CN112668779A - Preceding vehicle motion state prediction method based on self-adaptive Gaussian process - Google Patents
Preceding vehicle motion state prediction method based on self-adaptive Gaussian process Download PDFInfo
- Publication number
- CN112668779A CN112668779A CN202011581188.5A CN202011581188A CN112668779A CN 112668779 A CN112668779 A CN 112668779A CN 202011581188 A CN202011581188 A CN 202011581188A CN 112668779 A CN112668779 A CN 112668779A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- front vehicle
- motion state
- training
- gaussian
- 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.)
- Pending
Links
Images
Landscapes
- Traffic Control Systems (AREA)
- Control Of Driving Devices And Active Controlling Of Vehicle (AREA)
Abstract
The invention discloses a preceding vehicle motion state prediction method based on a self-adaptive Gaussian process. Then, real-time driving data are collected, the HMM is used for identifying the type of the front vehicle, corresponding initial values and boundary constraints are selected according to the identification result, the hyper-parameters are adaptively updated through the real-time driving data, an adaptive Gaussian process is built, and finally the motion state of the front vehicle is predicted through the adaptive Gaussian process. The method has the advantages that the motion characteristics of the front vehicle are considered from the vehicle perspective, the influence of the characteristics of a driver on the state of the front vehicle is effectively avoided, the state of the front vehicle can be effectively predicted through the AOGPR model, the motion state of the front vehicle can be predicted for a long time, the uncertainty of the motion state of the front vehicle can be predicted, the expansibility is high, and the method can adapt to different driving scenes.
Description
Technical Field
The invention relates to the technical field of vehicle state prediction, in particular to a preceding vehicle motion state prediction method based on a self-adaptive Gaussian process.
Background
The development of computer technology, sensing technology and automatic control technology is benefited, and the intelligent driving automobile is rapidly developed. The environmental perception serves as the 'eyes' of the automatic driving automobile, the traffic information of the road ahead is provided for the automobile, and a basis is provided for decision-planning-control of an intelligent driving system. In the actual driving environment, the following driving working condition is the most common, if the motion state of the front vehicle, such as the front vehicle speed, the relative distance between the two vehicles and the like, can be obtained in advance, the driving performance of the vehicle can be greatly improved, such as energy-saving control for reducing energy consumption, collision early warning control for improving driving safety, adaptive cruise control for improving comfort and the like. In the following vehicle working condition, the motion state of the vehicle is related to a plurality of factors such as traffic flow information, environmental information, driving style of a driver, vehicle type and the like, so that how to accurately estimate, judge and predict the motion state of the vehicle in front is a key technology for improving the longitudinal driving performance of the intelligent vehicle.
At present, some researchers at home and abroad have already carried out vehicle state prediction research, and the methods can be summarized into the following two types: one is a model-based prediction method, and the other is a data-driven prediction algorithm.
The prediction method based on the physical model assumes that the future state of the vehicle depends on the current motion state, and the vehicle state is predicted by combining a dynamic model or a kinematic model with Kalman Filtering (KF) and derivative algorithms thereof, such as Extended Kalman Filtering (EKF) and Unscented Kalman Filtering (UKF). In addition, the more complex the model is, the longer the calculation time is, and in the actual driving, it is difficult to accurately acquire the dynamic parameters of the front vehicle, which limits the application of the algorithm in the real-time embedded system.
The method is based on historical or online data of the vehicle, and utilizes advanced machine learning technology to estimate the vehicle state, and common algorithms comprise a BP neural network, a deep learning long short term memory network (LSTM) based on deep learning, a deep Bayesian network and the like. The algorithm has strong fitting capability and can predict the vehicle running state in a complex environment, but most models acquire model parameters through off-line training, so that the prediction error in an uncertain scene is increased, and in addition, due to the complex model structure and the need of optimizing too many parameters, the instantaneity is difficult to guarantee.
Therefore, a method for predicting the state of a preceding vehicle in various driving scenes during the following driving of the automatic driving vehicle is needed to improve the problems.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a preceding vehicle motion state prediction method based on a self-adaptive Gaussian process, which considers the motion characteristics of vehicles based on different types of preceding vehicles, effectively avoids the influence of the characteristics of drivers under different driving scenes on the preceding vehicle state prediction and can adapt to the preceding vehicle state prediction under different driving scenes.
The specific technical scheme is as follows:
in a first aspect, a method for predicting a preceding vehicle motion state based on an adaptive gaussian process is provided, which includes:
collecting natural driving data of a driver in different driving environments to construct a plurality of training sets, wherein all the training sets correspond to different front vehicle types one to one;
training the front vehicle recognition model through all training sets, and determining the hyperparametric initial values and the boundary constraints of Gaussian regression processes corresponding to different front vehicle types;
acquiring real-time driving data of a vehicle, and determining the type of the front vehicle through a trained front vehicle identification model;
and selecting corresponding initial values of the hyper-parameters and boundary constraints according to the recognition result, and performing adaptive optimization updating on the hyper-parameters through the real-time driving data to construct an adaptive Gaussian process to predict the motion state of the front vehicle.
With reference to the first aspect, in a first implementable manner of the first aspect, the constructing a plurality of training sets includes:
collecting natural driving data in a normal driving process;
extracting vehicle following condition data from the collected natural driving data according to preset conditions;
and classifying the extracted vehicle following condition data according to the type of the previous vehicle to obtain a plurality of training sets.
With reference to the first aspect, in a second implementable manner of the first aspect, the training the preceding vehicle recognition model through all training sets includes:
training the Gaussian mixture models corresponding to different vehicle types through corresponding training sets to obtain a plurality of trained Gaussian mixture models;
and constructing a hidden Markov model as the front vehicle identification model by using the trained mixed Gaussian models corresponding to different vehicle types and a Viterbi algorithm.
With reference to the second implementable manner of the first aspect, in a third implementable manner of the first aspect, the number of gaussians of the gaussian mixture model is determined by a bayesian criterion, and the parameter of the gaussian mixture model is determined by a maximum likelihood function.
With reference to the second implementable manner of the first aspect, in a fourth implementable manner of the first aspect, the determining a type of a leading vehicle by using the trained leading vehicle recognition model includes:
and inputting the real-time driving data into a hidden Markov model, and performing path backtracking through a Viterbi algorithm to determine the type of the preceding vehicle.
With reference to the first aspect, in a fifth implementation manner of the first aspect, the initial value and the boundary constraint of the hyper-parameter are obtained by performing discrete processing on sample data in a corresponding training set according to a preset time length.
With reference to the first aspect, in a sixth implementation manner of the first aspect, the following method is adopted to optimally update the hyper-parameter:
collecting real-time driving data within a certain time range to construct an optimized training set;
carrying out Gaussian process training by optimizing a training set, and solving Gaussian process parameters by using a maximum likelihood function;
and carrying out optimization solution based on a gradient optimization algorithm, and updating the hyper-parameter according to a solution result.
With reference to the first aspect, in a seventh implementation manner of the first aspect, the method further includes determining uncertainty of the prediction result by using a covariance matrix.
In a second aspect, a computer storage medium is provided, storing a computer program adapted to be loaded by a processor and to perform the prediction method described above.
Has the advantages that: under the multi-working condition, the motion characteristics of different types of vehicles in front are considered, the motion state of the front vehicle is considered from the aspect of vehicle types, the influence of the characteristics of a driver on the prediction of the state of the front vehicle is effectively avoided, the motion characteristics of the different types of vehicles are utilized to improve and optimize the Gaussian regression process, the improved model has self-adaptive learning capacity, the model parameters can be optimized and updated on line according to real-time driving data, the motion state of the front vehicle can be predicted for a long time, and the uncertainty of the motion state of the front vehicle can be predicted. The method has strong expansibility, can adapt to different driving scenes, and has good application prospect in future intelligent automobiles.
Drawings
In order to more clearly illustrate the embodiments of the present invention, the drawings, which are required to be used in the embodiments, will be briefly described below. In all the drawings, the elements or parts are not necessarily drawn to actual scale.
FIG. 1 is a schematic diagram of a prediction method according to an embodiment of the present invention;
FIG. 2 is a flow chart of a prediction method according to an embodiment of the present invention;
FIG. 3 is a flow chart of the construction of the training set of FIG. 2;
FIG. 4 is a flow diagram for constructing a hidden Markov (HMM);
FIG. 5 is a flow chart of adaptive Gaussian regression model (AOGPR) update parameters;
fig. 6 is a schematic diagram of an adaptive gaussian process.
Detailed Description
Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and therefore are only examples, and the protection scope of the present invention is not limited thereby.
Fig. 2 is a flowchart of a method for predicting a vehicle ahead motion state based on an adaptive gaussian process, where the method includes:
step 3, collecting real-time driving data of the vehicle, and determining the type of the front vehicle through a trained front vehicle identification model;
and 4, selecting corresponding initial values of the hyper-parameters and boundary constraints according to the recognition results, performing adaptive optimization updating on the hyper-parameters through the real-time driving data, and constructing an adaptive Gaussian process to predict the motion state of the front vehicle.
Specifically, the method comprises the following steps:
firstly, as shown in fig. 1, natural driving data of a driver in different driving environments can be collected through an existing real vehicle data collection platform, and the natural driving data is calibrated and classified to determine training sets and test sets corresponding to different front vehicle types. The driving environment comprises an expressway, an intercity express, a national road and the like, and the collected natural driving data comprises main vehicle state information, front vehicle information and environment information. The main vehicle information comprises main vehicle speed, acceleration, steering wheel turning angle, yaw velocity and the like, the preceding vehicle information comprises relative longitudinal distance, relative transverse distance, relative longitudinal vehicle speed and preceding vehicle type, and the environment information comprises lane number, road type, traffic signs and the like.
And then, training the front vehicle identification model through the sample data in all the training sets to obtain the front vehicle identification model for identifying the type of the front vehicle. Because the initial value of the hyper-parameter and the boundary constraint in the existing Gaussian regression process are generally generated randomly, more uncertain factors exist. Therefore, in order to reduce the uncertain factors, in this embodiment, the hyperparametric initial value and the boundary constraint of the gaussian regression process may be determined through the training set, and in this embodiment, the gaussian regression may be set as:
f(x):GPR(m(x),k(x,x′))；
where x is an input variable, m (x) is a mean function representing a vehicle state trend, which may be used to represent a desired predicted state; the covariance function k (x, x') represents the variance of the different inputs x themselves, and can be used to represent the uncertainty corresponding to the desired prediction state. For the sake of easy calculation of the value of m (x) to 0, the covariance function is expressed as:
wherein, deltafIs the standard deviation of the signal, deltalIs a characteristic length, σnTo observe the standard deviation of noise, InIs the kronecker function.
And then, acquiring real-time driving data of the driver, and inputting the real-time driving data into the trained front vehicle recognition model so as to determine the type of the front vehicle.
And then, selecting a corresponding super-parameter initial value according to the type of the front vehicle at the identification position, training a Gaussian regression process by acquiring real-time driving data within a certain time range, updating the super-parameter of the Gaussian regression process, constructing an adaptive Gaussian process, and predicting the motion state of the front vehicle by using the adaptive Gaussian process.
Therefore, the motion characteristics of different types of vehicles in front under various driving scenes are considered, the motion state of the vehicle in front is predicted from the dimension of the vehicle, and the influence of the characteristics of a driver on the prediction of the state of the vehicle in front is effectively avoided. And the optimized self-adaptive Gaussian regression process has self-adaptive learning capacity, and model parameters can be optimized and updated on line according to real-time driving data so as to predict the motion state of the front vehicle for a long time.
In this embodiment, preferably, as shown in fig. 1 and fig. 3, the constructing the plurality of training sets includes:
step 1-1, collecting natural driving data in a normal driving process;
step 1-2, extracting vehicle following condition data from collected natural driving data according to preset conditions;
and 1-3, classifying the extracted vehicle following condition data according to the type of the previous vehicle to obtain a plurality of training sets.
Specifically, the method comprises the following steps:
firstly, natural driving data of a driver in different driving environments can be acquired through an existing real vehicle data acquisition platform;
then, vehicle following condition data are extracted according to preset conditions, wherein the preset conditions comprise:
1. the following situation must include two vehicles, i.e., the host vehicle and the preceding vehicle, and the host vehicle and the preceding vehicle travel on the same lane.
2. If other vehicles are cut in/out between the main vehicle and the front vehicle, the vehicle following condition is finished;
3. the relative longitudinal distance between the main vehicle and the front vehicle is less than 120m and more than 5 m;
4. the speed of the main vehicle and the front vehicle is more than 20km/h and less than 120 km/h;
5. the car following duration needs to be more than 30 s;
6. if the condition of rapid acceleration/rapid deceleration occurs, the condition can not be regarded as the following condition.
And finally, classifying the extracted following vehicle condition data according to the types of the previous vehicles to obtain data sets corresponding to different types of the previous vehicles. In this embodiment, the preceding vehicle type is set to 3 types, where the 3 types are Car, Bus, and Truck, respectively, the extracted following vehicle condition data are classified according to the 3 preceding vehicle types, and the obtained corresponding data set is calibrated, for example: Car-Car (C-C), indicates that the host Car is a passenger Car, the following front Car is a passenger Car, and the data set is marked as 1. Car-Bus (C-B) indicates that the host vehicle is a passenger vehicle, the following front vehicle is a Bus or a Bus, and the data set is labeled 2. Car-Truck (C-T), indicates that the host Car is a passenger Car, the following front Car is a Truck, and the data set is calibrated to be 3. And respectively dividing the calibrated data into a training set and a test set according to the ratio of 8:2, wherein the training set is used for model training, and the test set is used for model testing.
In this embodiment, preferably, the training the preceding vehicle recognition model through all training sets includes:
training the Gaussian mixture models corresponding to different vehicle types through corresponding training sets to obtain a plurality of trained Gaussian mixture models;
and constructing a hidden Markov model as the front vehicle identification model by using the trained mixed Gaussian models corresponding to different vehicle types and a Viterbi algorithm.
Specifically, a hidden markov model-based front vehicle type recognition model is built with the front vehicle type as a hidden variable, the main vehicle information and the front vehicle information as an observation variable, and as shown in fig. 1 and 4, the hidden markov model-based front vehicle type recognition model specifically comprises the following steps:
step 2-1, selecting the speed of the main vehicle, relative longitudinal distance and longitudinal speed of the front vehicle as an observation variable, wherein the expression is as follows:
O＝[vego,Δd,vp]wherein v ispLongitudinal speed of the front vehicle, Δ d relative longitudinal distance, vegoThe host vehicle speed.
Step 2-2, setting an initial state of the hidden Markov model, taking three vehicle types as hidden variables, respectively using 1, 2 and 3 to represent that the front vehicle types are Car, Bus and Truck, and the probability of the initial state of the three types is as follows:
π＝[p(St)＝1,p(St)＝2,p(St)＝3]
wherein, p (S)tI) initial probability when state is i, FiThe sample data size for state i.
Step 2-3, obtaining the transition probability matrix of the hidden Markov model through the training set
The state transition probability matrix represents the transition probabilities between states, as follows:
wherein, ai,j∈R3×3，Ti,jIndicating the number of transitions from state i to state j.
Step 2-4, setting an emission probability model of the hidden markov model, wherein in this embodiment, the emission probability may adopt a trained gaussian mixture model as the emission probability model, and the expression is as follows:
wherein, B (O)t) Is a mixed Gaussian model of emission probability density, omegakAs a single Gaussian weight coefficient, Bk(Ot|μk,Σk) As a single Gaussian probability density function, muk,ΣkRespectively being Gaussian functionsMean and covariance matrix, d is the observation variable dimension.
In this embodiment, preferably, the number of gaussians of the gaussian mixture model may be determined by a bayesian criterion, and the parameter of the gaussian mixture model may be determined by a maximum likelihood function. Wherein the Bayesian criterion is expressed as follows:
the likelihood function is expressed as follows:
in this embodiment, preferably, the determining the type of the leading vehicle through the trained leading vehicle recognition model includes:
and inputting the real-time driving data into a hidden Markov model, and performing path backtracking through a Viterbi algorithm to determine the type of the preceding vehicle.
Specifically, after real-time driving data of the vehicle is acquired through the real vehicle data platform, the real-time driving data can be input into a trained hidden Markov model, and the type of the vehicle ahead can be determined by tracing back the path through a Viterbi algorithm.
In this embodiment, preferably, the initial value and the boundary constraint of the hyper-parameter are obtained by performing discrete processing on sample data in the corresponding training set according to a preset time length. Specifically, the training set corresponding to each vehicle type may be subjected to discrete processing according to a certain time length to obtain an initial value of a hyper-parameter of the gaussian regression model corresponding to each vehicle type and a boundary constraint, where the boundary constraint is as follows:
lp≤ψ≤up
wherein psi is a hyper-parameter, psi ═ σn,δf,δl]Wherein lp is the lower constraint bound and up is the lower constraint bound.
In this embodiment, preferably, as shown in fig. 5, the following method is adopted to perform adaptive optimization updating on the hyper-parameters:
3-1, collecting driving data in real-time driving data for a certain time to construct an optimized training set;
3-2, performing Gaussian process training by using the initial values and the boundary constraints of the hyper-parameters through an optimized training set, and solving the parameters of the Gaussian process by using a maximum likelihood function;
and 3-3, carrying out optimization solution based on a gradient optimization algorithm, and updating the hyper-parameter according to a solution result.
Specifically, as shown in fig. 1 and fig. 6, assuming that the motion state of the vehicle at the previous time has a strong correlation with the future state, and the motion state at the adjacent time does not change abruptly, the hyper-parameter relationship of different prediction steps can be expressed by the following formula:
wherein psi1ini＝ψtypeInitial value of hyper-parameter, psi, of each type of Gaussian regression processk-NcIs the over-parameter value, ω, of the k-Nth prediction stepk-NcFor corresponding weight coefficients,. psikiniAnd (5) the initial value of the hyperparameter predicted in the k step. As can be seen from the above assumption conditions, the closer the separation time is, the larger the weight coefficient is, and the farther the separation time is, the smaller the weight coefficient is. According to the log-likelihood function and the partial derivatives of the log-likelihood function to each unknown parameter, and according to the initial value of the hyper-parameter and the boundary constraint determined through the training set, the optimized hyper-parameter value can be obtained by utilizing a gradient optimization algorithm, and the method specifically comprises the following steps:
first, the real-time driving data of the first 5 seconds (τ ═ 5) is collected as an optimized training set to learn the gaussian regression process, where the training set is: t isGP(X,Y)＝{xi,yiI ═ 1: N }, N is the data length, and the unknown parameter is Ψ ═ σn,δf,δl]The log-edge likelihood function is:
wherein the content of the first and second substances,
then, the Gaussian process parameters are obtained by utilizing the maximum likelihood function
And finally, carrying out optimization solution on the hyper-parameters based on a gradient optimization algorithm, and solving out the optimized hyper-parameter values, wherein the partial derivative of each unknown quantity is as follows:
and predicting the motion state of the front vehicle in s seconds in the future by adopting the following prediction expression through the updated super parameters. In this embodiment, the prediction of the motion state of the preceding vehicle mainly includes the relative longitudinal distance of the preceding vehicle and the speed of the preceding vehicle, and the prediction expression is as follows:
wherein the content of the first and second substances,
in order to predict the mean value of the mean,
is a covariance matrix.
In this embodiment, it is preferable that the method further includes obtaining uncertainty of the prediction result by using a covariance matrix. Specifically, after the prediction result is obtained, the uncertainty of the prediction result can be obtained through the covariance matrix, and specifically, the following calculation formula can be adopted:
a computer storage medium storing a computer program adapted to be loaded by a processor and to perform the prediction method described above.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; such modifications and substitutions do not depart from the spirit and scope of the present invention, and they should be construed as being included in the following claims and description.
Claims (9)
1. A preceding vehicle motion state prediction method based on an adaptive Gaussian process is characterized by comprising the following steps:
collecting natural driving data of a driver in different driving environments to construct a plurality of training sets, wherein all the training sets correspond to different front vehicle types one to one;
training the front vehicle recognition model through all training sets, and determining the hyperparametric initial values and the boundary constraints of Gaussian regression processes corresponding to different front vehicle types;
acquiring real-time driving data of a vehicle, and determining the type of the front vehicle through a trained front vehicle identification model;
and selecting corresponding initial values of the hyper-parameters and boundary constraints according to the recognition result, and performing adaptive optimization updating on the hyper-parameters through the real-time driving data to construct an adaptive Gaussian process to predict the motion state of the front vehicle.
2. The adaptive Gaussian process-based preceding vehicle motion state prediction method according to claim 1, wherein the constructing a plurality of training sets comprises:
collecting natural driving data in a normal driving process;
extracting vehicle following condition data from the collected natural driving data according to preset conditions;
and classifying the extracted vehicle following condition data according to the type of the previous vehicle to obtain a plurality of training sets.
3. The method according to claim 1, wherein the training of the preceding vehicle recognition model through all training sets comprises:
training the Gaussian mixture models corresponding to different vehicle types through corresponding training sets to obtain a plurality of trained Gaussian mixture models;
and constructing a hidden Markov model as the front vehicle identification model by using the trained mixed Gaussian models corresponding to different vehicle types and a Viterbi algorithm.
4. The adaptive Gaussian process-based preceding vehicle motion state prediction method according to claim 3, characterized in that the Gaussian number of the Gaussian mixture model is determined by Bayesian criterion, and the parameters of the Gaussian mixture model are determined by maximum likelihood function.
5. The method according to claim 3, wherein the determining the type of the leading vehicle through the trained leading vehicle recognition model comprises:
and inputting the real-time driving data into a hidden Markov model, and performing path backtracking through a Viterbi algorithm to determine the type of the preceding vehicle.
6. The method according to claim 1, wherein the hyper-parameter initial value and the boundary constraint are obtained by performing discrete processing on sample data in a corresponding training set according to a preset time length.
7. The adaptive Gaussian process-based preceding vehicle motion state prediction method according to claim 1, characterized in that the hyper-parameters are optimally updated by adopting the following method:
collecting real-time driving data of a certain time to construct an optimized training set;
carrying out Gaussian process training by optimizing a training set, and solving Gaussian process parameters by using a maximum likelihood function;
and carrying out optimization solution based on a gradient optimization algorithm, and updating the hyper-parameter according to a solution result.
8. The adaptive Gaussian process-based preceding vehicle motion state prediction method according to claim 1, further comprising determining uncertainty of prediction results using a covariance matrix.
9. A computer storage medium, characterized in that a computer program is stored, which computer program is adapted to be loaded by a processor and to carry out the prediction method according to any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011581188.5A CN112668779A (en) | 2020-12-28 | 2020-12-28 | Preceding vehicle motion state prediction method based on self-adaptive Gaussian process |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011581188.5A CN112668779A (en) | 2020-12-28 | 2020-12-28 | Preceding vehicle motion state prediction method based on self-adaptive Gaussian process |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112668779A true CN112668779A (en) | 2021-04-16 |
Family
ID=75410833
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011581188.5A Pending CN112668779A (en) | 2020-12-28 | 2020-12-28 | Preceding vehicle motion state prediction method based on self-adaptive Gaussian process |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112668779A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113299158A (en) * | 2021-05-10 | 2021-08-24 | 易显智能科技有限责任公司 | Driving stability teaching method and system |
| CN113428141A (en) * | 2021-07-15 | 2021-09-24 | 东风汽车集团股份有限公司 | Intelligent detection method and system for timely response of emergency cut-in of front vehicle |
| CN113561976A (en) * | 2021-08-19 | 2021-10-29 | 湖南大学无锡智能控制研究院 | Vehicle energy-saving prediction adaptive cruise control method and device based on feedback optimization |
| CN113928313A (en) * | 2021-10-08 | 2022-01-14 | 南京航空航天大学 | Intelligent vehicle following control method and system suitable for heterogeneous traffic |
| CN114162132A (en) * | 2021-12-07 | 2022-03-11 | 吉林大学 | Driving mode identification method based on subjective and objective evaluation |
| CN117290801A (en) * | 2023-11-27 | 2023-12-26 | 浪潮软件科技有限公司 | Sequential monitoring index anomaly detection method based on Gaussian process regression |
-
2020
- 2020-12-28 CN CN202011581188.5A patent/CN112668779A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113299158A (en) * | 2021-05-10 | 2021-08-24 | 易显智能科技有限责任公司 | Driving stability teaching method and system |
| CN113428141A (en) * | 2021-07-15 | 2021-09-24 | 东风汽车集团股份有限公司 | Intelligent detection method and system for timely response of emergency cut-in of front vehicle |
| CN113561976A (en) * | 2021-08-19 | 2021-10-29 | 湖南大学无锡智能控制研究院 | Vehicle energy-saving prediction adaptive cruise control method and device based on feedback optimization |
| CN113561976B (en) * | 2021-08-19 | 2022-04-19 | 湖南大学无锡智能控制研究院 | Vehicle energy-saving prediction adaptive cruise control method and device based on feedback optimization |
| CN113928313A (en) * | 2021-10-08 | 2022-01-14 | 南京航空航天大学 | Intelligent vehicle following control method and system suitable for heterogeneous traffic |
| CN114162132A (en) * | 2021-12-07 | 2022-03-11 | 吉林大学 | Driving mode identification method based on subjective and objective evaluation |
| CN114162132B (en) * | 2021-12-07 | 2023-11-21 | 吉林大学 | Driving mode identification method based on subjective and objective evaluation |
| CN117290801A (en) * | 2023-11-27 | 2023-12-26 | 浪潮软件科技有限公司 | Sequential monitoring index anomaly detection method based on Gaussian process regression |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112668779A (en) | Preceding vehicle motion state prediction method based on self-adaptive Gaussian process | |
| CN110949398B (en) | Method for detecting abnormal driving behavior of first-vehicle drivers in vehicle formation driving | |
| CN107169567B (en) | Method and device for generating decision network model for automatic vehicle driving | |
| CN107229973B (en) | Method and device for generating strategy network model for automatic vehicle driving | |
| CN111079590B (en) | Peripheral vehicle behavior pre-judging method of unmanned vehicle | |
| Berndt et al. | Driver intention inference with vehicle onboard sensors | |
| CN112242059B (en) | Intelligent decision-making method for unmanned vehicle based on motivation and risk assessment | |
| CN112793576B (en) | Lane change decision method and system based on rule and machine learning fusion | |
| CN112677982B (en) | Vehicle longitudinal speed planning method based on driver characteristics | |
| CN111783943B (en) | LSTM neural network-based driver braking strength prediction method | |
| CN113722835B (en) | Personification random lane change driving behavior modeling method | |
| CN110490275A (en) | A kind of driving behavior prediction technique based on transfer learning | |
| US11560146B2 (en) | Interpreting data of reinforcement learning agent controller | |
| CN111907523A (en) | Vehicle following optimization control method based on fuzzy reasoning | |
| CN115186594A (en) | Energy-saving speed optimization method under influence of man-vehicle-road coupling | |
| Abdelrahman et al. | Driver behavior classification in crash and near-crash events using 100-CAR naturalistic data set | |
| CN115279643A (en) | On-board active learning method and apparatus for training a perception network of an autonomous vehicle | |
| CN113761715B (en) | Method for establishing personalized vehicle following model based on Gaussian mixture and hidden Markov | |
| CN114148349B (en) | Vehicle personalized following control method based on generation of countermeasure imitation study | |
| CN113954844A (en) | Intelligent automobile man-machine driving mode switching system | |
| KR102570295B1 (en) | Vehicle and control method thereof | |
| Zan et al. | Lane Change Intention Recognition for Human Driving Vehicles under Moving Bottleneck on Freeway | |
| CN117932234B (en) | Data processing method and system for manufacturing brake calibration table | |
| CN115195791B (en) | Unmanned driving speed control method and device based on big data | |
| Hou | Study on Anthropomorphic Lane Changing Decision Making for Smart Trucks Based on Driving Behavior |
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 |