CN111547064B - Driving style recognition and classification method for automobile adaptive cruise system - Google Patents
Driving style recognition and classification method for automobile adaptive cruise system Download PDFInfo
- Publication number
- CN111547064B CN111547064B CN202010454845.3A CN202010454845A CN111547064B CN 111547064 B CN111547064 B CN 111547064B CN 202010454845 A CN202010454845 A CN 202010454845A CN 111547064 B CN111547064 B CN 111547064B
- Authority
- CN
- China
- Prior art keywords
- vehicle
- driver
- driving style
- speed
- experimental
- 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
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 10
- 238000012549 training Methods 0.000 claims abstract description 25
- 238000002474 experimental method Methods 0.000 claims description 82
- 238000012360 testing method Methods 0.000 claims description 52
- 210000002569 neuron Anatomy 0.000 claims description 27
- 239000013598 vector Substances 0.000 claims description 26
- 230000001133 acceleration Effects 0.000 claims description 16
- 238000013139 quantization Methods 0.000 claims description 12
- 230000014509 gene expression Effects 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 4
- 238000013459 approach Methods 0.000 claims description 3
- 230000036461 convulsion Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000003068 static effect Effects 0.000 claims 1
- 230000007935 neutral effect Effects 0.000 description 3
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
- B60W40/09—Driving style or behaviour
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2520/00—Input parameters relating to overall vehicle dynamics
- B60W2520/10—Longitudinal speed
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2520/00—Input parameters relating to overall vehicle dynamics
- B60W2520/10—Longitudinal speed
- B60W2520/105—Longitudinal acceleration
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2554/00—Input parameters relating to objects
- B60W2554/80—Spatial relation or speed relative to objects
- B60W2554/802—Longitudinal distance
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2554/00—Input parameters relating to objects
- B60W2554/80—Spatial relation or speed relative to objects
- B60W2554/804—Relative longitudinal speed
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/80—Technologies aiming to reduce greenhouse gasses emissions common to all road transportation technologies
- Y02T10/84—Data processing systems or methods, management, administration
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Mathematical Physics (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention belongs to the technical field of automobile driver driving style identification, and particularly relates to a driving style identification and classification method for an automobile adaptive cruise system; acquiring driver data of the following driving conditions of the automobile, clustering the driver data by using an SOM and K-means algorithm, identifying the driving style of each driver off line, training a driving style online classifier by using a driving style identification result, and performing online identification on the driving styles of different drivers by using the trained driving style online classifier; the trained online driving style classifier is combined with the automobile ACC system, so that the driving style of a driver can be identified online, the automobile ACC system can make corresponding adjustment aiming at the drivers with different driving styles, the individual requirements of different drivers with different driving styles are met, the comfort and the acceptance of the automobile ACC system are improved, and the driving experience of the driver is improved.
Description
Technical Field
The invention belongs to the technical field of automobile driver driving style identification, and particularly relates to a driving style identification and classification method for an automobile adaptive cruise system.
Background
With the rapid development of automobile intellectualization and automation, automobile automatic driving technology is more and more concerned by people, and an automobile advanced assistant driving system is more and more applied to visible automobiles in the market as a primary stage of the automobile automatic driving technology. The automobile self-adaptive cruise (ACC) system is an important component of an advanced auxiliary driving system of an automobile, can control the automobile to automatically follow the front automobile, can reduce the driving burden of a driver to a great extent, and can improve the driving safety of the driver. However, the current ACC system of the automobile basically has only one fixed mode, and can only meet the average requirement of the driver, but cannot meet the individual requirement of the driver.
Disclosure of Invention
In order to overcome the problems, the invention provides a driving style identification and classification method for an automobile adaptive cruise system, which is an off-line identification and on-line classification method for the driving style of a driver of the automobile adaptive cruise system; the method comprises the steps of collecting driver data of an automobile following driving condition, clustering the driver data by using a self-organizing mapping neural network (SOM) and a K-means (K-means) algorithm, identifying the driving style of each driver off-line, training a driving style on-line classifier by using a driving style identification result to obtain a trained driving style on-line classifier, combining the trained driving style on-line classifier and an automobile ACC system, and realizing on-line identification of the driving style of the driver, so that the automobile ACC system can make corresponding adjustment aiming at the drivers with different driving styles, thereby meeting individual requirements of different drivers with different driving styles, improving the comfort and acceptance of the automobile ACC system, and simultaneously improving the driving experience of the driver.
The invention is realized by the following technical scheme:
a driving style recognition and classification method for an automobile self-adaptive cruise system is characterized by collecting driver data of running conditions of an automobile and the automobile, clustering the driver data by using an SOM and K-means algorithm, recognizing the driving style of each driver off line, training a driving style online classifier by using a driving style recognition result, and performing online recognition on the driving styles of different drivers by using the trained driving style online classifier; the method specifically comprises the following steps:
step one, car following experiment data acquisition
Firstly, randomly recruiting more than 50 drivers, respectively driving the same test vehicle to drive along with the same front vehicle on the same road section under the same vehicle following experiment working condition by each driver, wherein each driver can drive along with the front vehicle according to own style and habit, and acquiring four data of the speed of the test vehicle, the acceleration of the test vehicle, the distance between the test vehicle and the front vehicle and the relative speed of the test vehicle and the front vehicle of each driver in the vehicle following experiment;
step two, off-line recognition of the driving style of the driver
Taking a following vehicle distance minimum value, a following vehicle distance average value, an impact maximum value, an impact minimum value, a relatively far vehicle speed maximum value, a relatively close vehicle speed maximum value, a positive impact average value, a negative impact average value, a relatively far vehicle speed average value and a relatively close vehicle speed average value as characteristic variables for representing the driving style of drivers, calculating the driving style characteristic variables of the drivers according to the data of the drivers obtained in a following vehicle experiment, combining the driving style characteristic variables of each driver into a driving style characteristic variable vector, and arranging the driving style characteristic variable vectors of all the drivers into a label-free driver sample set to be identified;
determining the number of neurons in a competition layer of the SOM, and performing preliminary clustering on a label-free driver sample set by using the SOM to obtain the category of each driver; performing secondary clustering on the primary clustering result obtained by the SOM network by using a K-means algorithm to obtain a final off-line recognition result of the driving style of each driver in a following experiment;
step three, training the driving style on-line classifier
Forming a labeled driver sample set according to the off-line recognition result of the driving style of each driver in the following experiment obtained in the step two; the method comprises the steps of using a probabilistic neural network method as a driver driving style online classifier, training the classifier by using a formed labeled driver sample set to obtain a trained driver driving style online classifier, embedding the trained driver driving style online classifier into an automobile ACC system, providing driving style class information of a current driver for the automobile ACC system, and enabling the automobile ACC system to be switched to different modes according to the driving style class of the driver.
The following experiment in the step I comprises the following specific contents:
firstly, randomly recruiting more than 50 drivers;
secondly, selecting a following experiment working condition for the automobile self-adaptive cruise system, and selecting a dry and gentle road section with good road surface condition as an experiment road section;
the following experiment is completed by two vehicles together, one is a front pilot vehicle, which is called a front vehicle for short, and the other is an experiment following vehicle, which is called an experiment vehicle for short; during the experiment, all the drivers are respectively driven to drive the same experimental vehicle to run along with the same front vehicle on the same experimental road section, wherein the front vehicle is driven according to the designed experimental process, the drivers of the experimental vehicles run along with the front vehicle according to own style habits, and finally four kinds of data including the speed of the experimental vehicle, the acceleration of the experimental vehicle, the distance between the experimental vehicle and the front vehicle and the relative speed of the experimental vehicle and the front vehicle of each driver in the vehicle following experiment are collected.
The experimental process designed for the driving of the front vehicle in the step one is specifically as follows:
(1) accelerating the front vehicle to 30km/h from the standstill, and maintaining the speed for 15 s;
(2) accelerating the front vehicle to 50km/h and maintaining the speed for 15 s;
(3) accelerating the front vehicle to 70km/h, and maintaining the speed for 15 s;
(4) the front vehicle decelerates to 50km/h and keeps the speed for 15 s;
(5) the front vehicle decelerates to 30km/h and keeps the speed for 15 s;
(6) the front vehicle decelerates to a standstill.
In the first step, the following experiment needs to be carried out twice by each driver in the positive direction and the negative direction of the same experimental section, and the speed of the experimental vehicle, the acceleration of the experimental vehicle, the distance between the experimental vehicle and the front vehicle and the relative speed of the experimental vehicle and the front vehicle in the positive and negative experiments are collected as experimental data.
Determining the number of neurons in a competition layer of the SOM network in the second step specifically comprises the following steps: and calculating the numerical values of the quantization error and the topological error of the SOM network with different competition layer neuron numbers under the condition of data input aiming at the data acquired in the following experiment, and determining the competition layer neuron number of the SOM network by balancing the quantization error, the topological error and the limit of the number of the driver samples obtained in the experiment.
The off-line recognition of the driving style of the driver in the second step specifically comprises the following steps:
A. car following experiment data preprocessing
The driving style characteristic variables and the calculation formula are selected as follows,
(1) minimum distance d for following vehiclehmin: the following time interval is the time required for the head of the test vehicle to travel to the tail of the front vehicle at the current speed, and the following time interval is constant and positive;
wherein d issIs the vehicle distance, i.e. the distance between the tail of the preceding vehicle and the head of the experimental vehicle, vfFor the running speed of the experimental vehicle, n is the number of data points of each driver collected in the following experiment, and n is Te/Ts，TeFor the duration of the car-following experiment, TsThe operation period of the experimental data acquisition equipment is set;
(2) distance mean value when following vehicle
(3) Maximum value of impact JKmax: the impact degree is the derivative of the acceleration of the experimental vehicle, and the maximum value of the impact degree is always a positive value;
wherein a is the acceleration of the experimental vehicle, and t is time;
(4) minimum value of jerk JKmin: the minimum value of the impact degree is always a negative value;
(5) relatively far away from the maximum value of vehicle speed Δ vamax: the relative far-away speed is the relative speed when the test vehicle is far away from the front vehicle, namely the relative speed of the test vehicle is smaller than that of the front vehicle, and the relative far-away speed value is a constant positive value;
Δvamax＝max{(vp-vf)i,i＝1,2,...,n}
wherein v ispThe running speed v of the preceding vehiclefThe running speed of the test vehicle is obtained;
(6) relatively close to the maximum value of vehicle speed Δ vcmax: the relative approaching speed is the relative speed when the test vehicle approaches the front vehicle, namely the relative speed of the two vehicles when the speed of the test vehicle is greater than that of the front vehicle, and the relative departing speed value is constant to be a negative value;
Δvcmax＝min{(vp-vf)i,i＝1,2,...,n}
(7) average value of positive impact
Positive impact JKpRepresents a value when the impact is greater than zero; ,
wherein n iszCounting the number of data points with the impact degree larger than zero in the following experiment data;
(8) average value of negative impact
Negative impact JKnRepresents a value when the impact is less than zero;
wherein n isfCounting the number of data points with the impact degree smaller than zero in the following experimental data;
(9) relatively far from the average value of vehicle speed
Wherein n isaFor the number of data points when the speed of the experimental vehicle is less than that of the front vehicle in the following experimental data, delta vaIs relatively far away from the vehicle speed;
(10) relatively close to the average value of vehicle speed
Wherein n iscThe number of data points, delta v, of the following experimental data when the speed of the experimental vehicle is greater than that of the front vehiclecIs relatively close to the vehicle speed;
calculating the ten driving style characteristic variables of each driver according to the calculation formula and the four data recorded in the following experiment collected in the step one, and combining the driving style characteristic variables of each driver into a driving style characteristic variable vector, wherein each driving style characteristic variable vector represents a driver sample; combining all the driver samples to obtain a label-free driver sample set SuThe mathematical expression of the set is as follows,
wherein x isiRepresenting a driving style characteristic variable vector composed of 10 driving style characteristic variables, namely a driver sample, N representing the number of the driver samples, namely the number of drivers participating in the experiment, T representing onePerforming mathematical operation, namely solving the transposition of the matrix;
B. offline recognition of driver's driving style
Using SOM network for unlabeled driver sample set SuPerforming primary clustering to obtain a label-free driver sample set SuAll the driver samples in the system are divided into a plurality of categories, and the number of the categories is the same as the number of neurons of the set SOM network competition layer;
setting the number of clustering centers of a K-means algorithm, wherein the number represents a plurality of driving styles of drivers, carrying out secondary clustering on a primary clustering result obtained by the SOM network by using the K-means algorithm, and carrying out a label-free driver sample set SuAll the driver samples are divided into the driving style categories with corresponding quantity, and the label-free driver sample set S is realizeduOff-line identification of the driver's driving style in (1).
The training driving style on-line classifier in the third step specifically comprises the following steps:
according to the driving style off-line recognition result obtained in the step two, a label-free driver sample set SuEach driver sample in the system is given with a label of the driving style to which the driver sample belongs, and a labeled driver sample set S is obtainedlThe mathematical expression of the set is as follows,
wherein y isiRepresenting the driving style category to which the driver sample with the number i belongs;
set SlDivided into training set SltAnd test set SlvTwo parts, namely using a probabilistic neural network method as a driving style on-line classifier and using a training set SltTraining a driving style online classifier; after training is completed, test set SlvInputting the driving style to the trained driving style online classifier, testing the classification effect of the driving style online classifier, and outputting a result representing the driving style online classifier to each drivingPrediction of the driving style to which the sample of drivers belongs.
The invention has the beneficial effects that:
1. the method for identifying and classifying the driving style of the driver can provide training data set support for the online classifier of the driving style through the off-line identification of the driver data, does not need to subjectively judge the driving style of the driver, and can obtain a more accurate driving style classification result.
2. The driver driving style identification and classification method provided by the invention uses the SOM network and the K-means algorithm to perform twice clustering on driver data, so that the number of output clusters of the network is not limited when the SOM network is used for performing initial clustering, and a more accurate driver driving style identification result can be obtained.
3. The driving style online classifier in the method for identifying and classifying the driving style of the driver can be embedded into the front end of the automobile ACC system, so that the driving style class information of the driver is provided for the automobile ACC system, and the personalized function of the automobile ACC system is realized.
Drawings
FIG. 1 is a curve of the relationship between the quantization error and the topology error of the SOM network and the number of neurons in the competition layer of the SOM network in the embodiment.
FIG. 2 is a diagram illustrating the preliminary clustering result output by the SOM network in the embodiment.
FIG. 3 is the final clustering result output by the K-means algorithm in the embodiment.
FIG. 4 is a test set S of the driving style online classifier pair in the embodimentlvThe classification result of (1).
Detailed Description
The technical scheme of the invention is described in detail in the following with reference to the attached drawings:
example 1
A driving style identification and classification method for an adaptive cruise system of an automobile, the method comprising the steps of:
step one, vehicle following experiment data acquisition:
considering that the automobile adaptive cruise system mainly acts on a following driving scene, a group of typical following experimental working conditions are designed for data acquisition. The following experiment needs to be completed by two vehicles together, one is a front pilot vehicle, which is called a front vehicle for short; the other is an experiment following vehicle, which is called an experiment vehicle for short. The test car in this embodiment is equipped with an ABS system, an ESC system and a mobiley camera sensor. The Mobileye camera sensor is used for measuring the relative speed and distance information between the test vehicle and the front vehicle, the reference speed of the test vehicle, namely the speed of the test vehicle, and the acceleration, namely the longitudinal acceleration information of the test vehicle are respectively provided by the ABS and the ESC system, and the signals are collected and stored from the CAN bus by using the MicroAutoBox equipment and Controdesk software.
The following experiment is completed by two vehicles together, one is a front pilot vehicle, which is called a front vehicle for short, and the other is an experiment following vehicle, which is called an experiment vehicle for short;
the car following experiment specifically comprises: during the experiment, the road section which is dry, smooth and good in road surface condition is selected as the experimental road section. During an experiment, a front vehicle is driven by a professional driver according to a set experiment flow, a plurality of drivers recruited randomly drive a test vehicle to drive along with the front vehicle according to own style habits, the drivers drive the same test vehicle to drive along with the same front vehicle on the same vehicle-following experiment road section under the same vehicle-following experiment working condition, and four kinds of data including the speed of the test vehicle, the acceleration of the test vehicle, the distance between the test vehicle and the front vehicle and the relative speed of the test vehicle and the front vehicle in a vehicle-following experiment are collected;
more than 50 drivers need to be recruited to collect the data of the car following experiment before the experiment, and 66 drivers are recruited randomly to collect the data of the car following experiment in the embodiment. The experimental procedure of the driving of the front vehicle in the experiment is specifically as follows,
(1) accelerating the front vehicle to 30km/h from the standstill, and maintaining the speed for 15 s;
(2) accelerating the front vehicle to 50km/h and maintaining the speed for 15 s;
(3) accelerating the front vehicle to 70km/h, and maintaining the speed for 15 s;
(4) the front vehicle decelerates to 50km/h and keeps the speed for 15 s;
(5) the front vehicle decelerates to 30km/h and keeps the speed for 15 s;
(6) the front vehicle decelerates to a standstill.
In addition, in order to eliminate the influence of the road gradient on driving data, the following experiment needs to be carried out twice by each driver in the positive direction and the negative direction of the same experiment section, and the speed of an experiment vehicle, the acceleration of the experiment vehicle, the distance between the experiment vehicle and a front vehicle and the relative speed between the experiment vehicle and the front vehicle in the positive direction and the negative direction are collected to be used as experiment data.
Step two, off-line recognition of the driving style of the driver
A. Following experimental data preprocessing
Before offline recognition of the driving style of a driver by using driving data of a following experiment, characteristic variables representing the driving style of the driver need to be determined, the driving style characteristic variables of the driver are obtained through calculation according to the data of the driver obtained in the following experiment, the driving style characteristic variables of each driver are combined into a driving style characteristic variable vector, and the driving style characteristic variable vectors of all the drivers are arranged into a label-free driver sample set to be recognized.
The driving style characteristic variables and the calculation formula selected in the present embodiment are as follows,
(1) minimum distance d when following vehiclehmin: the following time interval is the time required for the head of the test vehicle to travel to the tail of the front vehicle at the current speed, and the following time interval is constant to be a positive value;
wherein d issIs the vehicle distance, i.e. the distance between the tail of the preceding vehicle and the head of the experimental vehicle, vfFor the running speed of the experimental vehicle, n is the number of data points of each driver collected in the following experiment, and n is Te/Ts，TeFor the duration of the car-following experiment, TsThe operation period of the experimental data acquisition equipment is set;
(2) distance mean value when following vehicle
(3) Maximum value of impact JKmax: the impact degree is the derivative of the acceleration of the experimental vehicle, and the maximum value of the impact degree is always a positive value;
wherein a is the acceleration of the experimental vehicle, and t is time;
(4) minimum value of jerk JKmin: the minimum value of the impact degree is always a negative value;
(5) relatively far away from the maximum value of vehicle speed Δ vamax: the relative far-away speed is the relative speed when the test vehicle is far away from the front vehicle, namely the relative speed of the two vehicles when the speed of the test vehicle is less than that of the front vehicle, and the relative far-away speed value is constant to a positive value;
Δvamax＝max{(vp-vf)i,i＝1,2,...,n}
wherein v ispIs the running speed, v, of the preceding vehiclefThe running speed of the test vehicle is obtained;
(6) relatively close to the maximum value of vehicle speed Δ vcmax: the relative approaching speed is the relative speed when the test vehicle approaches the front vehicle, namely the relative speed of the two vehicles when the speed of the test vehicle is greater than that of the front vehicle, and the relative departing speed value is constant to be a negative value;
Δvcmax＝min{(vp-vf)i,i＝1,2,...,n}
(7) average value of positive impact
Positive impact JKpRepresents a value when the impact is greater than zero; ,
wherein n iszCounting the number of data points with the impact degree larger than zero in the following experiment data;
(8) average value of negative impact
Negative impact JKnThe value of the impact degree is less than zero;
wherein n isfCounting the number of data points with the impact degree smaller than zero in the following experimental data;
(9) relatively far from the average value of vehicle speed
Wherein n isaFor the number of data points when the speed of the experimental vehicle is less than that of the front vehicle in the following experimental data, delta vaIs relatively far away from the vehicle speed;
(10) relatively close to the average value of vehicle speed
Wherein,ncthe number of data points, delta v, of the following experimental data when the speed of the experimental vehicle is greater than that of the front vehiclecIs relatively close to the vehicle speed;
calculating the ten driving style characteristic variables of each driver according to the calculation formula and the driver data recorded in the following experiment, and combining the driving style characteristic variables of each driver into a driving style characteristic variable vector, wherein each driving style characteristic variable vector represents a driver sample; combining all the driver samples to obtain a label-free driver sample set Su(i.e., a sample set of drivers of unknown driving style), the mathematical expression for this set is as follows,
wherein x isiThe driving style characteristic variable vector is composed of 10 driving style characteristic variables, namely, a driver sample, N represents the number of the driver samples, namely, the number of drivers participating in the experiment, and the number of the drivers participating in the experiment is 66 in the embodiment; t denotes a mathematical operation, i.e. transposing a matrix.
B. Offline recognition of driver's driving style
Obtaining a label-free driver sample set SuThereafter, the SOM network is used to sample a set S of unlabeled driversuPerforming primary clustering, and performing secondary clustering on a primary clustering result obtained by the SOM network by using a K-means algorithm to realize off-line identification of the driving style of a driver sample in a set;
using SOM network to unlabeled driver sample set SuBefore clustering, the number of neurons in a competition layer of the SOM network needs to be determined. The invention determines the number of neurons in a competition layer of the SOM network by balancing quantization errors, topological errors and the limit of the number of driver samples obtained in experiments, the definitions and expressions of the quantization errors and the topological errors are as follows,
quantization error is per input sample to winning nodeThe average distance of the points is used for measuring the resolution of the SOM network to the whole sample; wherein, the winning node refers to the neuron with the output of 1 in the competition layer of the network when each sample is input into the SOM network, and the quantization error EqThe expression of (a) is as follows,
in the formula xiIs a set SuThe driving style characteristic variable vector of (1), each xiAll represent a driver sample; w is ajAnd the weight vector of the winning node corresponding to each driver sample is obtained.
The topological error is the ratio of the distances from all samples to the winning node and the second matching node and is used for measuring the protection condition of the topological structure; the second matching node refers to the neuron closest to the input sample in the competition layer of the SOM network except the winning node, and the topological error EtThe expression of (a) is as follows,
wherein,
in the formula w1,…,wtAll weight vectors of the SOM network, t is the number of neurons of a competition layer of the SOM network, and wkFor each driver sample the weight vector, w, of the second matched nodelFor SOM networks other than wjAnd wkAny other weight vector, j, k, l, is wj,wk,wlThe number of the corresponding contention layer neuron.
The neurons of the SOM competition layer in this embodiment are arranged in a rectangular topological structure, and when the number of the neurons of the SOM network competition layer increases from 2 × 2 to 8 × 8, the change curves of the quantization error and the topological error of the SOM network are shown in fig. 1. As can be seen in FIG. 1, the quantization error decreases with increasing number of neurons in the competition layer; while the topology error tends to increase with the number of neurons in the competition layer. Each neuron in the competitive layer of the SOM network represents a class and, in order for the clustering to be meaningful, there are on average at least 2 driver samples in each class. The number of driver samples in this embodiment is 66, so the number of neurons in the competition layer needs to be selected from 2 × 2 to 5 × 5. In addition, in order to prevent one of the quantization error and the topology error of the SOM network in fig. 1 from being too large, the number of neurons in the competition layer of the SOM network can be selected from only 3 × 3 or 4 × 4, and either of the two is feasible. The SOM network competition layer selected in this embodiment is 16 neurons arranged according to 4 × 4.
Driver sample set S after normalization using the SOM network described aboveuAfter 200 batches of training, the SOM network compares the driver sample set SuThe preliminary clustering result of (2) is shown in fig. 2. The abscissa of fig. 2 represents a driver sample set SuThe serial numbers of the driver samples are integers from 1 to 66; the ordinate represents the category to which each driver sample of the SOM network output belongs. It can be seen that the SOM network integrates a sample set S of driversuThe 66 driver samples in (1) are divided into 16 categories, and the number of the categories is the same as the set number of neurons in the competition layer of the SOM network. In addition, the weight vectors w of 16 competition layer neurons of the SOM network after the clustering is finished1,…,wtI.e. 16 cluster center vectors of the SOM network.
The driving style of the driver is defined as aggressive, neutral and conservative, so the sample set SuThe 66 driver samples in (1) were clustered into three classes. Therefore, the primary clustering result of the SOM network is subjected to secondary clustering by using the K-means algorithm, and the number of clustering centers of the K-means algorithm is set to be 3. The input of the K-means algorithm is a preliminary clustering result of the SOM network, namely a clustering center vector set S of the SOM networkukThe mathematical expression of the set is as follows,
wherein, wiThe cluster center vectors of the SOM network are represented, t represents the number of the SOM cluster center vectors, and the number of the SOM cluster center vectors is 16 in the embodiment.
Using K-means algorithm to set SukThe final result after secondary clustering is shown in fig. 3. The abscissa of fig. 3 represents the number of each driver sample, numbered as an integer from 1 to 66; the ordinate represents the class to which each driver sample output by the K-means algorithm belongs. As can be seen, after secondary clustering is carried out on the primary clustering result of the SOM network through the K-means algorithm, the driver sample set S is collecteduThe 66 driver samples in (1) are divided into 3 categories, i.e. the driving style of the driver is identified offline, the three categories represent the three driving styles of aggressive, neutral and conservative, respectively.
Step three, training the online classifier of the driving style
According to the driving style off-line recognition result in fig. 3, a label-free driver sample set S can be obtaineduEach driver sample in the system is given with a label of the driving style to which the driver sample belongs, and a labeled driver sample set S is obtainedl(i.e., a sample set of drivers with known driving styles), the mathematical expression for this set is as follows,
wherein y isiRepresents the driving style class to which the driver sample numbered i belongs, and yiE {1,2,3}, 1 represents conservative driving style, 2 represents neutral driving style, and 3 represents aggressive driving style.
In order to test the trained driving style on-line classifier, a sample set SlIs divided into a training set SltAnd test set SlvTwo parts. Considering that the number of samples in this embodiment is not large, in order to obtain a better training effect, only the sample set S is selectedlRandomly selecting 6 driver samples as a test set SlvAnd the other 60Driver sample as training set Slt。
Using a probabilistic neural network as an on-line classifier of driving style and using a training set SltAnd training the driving style online classifier. After training is completed, test set SlvInputting the driving style to a trained online driving style classifier, testing the classification effect of the online driving style classifier, and outputting a result representing the prediction of the driving style of each driver sample by the classifier. The results of the online driving style classifier are shown in fig. 4. The needle-shaped graph in the figure represents the true driving style of each driver sample, i.e. y corresponding to each driver sampleiThe scattered points of the star represent the result output by the online driving style classifier. As can be seen, the spicular map completely coincides with the star scatter point, indicating that the driving style online classifier is paired with the test set SlvThe driving style class predictions of the 6 driver samples all yielded correct results.
Claims (2)
1. A driving style identification and classification method for an automobile adaptive cruise system is characterized by comprising the following steps:
step one, car following experiment data acquisition
Firstly, randomly recruiting more than 50 drivers, respectively driving the same test vehicle to drive along with the same front vehicle on the same road section under the same vehicle following experiment working condition by each driver, wherein each driver can drive along with the front vehicle according to own style and habit, and acquiring four data of the speed of the test vehicle, the acceleration of the test vehicle, the distance between the test vehicle and the front vehicle and the relative speed of the test vehicle and the front vehicle of each driver in the vehicle following experiment;
the experimental process designed for the driving of the front vehicle is as follows:
(1) accelerating the front vehicle to 30km/h from the standstill, and maintaining the speed for 15 s;
(2) accelerating the front vehicle to 50km/h and maintaining the speed for 15 s;
(3) accelerating the front vehicle to 70km/h, and maintaining the speed for 15 s;
(4) the front vehicle decelerates to 50km/h and keeps the speed for 15 s;
(5) the front vehicle decelerates to 30km/h, and the speed is maintained for 15 s;
(6) the front vehicle decelerates to be static;
in the following experiment, each driver needs to perform twice in the positive direction and the negative direction of the same experimental road section, and the speed of the experimental vehicle, the acceleration of the experimental vehicle, the distance between the experimental vehicle and the front vehicle and the relative speed of the experimental vehicle and the front vehicle in the positive direction and the negative direction are collected as experimental data;
step two, off-line recognition of the driving style of the driver
Taking a following vehicle distance minimum value, a following vehicle distance average value, an impact maximum value, an impact minimum value, a relatively far vehicle speed maximum value, a relatively close vehicle speed maximum value, a positive impact average value, a negative impact average value, a relatively far vehicle speed average value and a relatively close vehicle speed average value as characteristic variables for representing the driving style of drivers, calculating the driving style characteristic variables of the drivers according to the data of the drivers obtained in a following vehicle experiment, combining the driving style characteristic variables of each driver into a driving style characteristic variable vector, and arranging the driving style characteristic variable vectors of all the drivers into a label-free driver sample set to be identified;
determining the number of neurons of a competition layer of the SOM, and performing preliminary clustering on the unlabeled driver sample set by using the SOM to obtain the category of each driver; performing secondary clustering on the primary clustering result obtained by the SOM network by using a K-means algorithm to obtain a final off-line recognition result of the driving style of each driver in a following experiment;
the method for determining the number of neurons in the competition layer of the SOM network specifically comprises the following steps: calculating the numerical values of the quantization error and the topological error of the SOM network with different competition layer neuron numbers under the condition of data input aiming at the data acquired in the following experiment, and determining the competition layer neuron number of the SOM network by balancing the quantization error, the topological error and the limit of the number of the driver samples obtained in the experiment;
the method specifically comprises the following steps of performing off-line identification on the driving style of a driver:
A. car following experiment data preprocessing
The driving style characteristic variables and the calculation formula are selected as follows,
(1) minimum distance d when following vehiclehmin: the following time interval is the time required for the head of the test vehicle to travel to the tail of the front vehicle at the current speed, and the following time interval is constant to be a positive value;
wherein d issIs the vehicle distance, i.e. the distance between the tail of the preceding vehicle and the head of the experimental vehicle, vfFor the running speed of the experimental vehicle, n is the number of data points of each driver collected in the following experiment, and n is Te/Ts，TeFor the duration of the car-following experiment, TsThe operation period of the experimental data acquisition equipment is set;
(2) distance mean value when following vehicle
(3) Maximum value of impact JKmax: the impact degree is the derivative of the acceleration of the experimental vehicle, and the maximum value of the impact degree is always a positive value;
wherein a is the acceleration of the experimental vehicle, and t is time;
(4) minimum value of jerk JKmin: the minimum value of the impact degree is always a negative value;
(5) relatively far away from the maximum value of vehicle speed Δ vamax: the relative far-away speed is the relative speed when the test vehicle is far away from the front vehicle, namely the relative speed of the two vehicles when the speed of the test vehicle is less than that of the front vehicle, and the relative far-away speed value is constant to a positive value;
Δvamax＝max{(vp-vf)i,i＝1,2,...,n}
wherein v ispThe running speed v of the preceding vehiclefThe running speed of the test vehicle is obtained;
(6) relatively close to the maximum value of vehicle speed Δ vcmax: the relative approaching speed is the relative speed when the test vehicle approaches the front vehicle, namely the relative speed of the two vehicles when the speed of the test vehicle is greater than that of the front vehicle, and the relative departing speed value is constant to be a negative value;
Δvcmax＝min{(vp-vf)i,i＝1,2,...,n}
(7) average value of positive impact
Positive impact JKpThe value of the impact degree is larger than zero; ,
wherein n iszCounting the number of data points with the impact degree larger than zero in the following experiment data;
(8) average value of negative impact
Negative impact JKnRepresents a value when the impact is less than zero;
wherein n isfCounting the number of data points with the impact degree smaller than zero in the following experimental data;
(9) relatively far from the average value of vehicle speed
Wherein n isaFor the number of data points when the speed of the experimental vehicle is less than that of the front vehicle in the following experimental data, delta vaIs relatively far away from the vehicle speed;
(10) relatively close to the average value of vehicle speed
Wherein n iscThe number of data points, delta v, of the following experimental data when the speed of the experimental vehicle is greater than that of the front vehiclecIs relatively close to the vehicle speed;
calculating ten driving style characteristic variables of each driver according to the calculation formula and four data recorded in the following experiment collected in the step one, and combining the driving style characteristic variables of each driver into a driving style characteristic variable vector, wherein each driving style characteristic variable vector represents a driver sample; combining all the driver samples to obtain a label-free driver sample set SuThe mathematical expression of the set is as follows,
wherein x isiRepresenting a group of 10 driving style characteristic variablesForming a driving style characteristic variable vector, namely a driver sample, wherein N represents the number of the driver samples, namely the number of drivers participating in an experiment, and T represents mathematical operation, namely transposing the matrix;
B. offline recognition of driver's driving style
Using SOM network for unlabeled driver sample set SuPerforming primary clustering to obtain a label-free driver sample set SuAll driver samples in the system are divided into a plurality of categories, and the number of the categories is the same as the number of neurons of the set SOM network competition layer;
setting the number of clustering centers of a K-means algorithm, wherein the number represents a plurality of driving styles of drivers, carrying out secondary clustering on a primary clustering result obtained by the SOM network by using the K-means algorithm, and carrying out a label-free driver sample set SuAll the driver samples are divided into the driving style categories with corresponding quantity, and the label-free driver sample set S is realizeduOff-line identification of the driver's driving style in (1);
step three, training the driving style on-line classifier
Forming a labeled driver sample set according to the off-line recognition result of the driving style of each driver in the following experiment obtained in the step two; the method comprises the steps of using a probabilistic neural network method as a driver driving style online classifier, training the classifier by using a formed labeled driver sample set to obtain a trained driver driving style online classifier, embedding the trained driver driving style online classifier into an automobile ACC system, providing driving style class information of a current driver for the automobile ACC system, and enabling the automobile ACC system to be switched to different modes according to the driving style class of the driver;
wherein train the online classifier of the driving style, include specifically:
according to the driving style off-line recognition result obtained in the step two, a label-free driver sample set SuEach driver sample in the system is given with a label of the driving style to which the driver sample belongs, and a labeled driver sample set S is obtainedlThe mathematical expression of the set is as follows,
wherein y isiRepresenting the driving style class to which the sample of the driver with the number i belongs;
will gather SlDivided into training set SltAnd test set SlvTwo parts, namely, using a probabilistic neural network method as a driving style online classifier and using a training set SltTraining a driving style online classifier; after training is completed, test set SlvAnd inputting the driving style to the trained online driving style classifier, testing the classification effect of the online driving style classifier, and outputting a result representing the prediction of the online driving style classifier on the driving style of each driver sample by the online driving style classifier.
2. The driving style identification and classification method for the automobile adaptive cruise system according to claim 1, wherein the following experiment in the step one specifically comprises:
firstly, randomly recruiting more than 50 drivers;
secondly, selecting following experiment working conditions for the automobile self-adaptive cruise system, and selecting a dry and gentle road section with good road surface condition as an experiment road section;
the following experiment is completed by two vehicles together, one is a front pilot vehicle, which is called a front vehicle for short, and the other is an experiment following vehicle, which is called an experiment vehicle for short; during the experiment, all the drivers are respectively driven to drive the same experimental vehicle to run along with the same front vehicle on the same experimental road section, wherein the front vehicle is driven according to the designed experimental process, the drivers of the experimental vehicles run along with the front vehicle according to own style habits, and finally four kinds of data including the speed of the experimental vehicle, the acceleration of the experimental vehicle, the distance between the experimental vehicle and the front vehicle and the relative speed of the experimental vehicle and the front vehicle of each driver in the vehicle following experiment are collected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010454845.3A CN111547064B (en) | 2020-05-26 | 2020-05-26 | Driving style recognition and classification method for automobile adaptive cruise system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010454845.3A CN111547064B (en) | 2020-05-26 | 2020-05-26 | Driving style recognition and classification method for automobile adaptive cruise system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111547064A CN111547064A (en) | 2020-08-18 |
| CN111547064B true CN111547064B (en) | 2022-07-12 |
Family
ID=72000575
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010454845.3A Active CN111547064B (en) | 2020-05-26 | 2020-05-26 | Driving style recognition and classification method for automobile adaptive cruise system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111547064B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112248882B (en) * | 2020-10-20 | 2022-08-26 | 东风汽车集团有限公司 | Method for predicting driving range of pure electric vehicle |
| CN112560354B (en) * | 2020-12-26 | 2024-02-23 | 浙江天行健智能科技有限公司 | Car following behavior modeling method based on Gaussian process regression |
| CN112622920A (en) * | 2020-12-28 | 2021-04-09 | 华北科技学院 | Auxiliary driving system |
| CN112677983B (en) * | 2021-01-07 | 2022-04-12 | 浙江大学 | System for recognizing driving style of driver |
| CN113044046B (en) * | 2021-04-09 | 2022-09-06 | 联合汽车电子有限公司 | Driver style recognition method, system, electronic device and readable storage medium |
| CN113408431B (en) * | 2021-06-22 | 2023-03-10 | 青岛小鸟看看科技有限公司 | Intelligent driving evaluation training method and system based on eyeball tracking |
| CN113657752B (en) * | 2021-08-17 | 2022-05-31 | 西安航空学院 | Driving risk quantitative evaluation method for tunnels with different lengths based on driving behavior indexes |
| CN113947259B (en) * | 2021-11-09 | 2024-05-31 | 吉林大学 | GRU neural network-based method for predicting speeds of drivers in different styles |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101633359A (en) * | 2008-07-24 | 2010-01-27 | 通用汽车环球科技运作公司 | Adaptive vehicle control system with driving style recognition |
| CN101633358A (en) * | 2008-07-24 | 2010-01-27 | 通用汽车环球科技运作公司 | Adaptive vehicle control system with integrated driving style recognition |
| CN106570560A (en) * | 2016-11-02 | 2017-04-19 | 温州大学 | Driving style quantitative evaluation method based on standardized driving behaviors and phase space reconstruction |
| CN107016193A (en) * | 2017-04-06 | 2017-08-04 | 中国科学院自动化研究所 | Driver is with the expectation following distance computational methods in car behavioural analysis |
| CN108860148A (en) * | 2018-06-13 | 2018-11-23 | 吉林大学 | Self-adapting cruise control method based on driver's follow the bus characteristic Safety distance model |
| CN108944939A (en) * | 2017-05-17 | 2018-12-07 | 现代自动车株式会社 | For providing the method and system of driving directions |
| CN109367541A (en) * | 2018-10-15 | 2019-02-22 | 吉林大学 | A kind of intelligent vehicle class people's lane change decision-making technique based on driver behavior pattern |
| CN109624986A (en) * | 2019-03-01 | 2019-04-16 | 吉林大学 | A kind of the study cruise control system and method for the driving style based on pattern switching |
| CN109927725A (en) * | 2019-01-28 | 2019-06-25 | 吉林大学 | A kind of self-adaption cruise system and implementation method with driving style learning ability |
| CN110378483A (en) * | 2018-04-12 | 2019-10-25 | 百度（美国）有限责任公司 | The system and method for training machine learning model being deployed on analog platform |
| CN110858312A (en) * | 2018-08-23 | 2020-03-03 | 重庆大学 | Driver driving style classification method based on fuzzy C-means clustering algorithm |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9361409B2 (en) * | 2013-01-10 | 2016-06-07 | International Business Machines Corporation | Automatic driver modeling for integration of human-controlled vehicles into an autonomous vehicle network |
| CN107808027B (en) * | 2017-09-14 | 2020-11-24 | 上海理工大学 | Self-adaptive car following method based on improved model predictive control |
-
2020
- 2020-05-26 CN CN202010454845.3A patent/CN111547064B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101633359A (en) * | 2008-07-24 | 2010-01-27 | 通用汽车环球科技运作公司 | Adaptive vehicle control system with driving style recognition |
| CN101633358A (en) * | 2008-07-24 | 2010-01-27 | 通用汽车环球科技运作公司 | Adaptive vehicle control system with integrated driving style recognition |
| CN106570560A (en) * | 2016-11-02 | 2017-04-19 | 温州大学 | Driving style quantitative evaluation method based on standardized driving behaviors and phase space reconstruction |
| CN107016193A (en) * | 2017-04-06 | 2017-08-04 | 中国科学院自动化研究所 | Driver is with the expectation following distance computational methods in car behavioural analysis |
| CN108944939A (en) * | 2017-05-17 | 2018-12-07 | 现代自动车株式会社 | For providing the method and system of driving directions |
| CN110378483A (en) * | 2018-04-12 | 2019-10-25 | 百度（美国）有限责任公司 | The system and method for training machine learning model being deployed on analog platform |
| CN108860148A (en) * | 2018-06-13 | 2018-11-23 | 吉林大学 | Self-adapting cruise control method based on driver's follow the bus characteristic Safety distance model |
| CN110858312A (en) * | 2018-08-23 | 2020-03-03 | 重庆大学 | Driver driving style classification method based on fuzzy C-means clustering algorithm |
| CN109367541A (en) * | 2018-10-15 | 2019-02-22 | 吉林大学 | A kind of intelligent vehicle class people's lane change decision-making technique based on driver behavior pattern |
| CN109927725A (en) * | 2019-01-28 | 2019-06-25 | 吉林大学 | A kind of self-adaption cruise system and implementation method with driving style learning ability |
| CN109624986A (en) * | 2019-03-01 | 2019-04-16 | 吉林大学 | A kind of the study cruise control system and method for the driving style based on pattern switching |
Non-Patent Citations (1)
| Title |
|---|
| 《基于K-均值聚类算法的行驶工况构建方法》;秦大同等;《吉林大学学报(工学版)》;20160331;383-389 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111547064A (en) | 2020-08-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111547064B (en) | Driving style recognition and classification method for automobile adaptive cruise system | |
| CN111038485B (en) | Hybrid electric vehicle control method and system based on driving style recognition | |
| CN106971194B (en) | Driving intention recognition method based on improved HMM and SVM double-layer algorithm | |
| CN110949398B (en) | Method for detecting abnormal driving behavior of first-vehicle drivers in vehicle formation driving | |
| CN107492251B (en) | Driver identity recognition and driving state monitoring method based on machine learning and deep learning | |
| CN105741637B (en) | Four-wheel hub motor electric car automated steering control method | |
| CN110626339B (en) | Emergency obstacle avoidance automatic driving assistance torque calculation method based on driver intention | |
| CN112046489B (en) | Driving style identification algorithm based on factor analysis and machine learning | |
| CN110386144B (en) | GHMM/GGAP-RBF hybrid model for identifying driver braking intention and identification method | |
| CN112721949B (en) | Method for evaluating longitudinal driving personification degree of automatic driving vehicle | |
| CN112116100B (en) | Game theory decision method considering driver type | |
| CN108482481A (en) | Four motorized wheels and the four-wheel steering control method for turning to electric vehicle | |
| CN109255094A (en) | Commercial truck quality estimation method based on SVR_Adaboost innovatory algorithm | |
| CN111332362A (en) | Intelligent steer-by-wire control method integrating individual character of driver | |
| KR20180037468A (en) | Method for deciding a road surface using vehicle data | |
| CN111340074B (en) | Driver braking intention identification method based on improved RBF neural network | |
| CN116373881A (en) | In-transit driving style identification method based on driving behavior semantic understanding | |
| CN115186594A (en) | Energy-saving speed optimization method under influence of man-vehicle-road coupling | |
| Zhu et al. | A comfort quantification method based on semi-supervised learning for automated vehicle at lane change scenarios | |
| CN114148349A (en) | Vehicle personalized following control method based on generation countermeasure simulation learning | |
| EP3967565A1 (en) | Methods and apparatuses for estimating an environmental condition | |
| Wang et al. | A data-driven estimation of driving style using deep clustering | |
| CN118004220A (en) | Intelligent automobile personalized obstacle avoidance control method and system | |
| Anil et al. | Driver behavior analysis using K-means algorithm | |
| CN114639238B (en) | Urban expressway traffic state estimation method based on normalized spectral clustering algorithm |
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 |