CN111717217A - Driver intention identification method based on probability correction - Google Patents
Driver intention identification method based on probability correction Download PDFInfo
- Publication number
- CN111717217A CN111717217A CN202010624121.9A CN202010624121A CN111717217A CN 111717217 A CN111717217 A CN 111717217A CN 202010624121 A CN202010624121 A CN 202010624121A CN 111717217 A CN111717217 A CN 111717217A
- Authority
- CN
- China
- Prior art keywords
- driver
- probability
- intention
- data
- correction
- 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.)
- Granted
Links
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
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
-
- 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
-
- 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
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W2050/0001—Details of the control system
- B60W2050/0019—Control system elements or transfer functions
- B60W2050/0028—Mathematical models, e.g. for simulation
- B60W2050/0029—Mathematical model of the driver
-
- 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
- B60W2510/00—Input parameters relating to a particular sub-units
- B60W2510/06—Combustion engines, Gas turbines
- B60W2510/0604—Throttle position
-
- 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
- B60W2510/00—Input parameters relating to a particular sub-units
- B60W2510/18—Braking system
-
- 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
- B60W2520/12—Lateral 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
- B60W2540/00—Input parameters relating to occupants
- B60W2540/18—Steering angle
-
- 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
- B60W2540/00—Input parameters relating to occupants
- B60W2540/223—Posture, e.g. hand, foot, or seat position, turned or inclined
-
- 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
- B60W2540/00—Input parameters relating to occupants
- B60W2540/225—Direction of gaze
-
- 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
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Human Computer Interaction (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to a driver intention identification method based on probability correction, and belongs to the field of unmanned automobiles. The method comprises the following steps: s1: collecting and preprocessing driving behavior data; s2: acquiring environmental road information, selecting characteristic data, carrying out primary identification on the intention of a driver by using an MGHMM model, and calculating to obtain initial probabilities of observation sequences corresponding to the intention models of the drivers; s3: initial probability P of driver intention model by combining environmental road information and collected driver data1Correction is carried out to respectively obtain correction probability P2And P3(ii) a S4: and inputting the corrected probability of the driver intention model into a PSO-SVM classifier for classification and identification, and identifying the final driver intention. The invention greatly improves the accuracy and the practicability of the intention recognition of the driver. The invention can be used for drivingAnd the development and design of the driver assistance system realize the man-vehicle cooperative control of the unmanned vehicle.
Description
Technical Field
The invention belongs to the field of unmanned automobiles, relates to the field of driver intention identification and machine learning, and particularly relates to a driver intention identification method based on probability correction.
Background
The rapid development of the transportation industry and the steady increase of the automobile holding capacity bring convenience to the travel and the life of people and also provide a severe test for road traffic safety. According to statistics, most road traffic safety accidents are directly or indirectly linked with driver misoperation, so that each automobile manufacturer continuously improves the active safety and the passive safety of own vehicle, thereby improving the safety performance of the vehicle. The driver's intention is accurately and quickly recognized, and a safer and more reliable collision assessment or collision early warning can be obtained, so that the safety performance of the vehicle is improved, and the life and property safety of the driver is ensured. Furthermore, the result of the driver intention recognition can also be used to activate various driver assistance systems, such as lane keeping systems, lane change assistance systems, cruise control systems, etc.
In recent years, under the large background of the vigorous development of intelligent networked automobiles, human-vehicle cooperative control is widely concerned by researchers at home and abroad. On the one hand, there are many key technologies to achieve the L5 level of fully autonomous driving, and on the other hand, the driving will of the driver with driving interest cannot be ignored and deprived. The existing driver intention identification method has the problems of single identification information, low accuracy, poor practicability and the like, and does not fully utilize man-vehicle-road comprehensive information. Therefore, the driver intends to identify the key technology of the man-vehicle cooperative control, and needs to be further improved and optimized so as to improve the safety and reliability of the man-vehicle cooperative control.
Disclosure of Invention
In view of the above, the present invention provides a driver intention identification method based on probability correction, which aims at solving the problems of single identification information, insufficient utilization of man-car-road comprehensive information, low accuracy, poor practicability, etc. of the existing driver intention identification method, improves the safety performance of vehicles, ensures the life and property safety of drivers, perfects and optimizes a man-car cooperative control method, and provides technical support for the development of unmanned driving technology.
In order to achieve the purpose, the invention provides the following technical scheme:
a driver intention recognition method based on probability correction comprises the following steps:
s1: collecting data of driving behaviors, including vehicle data and driver data, and performing corresponding preprocessing;
s2: obtaining environment road information, selecting characteristic data, carrying out primary recognition of driver intentions by using a Multi-dimensional Gaussian Hidden Markov Model (MGHMM), and calculating to obtain initial probabilities P of observation sequences corresponding to the driver intention models1;
S3: initial probability P of driver intention model by combining environmental road information and collected driver data1Correction is carried out to respectively obtain correction probability P2And P3;
S4: inputting the corrected probability P of the driver intention model into the SVM of which the parameters are optimized by a Particle Swarm Optimization (PSO), and carrying out classification and identification by a PSO-SVM classifier so as to identify the final driver intention.
Further, the step S1 specifically includes the following steps:
s11: collecting driving behavior data comprising vehicle data and driver data; vehicle data includes, but is not limited to, steering wheel angle, rate of change of steering wheel angle, throttle opening, brake pedal effort, speed, lateral speed, longitudinal speed, yaw rate, yaw angle, pitch angle, roll angle, and the like; driver data includes, but is not limited to, driver gaze number, gaze time, driver average saccade time, average saccade angle, average saccade velocity, driver eye horizontal movement, vertical movement, driver head yaw movement, roll movement, pitch movement, etc.;
s12: and preprocessing the acquired data, including but not limited to abnormal data elimination, missing value filling, filtering, data size range unification, class classification and the like.
Further, the step S2 specifically includes the following steps:
s21: acquiring environmental road information according to a high-precision map and an environmental perception system, and selecting characteristic data for driver intention identification;
s22: training various driver intention models by using training data, and solving parameters of an MGHMM model by adopting a Baum-Welch algorithm, wherein the parameter is (pi, A, C, mu and U), wherein pi is initial state probability distribution, A is a state transition matrix, C is a mixed Gaussian element covariance matrix, mu is a mixed Gaussian element mean matrix, and U is a mixed Gaussian element covariance matrix;
s23: calculating output probability P of observation sequence to MGHMM model by adopting forward probability1(o | λ), performing primary recognition of the driver intention, and calculating the probability of each driver intention model corresponding to the observation sequence:
wherein, αT(i) Representing the forward probability, N being the number of states in the model, and T being the observation sequence length.
Further, the step S21 specifically includes: the method comprises the steps that (1) gradient, flatness and attachment coefficient of a road, position, running speed, direction and the like of surrounding vehicles are extracted from a high-precision map and used for selecting vehicle characteristic data; if the gradient change is large, the accelerator opening is not taken as characteristic data; when the road is uneven, the pitch angle is removed from the characteristic data.
Further, the step S3 specifically includes the following steps:
s31: calculating the correction summary of the driver by using the environmental road information acquired by the environmental perception system and the high-precision mapRate P2;
S32: obtaining a driver intention correction probability P using driver data3;
S33: initial probability P output by using MGHMM model1And the correction probability P2、P3And obtaining the final probability P of the driver intention model as follows: p ═ P1+P2+P3。
Further, the step S31 specifically includes: based on the environmental road information, a probability correction value of the driver intention model is obtained, and the accuracy of driver intention identification is improved. If the high-precision map displays that the front road can only turn left, a larger left-turn probability correction value is generated based on the information, so that the left-turn probability of the final intention probability value of the driver is larger; when the environment perception system perceives that the left and right sides of the vehicle are provided with the vehicle to block the left and right turning of the vehicle, a larger straight-going probability correction value is generated based on the road condition information, so that the straight-going probability of the final intention probability value of the driver is larger.
Further, the step S32 specifically includes: in order to identify the driving intention, the motion states and the motion features of eyes and heads in the driver data are extracted, the relevance between each motion state and each feature and the driving intention is obtained by a statistical method, the motion states and the features with strong relevance are selected as feature data, the driver intention is identified by using an MGHMM model, and the driver intention correction probability P is obtained3。
Further, the step S4 specifically includes the following steps:
s41: and (3) optimizing a penalty parameter c and a kernel function parameter g of the support vector machine by adopting a PSO algorithm to obtain the optimal classifier parameters as far as possible, wherein the update speed and the update position of the particles in the PSO algorithm are compared by the following formula:
where i is an arbitrary ith particle (i ═ 1,2, …, n), M is the dimension of the solution space (M ═ 1,2, …, M), k represents the number of times the current iteration has been performed, and the position of the ith particle is xi=(xi1,xi2,...,xiM)TVelocity vi=(vi1,vi2,...,viM)TIndividual extreme value of pi=(pi1,pi2,...,piM)TPopulation extremum is pg=(pg1,pg2,...,pgM)T,r1、r2Is [0,1 ]]A random number in between, ω is a non-negative inertial weight factor, c1、c2Is a learning factor;
s42: and (2) using a support vector machine, taking the probability value arrays of the final intentions of various drivers as input feature vectors, inputting the input feature vectors into a PSO-SVM classifier for classification and recognition to obtain a final intention recognition result of the drivers, wherein the decision function of the support vector machine is as follows:
wherein, in the known sample set G { (x)i,yi) I ═ 1, …, l }, (x)i,yi) Represents an arbitrary ith sample, αiIs Lagrange multiplier, K (x, x)i) B is the offset for the support vector machine kernel.
The invention has the beneficial effects that: compared with the prior art, the method has the advantages that,
1) the invention selects the characteristic data by referring to the environment and road condition information, thereby reducing the model calculation amount and improving the identification speed of the method.
2) The invention utilizes the driver data to carry out the probability correction of the driver intention model, thereby improving the accuracy of intention identification.
3) The method and the device preprocess the driver data, select the strong correlation data and identify the intention of the driver by using the selected strong correlation characteristic data.
4) The invention introduces the driver data recognition result to carry out the probability correction of the driver intention model, thereby improving the accuracy of intention recognition.
5) The method uses the corrected final driver intention recognition probability to recognize the driver intention, and uses a PSO-SVM classifier to obtain a final intention recognition result.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention may be realized and attained by the means of the instrumentalities and combinations particularly pointed out hereinafter.
Drawings
For the purposes of promoting a better understanding of the objects, aspects and advantages of the invention, reference will now be made to the following detailed description taken in conjunction with the accompanying drawings in which:
FIG. 1 is a general flow diagram of a driver intent recognition method of the present invention;
FIG. 2 is a schematic diagram of hidden Markov (HMM);
FIG. 3 is a Particle Swarm Optimization (PSO) flow chart;
FIG. 4 is a schematic diagram of a Support Vector Machine (SVM).
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It should be noted that the drawings provided in the following embodiments are only for illustrating the basic idea of the present invention in a schematic way, and the features in the following embodiments and examples may be combined with each other without conflict.
Referring to fig. 1 to 4, fig. 1 is a general schematic diagram of a preferred driver intention recognition method based on probability correction according to the present invention, and the method specifically includes the following steps:
s1: and (3) acquiring driving behavior data (including vehicle data and driver data) by using the vehicle-mounted sensor, the eye tracker and the built-in camera, and performing corresponding preprocessing. The method specifically comprises the following steps:
s11: performing driving behavior data acquisition including vehicle data and driver data; vehicle data includes, but is not limited to, steering wheel angle, rate of change of steering wheel angle, throttle opening, brake pedal effort, speed, lateral speed, longitudinal speed, yaw rate, yaw angle, pitch angle, roll angle, and the like.
S12: the driver data includes, but is not limited to, the number of times the driver gazes the left and right rear-view mirrors, the gazing time, the average saccade angle, the average saccade speed, the horizontal and vertical movements of the driver's eyes, the yaw, roll and pitch movements of the driver's head, and the like.
S13: and preprocessing the acquired data, including but not limited to abnormal data elimination, missing value filling, filtering, data size range unification, class classification and the like.
S2: according to the information obtained by the high-precision map and the environment perception system, the characteristic data is selected, the preliminary identification of the intention of the driver is carried out by using a multi-dimensional Gaussian hidden Markov model (MGHMM), and the initial probability P of the observation sequence corresponding to each intention model of the driver is obtained through calculation1. The method specifically comprises the following steps:
s21: selecting characteristic data for identifying the intention of the driver according to the high-precision map and the information acquired by the environment perception system; the method specifically comprises the following steps: the gradient, flatness and attachment coefficient of a road, the position, running speed, direction and the like of a surrounding vehicle are extracted from the high-precision map and used for selecting the vehicle characteristic data. If the change of the gradient is large, the opening degree of the accelerator is no longer used as characteristic data, and when the road is uneven, the pitch angle is removed from the characteristic data.
S22: training various driver intention models by using training data, and solving parameters of the MGHMM (lambda is equal to (pi, A, C, mu, U) by adopting a Baum-Welch algorithm, wherein pi is initial state probability distribution, A is a state transition matrix, C is a mixed Gaussian element covariance matrix, mu is a mixed Gaussian element mean matrix, and U is a mixed Gaussian element covariance matrix.
S23: calculating output probability P of observation sequence to MGHMM model by adopting forward probability1(o | λ), performing primary recognition of the driver intention, and calculating the probability of each driver intention model corresponding to the observation sequence:
wherein, αT(i) Representing the forward probability, N being the number of states in the model, and T being the observation sequence length.
S3: the initial probability of the driver intention model is corrected by combining the environmental road information constructed by the environmental perception system and the high-precision map, the driver data collected by the eye tracker and the camera, and the corrected probabilities P are respectively obtained2,P3. The method specifically comprises the following steps:
s31: calculating the probability correction value P of the driver by using the environmental perception system and the environmental road information of the high-precision map2。
Based on the environmental road information, a probability correction value of the driver intention model is obtained, and the accuracy of driver intention identification is improved. If the high-precision map displays that the front road can only turn left, a larger left-turn probability correction value is generated based on the information, so that the left-turn probability of the final intention probability value of the driver is larger; when the environment perception system perceives that the left and right sides of the vehicle are provided with the vehicle to block the left and right turning of the vehicle, a larger straight-going probability correction value is generated based on the road condition information, so that the straight-going probability of the final intention probability value of the driver is larger.
S32: obtaining a driver intention probability correction value P using driver data3。
For recognition of driving intention, eye and head movements in driver data are extractedObtaining the relevance between each motion state and characteristic and the driving intention by a statistical method, selecting the motion state and characteristic with strong relevance as characteristic data, identifying the driver intention by using MGHMM (media gateway motion map model), and obtaining the driver intention correction probability P3。
S33: probability P output by MGHMM model1And probability correction value P2,P3And obtaining a final probability value P of the intention of the driver as follows: p ═ P1+P2+P3。
S4: and inputting the corrected probability P of the driver intention model into a Support Vector Machine (SVM) with parameters optimized by a Particle Swarm Optimization (PSO), and carrying out classification and identification by a PSO-SVM classifier so as to identify the final driver intention. The method specifically comprises the following steps:
s41: and optimizing a punishment parameter c and a kernel function parameter g of the support vector machine by adopting a particle swarm optimization algorithm to obtain the optimal classifier parameter as far as possible, wherein the update speed and the position of the particles in the particle swarm optimization algorithm are compared by the following formula:
where i is an arbitrary ith particle (i ═ 1,2, …, n), M is the dimension of the solution space (M ═ 1,2, …, M), k represents the number of times the current iteration has been performed, and the position of the ith particle is xi=(xi1,xi2,...,xiM)TVelocity vi=(vi1,vi2,...,viM)TIndividual extreme value of pi=(pi1,pi2,...,piM)TPopulation extremum is pg=(pg1,pg2,...,pgM)T,r1、r2Is [0,1 ]]A random number in between, ω is a non-negative inertial weight factor, c1、c2Is a learning factor;
s42: and (2) using a support vector machine, taking the probability value arrays of the final intentions of various drivers as input feature vectors, inputting the input feature vectors into a PSO-SVM classifier for classification and recognition to obtain a final intention recognition result of the drivers, wherein the decision function of the support vector machine is as follows:
wherein, in the known sample set G { (x)i,yi) I ═ 1, …, l }, (x)i,yi) Represents an arbitrary ith sample, αiIs Lagrange multiplier, K (x, x)i) B is the offset for the support vector machine kernel.
Finally, the above embodiments are only intended to illustrate the technical solutions of the present invention and not to limit the present invention, and although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions, and all of them should be covered by the claims of the present invention.
Claims (7)
1. A driver intention recognition method based on probability correction, characterized by comprising the steps of:
s1: collecting data of driving behaviors, including vehicle data and driver data, and performing corresponding preprocessing;
s2: obtaining environment road information, selecting characteristic data, carrying out primary recognition of driver intentions by using a Multi-dimensional Gaussian Hidden Markov Model (MGHMM), and calculating to obtain initial probabilities P of observation sequences corresponding to the driver intention models1;
S3: initial probability P of driver intention model by combining environmental road information and collected driver data1Correction is carried out to respectively obtain correction probability P2And P3;
S4: inputting the corrected probability P of the driver intention model into the SVM of which the parameters are optimized by a Particle Swarm Optimization (PSO), and carrying out classification and identification by a PSO-SVM classifier so as to identify the final driver intention.
2. The probability correction-based driver intention recognition method as claimed in claim 1, wherein the step S1 specifically comprises the steps of:
s11: collecting driving behavior data comprising vehicle data and driver data; the vehicle data includes a steering wheel angle, a rate of change of the steering wheel angle, an accelerator opening, a brake pedal force, a velocity, a lateral velocity, a longitudinal velocity, a yaw rate, a yaw angle, a pitch angle, and a roll angle; the driver data comprises the fixation times and fixation time of the left and right rear-view mirrors of the driver, the average saccade time, the average saccade angle and the average saccade speed of the driver, the horizontal movement and the vertical movement of the eyes of the driver, the yaw movement, the roll movement and the pitch movement of the head of the driver;
s12: and preprocessing the acquired data.
3. The probability correction-based driver intention recognition method as claimed in claim 1, wherein the step S2 specifically comprises the steps of:
s21: acquiring environmental road information according to a high-precision map and an environmental perception system, and selecting characteristic data for driver intention identification;
s22: training various driver intention models by using training data, and solving parameters of an MGHMM model by adopting a Baum-Welch algorithm, wherein the parameter is (pi, A, C, mu and U), wherein pi is initial state probability distribution, A is a state transition matrix, C is a mixed Gaussian element covariance matrix, mu is a mixed Gaussian element mean matrix, and U is a mixed Gaussian element covariance matrix;
s23: calculating output probability P of observation sequence to MGHMM model by adopting forward probability1(o | λ), performing primary recognition of the driver intention, and calculating the probability of each driver intention model corresponding to the observation sequence:
wherein, αT(i) Representing the forward probability, N being the number of states in the model, and T being the observation sequence length.
4. The method for recognizing the driver' S intention based on the probability correction according to claim 3, wherein the step S21 specifically includes: the characteristic data extracted from the high-precision map comprises the gradient, the flatness and the adhesion coefficient of a road, and the position, the running speed and the direction of surrounding vehicles; if the gradient change is large, the accelerator opening is not taken as characteristic data; when the road is uneven, the pitch angle is removed from the characteristic data.
5. The probability correction-based driver intention recognition method according to claim 3, wherein the step S3 specifically comprises the steps of:
s31: calculating the correction probability P of the driver by using the environmental road information acquired by the environmental perception system and the high-precision map2;
S32: obtaining a driver intention correction probability P using driver data3;
S33: initial probability P output by using MGHMM model1And the correction probability P2、P3And obtaining the final probability P of the driver intention model as follows: p ═ P1+P2+P3。
6. The method for recognizing the driver' S intention based on the probability correction as claimed in claim 5, wherein the step S32 specifically includes: obtaining the relevance between each motion state and characteristic and the driving intention by a statistical method, selecting the motion state and characteristic with strong relevance as characteristic data, identifying the driver intention by using an MGHMM model, and obtaining the driver intention correction probability P3。
7. The probability correction-based driver intention recognition method according to claim 5, wherein the step S4 specifically comprises the steps of:
s41: and optimizing a penalty parameter c and a kernel function parameter g of the support vector machine by adopting a PSO algorithm to obtain an optimal classifier parameter, wherein the update speed and the position of the particles in the PSO algorithm are compared by the following formula:
wherein i is any ith particle, i is 1,2, …, n; m is the dimension of the solution space, M is 1,2, …, M, k represents the number of times the current iteration is performed, and the position of the ith particle is xi=(xi1,xi2,...,xiM)TVelocity vi=(vi1,vi2,...,viM)TIndividual extreme value of pi=(pi1,pi2,...,piM)TPopulation extremum is pg=(pg1,pg2,...,pgM)T,r1、r2Is [0,1 ]]A random number in between, ω is a non-negative inertial weight factor, c1、c2Is a learning factor;
s42: and (2) using a support vector machine, taking the probability value arrays of the final intentions of various drivers as input feature vectors, inputting the input feature vectors into a PSO-SVM classifier for classification and recognition to obtain a final intention recognition result of the drivers, wherein the decision function of the support vector machine is as follows:
wherein, in the known sample set G { (x)i,yi) I ═ 1, …, l }, (x)i,yi) Represents an arbitrary ith sample, αiIs Lagrange multiplier, K (x, x)i) As a support vectorKernel function, b is the offset.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010624121.9A CN111717217B (en) | 2020-06-30 | 2020-06-30 | Driver intention identification method based on probability correction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010624121.9A CN111717217B (en) | 2020-06-30 | 2020-06-30 | Driver intention identification method based on probability correction |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111717217A true CN111717217A (en) | 2020-09-29 |
CN111717217B CN111717217B (en) | 2022-11-08 |
Family
ID=72571061
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010624121.9A Active CN111717217B (en) | 2020-06-30 | 2020-06-30 | Driver intention identification method based on probability correction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111717217B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112744226A (en) * | 2021-01-18 | 2021-05-04 | 国汽智控(北京)科技有限公司 | Automatic driving intelligent self-adaption method and system based on driving environment perception |
CN112926404A (en) * | 2021-01-29 | 2021-06-08 | 吉林大学 | Active interactive human-vehicle passing system and method |
CN113386775A (en) * | 2021-06-16 | 2021-09-14 | 杭州电子科技大学 | Driver intention identification method considering human-vehicle-road characteristics |
CN113920734A (en) * | 2021-10-20 | 2022-01-11 | 长安大学 | Lane change early warning method based on logistic model |
CN115264048A (en) * | 2022-07-26 | 2022-11-01 | 重庆大学 | Automatic transmission intelligent gear decision design method based on data mining |
Citations (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005252859A (en) * | 2004-03-05 | 2005-09-15 | Kddi Corp | Scene-splitting device for dynamic image data |
US20050273215A1 (en) * | 2004-06-02 | 2005-12-08 | Nissan Motor Co., Ltd. | Adaptive intention estimation method and system |
JP2009262702A (en) * | 2008-04-23 | 2009-11-12 | Fuji Heavy Ind Ltd | Safe driving support system |
KR20130089351A (en) * | 2012-02-02 | 2013-08-12 | 자동차부품연구원 | A system to detect and to decide the lateral lane change intention of drivers |
JP2015056058A (en) * | 2013-09-12 | 2015-03-23 | 株式会社デンソー | Driving support device |
CN104494600A (en) * | 2014-12-16 | 2015-04-08 | 电子科技大学 | SVM (support vector machine) algorithm-based driver intention recognition method |
CN106682682A (en) * | 2016-10-20 | 2017-05-17 | 北京工业大学 | Method for optimizing support vector machine based on Particle Swarm Optimization |
CN106740864A (en) * | 2017-01-12 | 2017-05-31 | 北京交通大学 | A kind of driving behavior is intended to judge and Forecasting Methodology |
JP2017117096A (en) * | 2015-12-22 | 2017-06-29 | 三菱自動車工業株式会社 | Vehicular drive operation monitor apparatus |
CN106971194A (en) * | 2017-02-16 | 2017-07-21 | 江苏大学 | A kind of driving intention recognition methods based on the double-deck algorithms of improvement HMM and SVM |
US20180018871A1 (en) * | 2016-07-12 | 2018-01-18 | Denso Corporation | Driving assist system |
US20180050702A1 (en) * | 2015-03-31 | 2018-02-22 | Hitachi Automotive Systems, Ltd. | Automatic driving control device |
US20180053102A1 (en) * | 2016-08-16 | 2018-02-22 | Toyota Jidosha Kabushiki Kaisha | Individualized Adaptation of Driver Action Prediction Models |
CN107967486A (en) * | 2017-11-17 | 2018-04-27 | 江苏大学 | A kind of nearby vehicle Activity recognition method based on V2V communications with HMM-GBDT mixed models |
CN108806021A (en) * | 2018-06-12 | 2018-11-13 | 重庆大学 | Electric vehicle target road section energy consumption prediction technique based on physical model and roadway characteristic parameter |
CN108995655A (en) * | 2018-07-06 | 2018-12-14 | 北京理工大学 | A kind of driver's driving intention recognition methods and system |
CN109050537A (en) * | 2018-08-01 | 2018-12-21 | 合肥工业大学 | A kind of compound driving intention recognition methods |
US20190212138A1 (en) * | 2016-06-30 | 2019-07-11 | Bridgestone Corporation | Road surface state determination method and road surface state determination apparatus |
US20190337522A1 (en) * | 2018-05-04 | 2019-11-07 | Ford Global Technologies, Llc | Determining vehicle driving behavior |
CN110427850A (en) * | 2019-07-24 | 2019-11-08 | 中国科学院自动化研究所 | Driver's super expressway lane-changing intention prediction technique, system, device |
CN110766057A (en) * | 2019-10-09 | 2020-02-07 | 南京理工大学 | Gesture recognition device and method |
CN110852281A (en) * | 2019-11-13 | 2020-02-28 | 吉林大学 | Driver lane change intention identification method based on Gaussian mixture hidden Markov model |
US20200189574A1 (en) * | 2017-06-02 | 2020-06-18 | Toyota Motor Europe | Driving assistance method and system |
-
2020
- 2020-06-30 CN CN202010624121.9A patent/CN111717217B/en active Active
Patent Citations (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005252859A (en) * | 2004-03-05 | 2005-09-15 | Kddi Corp | Scene-splitting device for dynamic image data |
US20050273215A1 (en) * | 2004-06-02 | 2005-12-08 | Nissan Motor Co., Ltd. | Adaptive intention estimation method and system |
JP2009262702A (en) * | 2008-04-23 | 2009-11-12 | Fuji Heavy Ind Ltd | Safe driving support system |
KR20130089351A (en) * | 2012-02-02 | 2013-08-12 | 자동차부품연구원 | A system to detect and to decide the lateral lane change intention of drivers |
JP2015056058A (en) * | 2013-09-12 | 2015-03-23 | 株式会社デンソー | Driving support device |
CN104494600A (en) * | 2014-12-16 | 2015-04-08 | 电子科技大学 | SVM (support vector machine) algorithm-based driver intention recognition method |
US20180050702A1 (en) * | 2015-03-31 | 2018-02-22 | Hitachi Automotive Systems, Ltd. | Automatic driving control device |
JP2017117096A (en) * | 2015-12-22 | 2017-06-29 | 三菱自動車工業株式会社 | Vehicular drive operation monitor apparatus |
US20190212138A1 (en) * | 2016-06-30 | 2019-07-11 | Bridgestone Corporation | Road surface state determination method and road surface state determination apparatus |
US20180018871A1 (en) * | 2016-07-12 | 2018-01-18 | Denso Corporation | Driving assist system |
US20180053102A1 (en) * | 2016-08-16 | 2018-02-22 | Toyota Jidosha Kabushiki Kaisha | Individualized Adaptation of Driver Action Prediction Models |
CN106682682A (en) * | 2016-10-20 | 2017-05-17 | 北京工业大学 | Method for optimizing support vector machine based on Particle Swarm Optimization |
CN106740864A (en) * | 2017-01-12 | 2017-05-31 | 北京交通大学 | A kind of driving behavior is intended to judge and Forecasting Methodology |
CN106971194A (en) * | 2017-02-16 | 2017-07-21 | 江苏大学 | A kind of driving intention recognition methods based on the double-deck algorithms of improvement HMM and SVM |
US20200189574A1 (en) * | 2017-06-02 | 2020-06-18 | Toyota Motor Europe | Driving assistance method and system |
CN107967486A (en) * | 2017-11-17 | 2018-04-27 | 江苏大学 | A kind of nearby vehicle Activity recognition method based on V2V communications with HMM-GBDT mixed models |
US20190337522A1 (en) * | 2018-05-04 | 2019-11-07 | Ford Global Technologies, Llc | Determining vehicle driving behavior |
CN108806021A (en) * | 2018-06-12 | 2018-11-13 | 重庆大学 | Electric vehicle target road section energy consumption prediction technique based on physical model and roadway characteristic parameter |
CN108995655A (en) * | 2018-07-06 | 2018-12-14 | 北京理工大学 | A kind of driver's driving intention recognition methods and system |
CN109050537A (en) * | 2018-08-01 | 2018-12-21 | 合肥工业大学 | A kind of compound driving intention recognition methods |
CN110427850A (en) * | 2019-07-24 | 2019-11-08 | 中国科学院自动化研究所 | Driver's super expressway lane-changing intention prediction technique, system, device |
CN110766057A (en) * | 2019-10-09 | 2020-02-07 | 南京理工大学 | Gesture recognition device and method |
CN110852281A (en) * | 2019-11-13 | 2020-02-28 | 吉林大学 | Driver lane change intention identification method based on Gaussian mixture hidden Markov model |
Non-Patent Citations (6)
Title |
---|
LIU YUN: "Driving Intentions Identification Based on", 《2011 SECOND INTERNATIONAL CONFERENCE ON DIGITAL MANUFACTURING & AUTOMATION》 * |
LIU YUN: "Driving Intentions Identification Based on", 《2011 SECOND INTERNATIONAL CONFERENCE ON DIGITAL MANUFACTURING & AUTOMATION》, 20 October 2011 (2011-10-20), pages 460 - 464 * |
宋晓琳等: "基于HMM-SVM的驾驶员换道意图辨识研究", 《电子测量与仪器学报》, no. 01, 15 January 2016 (2016-01-15), pages 58 - 64 * |
张亚桥等: "基于驾驶员意图识别的微型纯电动货车控制策略", 《河南科技大学学报(自然科学版)》 * |
张亚桥等: "基于驾驶员意图识别的微型纯电动货车控制策略", 《河南科技大学学报(自然科学版)》, no. 05, 21 June 2018 (2018-06-21), pages 31 - 37 * |
王忻: "基于改进的隐马尔可夫模型交通拥堵识别研究", 《兰州交通大学学报》, no. 05, 15 October 2018 (2018-10-15), pages 29 - 34 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112744226A (en) * | 2021-01-18 | 2021-05-04 | 国汽智控(北京)科技有限公司 | Automatic driving intelligent self-adaption method and system based on driving environment perception |
CN112926404A (en) * | 2021-01-29 | 2021-06-08 | 吉林大学 | Active interactive human-vehicle passing system and method |
CN112926404B (en) * | 2021-01-29 | 2022-06-24 | 吉林大学 | Active interactive human-vehicle passing system and method |
CN113386775A (en) * | 2021-06-16 | 2021-09-14 | 杭州电子科技大学 | Driver intention identification method considering human-vehicle-road characteristics |
CN113386775B (en) * | 2021-06-16 | 2022-06-17 | 杭州电子科技大学 | Driver intention identification method considering human-vehicle-road characteristics |
CN113920734A (en) * | 2021-10-20 | 2022-01-11 | 长安大学 | Lane change early warning method based on logistic model |
CN113920734B (en) * | 2021-10-20 | 2023-02-28 | 长安大学 | Lane change early warning method based on logistic model |
CN115264048A (en) * | 2022-07-26 | 2022-11-01 | 重庆大学 | Automatic transmission intelligent gear decision design method based on data mining |
Also Published As
Publication number | Publication date |
---|---|
CN111717217B (en) | 2022-11-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111717217B (en) | Driver intention identification method based on probability correction | |
CN108995655B (en) | Method and system for identifying driving intention of driver | |
CN108995653B (en) | Method and system for identifying driving style of driver | |
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 | |
CN108773376B (en) | Automobile multi-target layered cooperative control and optimization method fusing driving intentions | |
CN106184223A (en) | Automatic driving control method and device and automobile | |
CN111332362B (en) | Intelligent steer-by-wire control method integrating individual character of driver | |
CN110626339B (en) | Emergency obstacle avoidance automatic driving assistance torque calculation method based on driver intention | |
AU2019205817A1 (en) | Software validation for autonomous vehicles | |
CN104494600B (en) | A kind of Driver intention recognition method based on SVM algorithm | |
CN109145719B (en) | Driver fatigue state identification method and system | |
CN109727490B (en) | Peripheral vehicle behavior self-adaptive correction prediction method based on driving prediction field | |
CN112116100B (en) | Game theory decision method considering driver type | |
CN113291311B (en) | Method for detecting abnormal behavior of driver under emergency collision avoidance working condition and storage medium | |
CN107521501A (en) | Driver assistance system decision-making technique, system based on game theory and other | |
US11454971B2 (en) | Methods and systems for learning user preferences for lane changes | |
CN111559378A (en) | Four-wheel independent-drive electric vehicle dynamics control method considering driver characteristics | |
CN111830962A (en) | Interpretation data for reinforcement learning agent controller | |
CN109515441B (en) | Vehicle speed control system for intelligent driving vehicle | |
KR20150066308A (en) | Apparatus and method for determining driving condition of deiver | |
US20230219569A1 (en) | Personalized vehicle operation for autonomous driving with inverse reinforcement learning | |
CN116767237A (en) | Fraud detection for hands-on automated driving | |
CN115384539A (en) | Method and device for generating intelligent driving strategy of vehicle, vehicle and storage medium | |
JP2022552938A (en) | In-vehicle cluster tracking system |
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 |