges and warning system | |
| US9767622B2 (en) | System and a method for improved car prognosis | |
| CN105324275B (en) | Movement pattern device and movement pattern method | |
| Tageldin et al. | Can time proximity measures be used as safety indicators in all driving cultures? Case study of motorcycle safety in China | |
| CN110488802A (en) | A kind of automatic driving vehicle dynamic behaviour decision-making technique netted under connection environment | |
| CN105761149A (en) | Vehicle insurance system and vehicle insurance charging method | |
| CN109643494A (en) | Automatic driving vehicle, the parking method of automatic driving vehicle and program | |
| CN104246850B (en) | Crawl decision maker | |
| CN106644503A (en) | Intelligent vehicle planning capacity testing platform | |
| Sun et al. | An integrated solution for lane level irregular driving detection on highways | |
| CN106184220B (en) | Abnormal driving detection method in a kind of track based on vehicle location track | |
| US10249112B2 (en) | Vehicle state monitoring apparatus, vehicle state monitoring system, and vehicle state monitoring method | |
| CN104192063B (en) | Vehicle safe driving caution system and corresponding alarming method for power | |
| CN106373461A (en) | Driver training management service system and method | |
| CN106209546A (en) | Based on binocular camera and area array cameras automatic with car system | |
| Wu et al. | L3: Sensing driving conditions for vehicle lane-level localization on highways | |
| CN106427825A (en) | Automobile, user terminal, as well as automobile safety monitoring method and automobile safety monitoring system based on traveling data | |
| CN108960083A (en) | Based on automatic Pilot objective classification method combined of multi-sensor information and system | |
| CN107662613B (en) | A kind of extreme driving behavior recognition methods and system based on mobile intelligent perception | |
| CN108482385B (en) | A kind of mobile unit operational risk discrimination method considering traffic environment complexity | |
| CN109591825A (en) | A kind of driving fatigue detection method, device and vehicle | |
| CN109720353A (en) | A kind of driving behavior detection method based on smart phone |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20181228 Termination date: 20201123 |