Driver behavior analysis and prediction method based on causal inference
By constructing a driving index evaluation system and applying causal inference theory, combined with dual machine learning models, the driver behavior problem that is difficult to in-depth causal estimation in the existing technology is solved, and a comprehensive analysis and prediction of driver behavior in AV and HV environments is achieved. , improving traffic safety and efficiency.
Patent Information
- Application Number
- CN202411088252.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-08-08
AI Technical Summary
The existing driver behavior analysis methods are difficult to conduct in-depth causal estimation in AV and HV mixed environments, especially when dealing with counterfactual data in multiple scenarios, it is difficult to meet the requirements of complex simulation experiments and cannot effectively improve road traffic efficiency. and the social recognition of AV.
Build a driving index evaluation system, including the front distance, the imminent collision time, speed and brake amplitude, and eye tracking information, and apply causal inference theory and dual machine learning model to output causal estimates of driving behavior through data preprocessing and causal inference models. , the decomposition condition is estimated as the prediction subtask of the result and process, estimating ATE and CATE.
In-depth analysis and prediction of driver behavior has been achieved, traffic safety and driving efficiency have been improved, analysis accuracy and reliability have been enhanced, multiple factors can be considered more comprehensively, and the basis for harmonious coexistence between AV and HV has been enhanced.
Smart Images

Figure CN119037444A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of human factor analysis in traffic engineering, and more specifically relates to a driver behavior analysis and prediction method based on causal inference. Background Art
[0002] In recent years, with the acceleration of urbanization, traffic congestion and frequent traffic accidents have become serious urban diseases. The development of AV provides an innovative solution. Its core advantage is that it can effectively reduce traffic accidents, optimize traffic flow and improve road use efficiency through advanced perception systems and intelligent algorithms. AV can analyze traffic conditions in real time, automatically adjust driving routes and speeds, and avoid traffic congestion and potential collisions. In addition, autonomous driving technology can reduce driver fatigue and human errors, further improving road safety. With the continuous advancement and maturity of technology, AV is expected to become an important part of the future urban transportation system, greatly improving urban traffic conditions and promoting sustainable development. In the near future, AV will gradually coexist with HV in complex traffic scenes and road environments. Before autonomous driving technology is widely used, it is of great practical significance to analyze and judge the interaction between AV and HV. For example, after studying how drivers react to AVs merging into traffic while driving, autonomous driving companies can improve the systems and algorithms of autonomous driving vehicles based on their driving behavior; traffic planning and management departments can also improve road designs that are more friendly to AVs and HVs based on this. However, the existing analysis methods of the impact of AV and HV on driver behavior either remain at the macro level or only conduct relatively simple driving index analysis without in-depth modeling and mining of driver behavior. It is difficult to effectively meet the requirements of complex and repeatable simulation experiments, especially when counterfactual data in multiple scenarios (i.e., data that does not exist in actual experiments) needs to be processed. Therefore, it is necessary to define a comprehensive and integrated driver behavior analysis and modeling framework to promote the appearance design and algorithm design of AV, thereby improving road traffic efficiency and AV social acceptance, and laying a solid foundation for the harmonious coexistence of AV and HV in the future. Summary of the invention
[0003] The technical problem to be solved by the present invention is to propose a driver behavior analysis and prediction method based on causal inference, so as to achieve a comprehensive and comprehensive causal estimation of driver behavior in an AV and HV mixed traffic environment.
[0004] In order to achieve the above object, the present invention is implemented by adopting the following technical scheme: the method comprises:
[0005] Construct a driving index evaluation system, including four types of indicators: headway, time to collision, speed and braking range, and eye tracking information;
[0006] Apply causal inference theory to driving behavior analysis and study the differences in driving behavior of human drivers when encountering AVs and HVs;
[0007] For a specific driving behavior evaluation indicator, the experimental data of the driver in the AV scenario and the HV scenario are collected respectively. After data preprocessing, the data is input into the causal inference model based on dual machine learning, which can output the causal estimation of the driving behavior.
[0008] In one embodiment, the headway is the time interval between two consecutive vehicles passing the same point; the time interval is calculated as the spatial interval between the leading vehicle and the trailing vehicle divided by the speed of the trailing vehicle, as shown below:
[0009]
[0010] Where X L (t) represents the position of the preceding vehicle at time t, X F (t) is the position of the following vehicle at time t, V F (t) represents the speed of the following vehicle at time t.
[0011] In one embodiment, the time to collision is defined as "the time remaining before a collision between two vehicles if a collision occurs", as described below:
[0012]
[0013] Where X L (t) represents the position of the preceding vehicle at time t, X F (t) is the position of the rear vehicle at time t, L is the length of the front vehicle, V F (t) represents the speed of the following vehicle at time t, V L (t) represents the speed of the preceding vehicle at time t;
[0014] Headway time is mainly used to measure following distance, that is, collision time is used to indicate the collision risk and safety level in following situations.
[0015] In one embodiment, the learning task of the dual machine learning based causal inference model is decomposed into two stages:
[0016] (1) Decompose the conditional estimation of the residual into two prediction subtasks: the result and the processing;
[0017] (2) Obtain estimates of ATE and CATE through the final prediction model with outcome and treated residuals.
[0018] In one embodiment, the causal inference model based on dual machine learning has the following specific steps:
[0019] Y=θ(X)·T+g(X)+∈#(1)
[0020] T=f(X)+η#(2)
[0021] E[∈|X]=0; E[η|X]=0; E[η·∈|
[0022] In equation (1), Y is the outcome, which is the characteristics of the vehicle driven by the driver, including speed, distance from AV or HV, θ(X) is the treatment effect of each X, representing the causal effect of treatment T on the outcome Y, T is the treatment method, that is, the type of vehicle, which is AV or HV in the T-intersection scenario, and g(X) is the function used to control the effects of covariates and confounding factors. ∈ is the random error, which represents the randomness in Y that cannot be explained by the model.
[0023] In equation (2), T is the treatment mentioned above, f(X) is the model that represents how T depends on the covariate X, and η is a random error similar to ∈ in equation (1);
[0024] In formula (3), given X, the expected values of the error terms ∈ and η are zero, which means that X contains all relevant information that affects Y and T; X is the background data of the driver;
[0025] The purpose of the process is to estimate the constant marginal CATE, that is, θ(X); the idea of estimating θ(X) is as follows, and the equation is rewritten as:
[0026] YE[Y|X]=θ(X)·(TE[TIX])+∈#(4)
[0027] In equation (4), E[Y|X] and E[T|X] are two conditional expectation functions, both of which are non-parametric regression tasks;
[0028] Therefore, according to q(X) = E[Y|X] and f(X) = E[T|X], the residual is calculated:
[0029]
[0030] It is then related by the following equation:
[0031]
[0032] Subsequently, since (E[η·∈|X]=0), estimating θ(X) is the final regression problem, i.e., regressing Y on X, T (although the model is linear in T), i.e.:
[0033]
[0034] At this point, the value of CATE (i.e., θ(X)) is obtained.
[0035] In one embodiment, the speed and braking amplitude are indicators for measuring the driver's driving behavior and indicating his driving state; the driving state of the vehicle driven by the driver is clearly indicated by its speed;
[0036] Driver braking behavior reflects the actions taken by drivers when approaching vehicles or objects in the surrounding area that are too close to them.
[0037] In one embodiment, the eye tracking information includes: gaze direction, vergence, eye opening, pupil diameter, pupil position, focus object position, focus object type and name, focus object distance from the eyeball, and focus object gaze duration.
[0038] By constructing a driving index evaluation system and applying causal inference theory to driving behavior analysis, we can deeply study the differences in driving behavior of human drivers when encountering autonomous driving (AV) and manual driving (HV). And we can analyze specific driving behavior evaluation indicators and predict the driver's driving behavior, so as to enhance traffic safety and improve driving efficiency. The causal inference model using dual machine learning can more accurately estimate the causal effect of driving behavior, and the model fitting accuracy is high, and the prediction results are more reliable. In addition, this method also takes into account multiple factors such as eye tracking information, which is more comprehensive when analyzing driver behavior and can achieve more accurate driver behavior prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 is a flow chart of the method of the present invention;
[0040] Figure 2 This is a diagram of the driving index evaluation system of the present invention;
[0041] Figure 3 This is a flow chart of the causal effect estimation method based on dual machine learning of the present invention;
[0042] Figure 4 The figure is a flow chart of the construction process of the heterogeneity analysis decision tree of the present invention. DETAILED DESCRIPTION
[0043] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Typical embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in the present invention. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.
[0044] Unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.
[0045] like Figure 1 As shown, a driver behavior analysis and prediction method based on causal inference includes:
[0046] S1. Construct a driving index evaluation system, which includes four types of indicators: headway, time to collision, speed and braking amplitude, and eye tracking information.
[0047] like Figure 2 As shown in the figure, in order to measure the driving behavior of the driver, a driving index evaluation system needs to be constructed. In this system, there are four types of indicators, namely: Time Headway, Time To Collision (TTC), speed and braking amplitude, and eye tracking information. The first two indicators are applicable to the interaction between the vehicle driven by the driver and other vehicles, and the research objects of the latter two indicators are the vehicle driven by the driver or the driver himself. Time Headway is used to measure the following distance between the vehicle driven by the driver and other vehicles; TTC is used to indicate the collision risk and safety level in the following situation; speed and braking amplitude directly indicate the state of the vehicle driven by the driver, that is, the speed and braking amplitude of the vehicle; eye tracking information reflects the distribution law of the driver's eye movement, including the gaze direction, convergence, eye opening, pupil diameter, pupil position, focus object position, type and name of the focus object, distance of the focus object from the eyeball, gaze duration of the focus object, etc. Through these eye movement data indicators, the deep behavioral factors of the driver can be excavated.
[0048] S101, Calculate Time Headway
[0049] Time Headway is defined as the time interval between two consecutive vehicles passing the same point. It provides important guidance for road design and management, which depends on traffic flow composition and driving behavior. The time headway is calculated as the spatial separation between the leading and following vehicles divided by the speed of the following vehicle, as follows:
[0050]
[0051] Where X L (t) represents the position of the preceding vehicle at time t, X F (t) is the position of the following vehicle at time t, V F (t) represents the speed of the following vehicle at time t.
[0052] S102: Calculate Time To Collision (TTC)
[0053] TTC is a common indicator used to represent safety, which is defined as "the time remaining before a collision between two vehicles if a collision occurs". It reflects the degree of driving safety and the risk of collision. Mathematically, it is described as follows:
[0054]
[0055] Where X L (t) represents the position of the preceding vehicle at time t, X F (t) is the position of the rear vehicle at time t, L is the length of the front vehicle, V F (t) represents the speed of the following vehicle at time t, V L (t) represents the speed of the preceding vehicle at time t.
[0056] In the present invention, the headway time is mainly used to measure the following distance, and TTC is used to indicate the collision risk and safety level in the following situation.
[0057] S103, speed and braking range
[0058] In order to measure the driver's driving behavior, indicators that directly represent their driving status are needed. The driving status of the vehicle driven by the driver is clearly represented by its speed. In addition, the driver's braking behavior reflects the measures taken by the driver when approaching the vehicle or objects in the surrounding area that are too close to them, thereby minimizing the possibility of collision. In this category of indicators, the research object is the vehicle driven by the driver itself.
[0059] S104, Eye Tracking Information
[0060] Compared with the above-mentioned traditional driving behavior evaluation indicators, it is crucial to also take into account the eye tracking data of driving habits, which can dig out the deep behavioral factors of the driver. Eye tracking information includes: eye gaze direction, vergence, eye openness, pupil diameter, pupil position, focal object position, focal object type and name, focal object distance from the eyeball, focal object gaze duration, etc. These eye movement information data are relatively rich. In actual analysis, it is not necessary to use all of them, and they can be flexibly adjusted according to actual conditions. The following is a specific use case description.
[0061] In order to analyze the behavior of the driver's eye information rules, five indicators are used: the type and name of the focus object, the distance of the focus object from the eyeball, the longest gaze time among all focus objects, the number of types of all focus objects, and the pupil position of the eye focus. The first indicator can directly reflect the type of focus object; the second indicator can reflect the distance law of the driver's gaze on the object; based on the first indicator, the fourth indicator can be further obtained, which can reflect the driver's concentration together with the third indicator. Specifically, the larger the third indicator is, the smaller the fourth indicator is, indicating that the driver is more focused during driving; the fifth indicator can reflect the distribution law of the eyeball from a macro level.
[0062] like Figure 3 As shown in S2, causal inference theory is applied to driving behavior analysis to study the differences in driving behaviors of human drivers when encountering AVs and HVs.
[0063] For two variables, correlation does not necessarily mean causality; however, causality is usually manifested as correlation at the statistical level. Causal inference is such a statistical theory and method for evaluating causal effects and using data to discover causal relationships between variables. For decades, it has been an important research topic in many fields such as statistics, computer science, education, public policy, and economics. In the present invention, based on the above-mentioned driving index evaluation system, causal inference theory is applied to driving behavior analysis to comprehensively study the differences in driving behavior of human drivers when encountering AV and HV. Specifically, for a specific driving behavior evaluation index, the experimental data of the driver in the AV scenario and the HV scenario are collected respectively, and after certain data preprocessing, they are input into the causal inference model based on dual machine learning adopted by the present invention, which can output the causal estimate of driving behavior.
[0064] Basic concepts and symbols of causal inference, defining basic concepts and symbols, including units, treatments, observations, and counterfactuals.
[0065] Unit. A unit is the atomic object of study. A unit can be a collection of physical objects, companies, or individuals at a specific point in time, such as a classroom or a market. In this invention, the unit refers to the driver of the test vehicle in the experiment, i.e., the subject.
[0066] Treatment. A treatment is an action imposed on a unit. In the present invention, the treatment is the class of vehicle, i.e., AV or HV, that interacts with the vehicle driven by the driver. Therefore, the treatment is a binary variable, defined as T=1 if the vehicle interacting with the test vehicle is an AV, and T=0 if it is an HV. If T=1, the collected experimental data or samples belong to the treatment group, otherwise they belong to the control group.
[0067] Observed outcome: The observed outcome is the result of the actual treatment taken by the unit, which is also called the actual outcome. Y(T=t) represents the actual outcome when treatment T exists.
[0068] Counterfactual outcome. A counterfactual outcome is the result of the unit taking a different treatment (i.e., the opposite of the actual treatment). The difference between an observed outcome and a counterfactual outcome is that the observed outcome actually occurred, while the counterfactual outcome did not actually exist.
[0069] Pre-treatment variables. Pre-treatment variables are variables that are not affected by the treatment. Pre-treatment variables are also called background variables, they can be the distance between different vehicles, the speed of vehicles before the treatment time, etc.
[0070] Confounders. Confounders are variables that affect both the treatment and the outcome. The relationship between the variables is implicit, not linear. Confounders interfere with the effect of the treatment on the outcome and can lead to erroneous conclusions. For example, girls who eat breakfast every day usually have lower body weight. Eating breakfast seems to help with weight loss, but it may be that other factors, such as regular exercise and adequate sleep, play an important role in weight loss. In this case, having a better lifestyle is a common cause of breakfast and being lighter, so we can also consider it a confounder of the causal relationship between breakfast and being lighter.
[0071] The evaluation indicator of causal effect, treatment effect, can be defined at the group, treatment group, control group and individual levels.
[0072] At the group level, the treatment effect is named the Average Treatment Effect (ATE), which is defined as:
[0073]
[0074] Among them, Y(T=1) represents the results of the treatment group at the group level, and Y(T=0) represents the results of the control group at the group level.
[0075] For the treated group, the treatment effect is named the average treatment effect on the treated group (ATT), which is defined as:
[0076]
[0077] in, and denote the true and counterfactual outcomes of the treatment groups, respectively.
[0078] For the control group, the treatment effect is named the average treatment effect on the treatment group (ATC), which is defined as:
[0079]
[0080] in, and denote the counterfactual and true results of the control group, respectively.
[0081] The basic assumptions of causal inference, the causal inference model has three major assumptions, namely:
[0082] Assumption 1: Stable Unit Treatment Value Assumption (SUTVA). Units will not affect each other. That is, the driving behavior of a driver will not be affected by other drivers.
[0083] Assumption 2: (Statistical) Ignorability. Given the background variable X, the treatment assignment T is independent of the potential outcome, which can be expressed mathematically as:
[0084] W||Y(=0),Y(T=1)|X
[0085] Assumption 3: Positivity. For any value of X, the treatment allocation is not deterministic, and the mathematical expression is:
[0086] P(T=t|X=x)>0,
[0087] That is, for a driver with any characteristics, it can be tested in both AV and HV scenarios. In the method mentioned in the present invention, the above three assumptions are all met.
[0088] Causal effect estimation based on dual machine learning methods. The potential heterogeneous effects of AV or HV on driving behavior are difficult to evaluate using traditional statistical methods (such as least squares regression). The dual machine learning method fundamentally solves this problem. It is a debiased orthogonal method used to estimate the average treatment effect (ATE) between treatment and outcome, and can further estimate the conditional average treatment effect (CATE) based on the high-dimensional observable features of the treatment samples. More importantly, it can further explain the heterogeneous effects based on decision trees. The following is an explanation of the principle of this method.
[0089] The dual machine learning task is decomposed into two stages:
[0090] (1) Decompose the conditional estimation of the residual into two prediction subtasks: the result and the processing.
[0091] (2) Obtain estimates of ATE and CATE through the final prediction model with outcome and treated residuals.
[0092] The steps are as follows:
[0093] Y=θ(X)·T+g(X)+∈#(1)
[0094] T=f(X)+η#(2)
[0095] E[∈|X]=0; E[η|X]=0; E[η·∈|
[0096] In equation (1), Y is the outcome, which can be the characteristics of the vehicle driven by the driver, such as speed, distance to AV or HV, etc. θ(X) is the treatment effect of each X, representing the causal effect of treatment T on the outcome Y. T is the treatment, that is, the type of vehicle, which is AV or HV in the T-intersection scenario. g(X) is the function used to control the effects of covariates and confounders. ∈ is the random error, which represents the randomness in Y that cannot be explained by the model.
[0097] In equation (2), T is the treatment mentioned above, f(X) is a model that represents how T depends on the covariate X, and η is a random error similar to ∈ in equation (1).
[0098] In formula (3), it is assumed that given X, the expected values of the error terms ∈ and η are zero, which means that X contains all relevant information that affects Y and T. In other words, the model has considered all confounding variables. In this study, X is the driver's background information, such as age and other data.
[0099] The purpose of the process is to estimate the constant marginal CATE, that is, θ(X). The idea of estimating θ(X) is as follows, and the equation is rewritten as:
[0100] YE[Y|X]=θ(X)·(TE[T|X])+∈#(4)
[0101] In equation (4), E[Y|X] and E[T|X] are two conditional expectation functions (both are non-parametric regression tasks).
[0102] Therefore, if we can estimate them according to q(X) = E[Y|X] and f(X) = E[T|X], we can calculate the residual:
[0103]
[0104] It is then related by the following equation:
[0105]
[0106] Subsequently, since (E[η·∈|X]=0), estimating θ(X) is the final regression problem, i.e., regressing Y on X, T (although the model is linear in T), i.e.:
[0107]
[0108] So far, the value of CATE (i.e., θ(X)) has been obtained. In addition, the causal effect estimation method based on the dual machine learning method can also make a causal tree to further conduct heterogeneity analysis. The construction process of the heterogeneity analysis decision tree is shown in the attached figure. Figure 4 shown.
[0109] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).
[0110] It should be understood that the detailed description of the technical solutions of the present invention by means of the preferred embodiments is illustrative rather than restrictive. A person skilled in the art may modify the technical solutions described in the embodiments, or replace some of the technical features by equivalents, based on reading the specification of the present invention; and these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A driver behavior analysis and prediction method based on causal inference, characterized by: The method includes: Construct a driving index evaluation system, including four types of indicators: headway, time to collision, speed and braking range, and eye tracking information; Apply causal inference theory to driving behavior analysis and study the differences in driving behavior of human drivers when encountering AVs and HVs; For a specific driving behavior evaluation indicator, the experimental data of the driver in the AV scenario and the HV scenario are collected respectively. After data preprocessing, the data is input into the causal inference model based on dual machine learning, which can output the causal estimation of the driving behavior.
2. The driver behavior analysis and prediction method based on causal inference according to claim 1 is characterized by: The headway is the time interval between two consecutive vehicles passing the same point; the time interval is calculated as the spatial interval between the leading vehicle and the trailing vehicle divided by the speed of the trailing vehicle, as shown below: Where X L (t) represents the position of the preceding vehicle at time t, X F (t) is the position of the following vehicle at time t, V F (t) represents the speed of the following vehicle at time t.
3. The driver behavior analysis and prediction method based on causal inference according to claim 1 is characterized by: The time to collision is defined as "the remaining time before a collision between two vehicles if a collision occurs", as described below: Where X L (t) represents the position of the preceding vehicle at time t, X F (t) is the position of the rear vehicle at time t, L is the length of the front vehicle, V F (t) represents the speed of the following vehicle at time t, VL ( t) represents the speed of the preceding vehicle at time t; Headway time is mainly used to measure following distance, that is, collision time is used to indicate the collision risk and safety level in following situations.
4. The driver behavior analysis and prediction method based on causal inference according to claim 1 is characterized by: The learning task of the dual machine learning-based causal inference model is decomposed into two stages: (1) Decompose the conditional estimation of the residual into two prediction subtasks: the result and the processing; (2) Obtain estimates of ATE and CATE through the final prediction model with outcome and treated residuals.
5. The driver behavior analysis and prediction method based on causal inference according to claim 1 is characterized by: The specific steps of the causal inference model based on dual machine learning are as follows: Y=θ(X)·T+g(X)+∈#(1) T=f(X)+η#(2) E[∈|X]=0; E[η|X]=0; E[η·∈| In equation (1), Y is the outcome, which is the characteristics of the vehicle driven by the driver, including speed, distance from AV or HV, θ(X) is the treatment effect of each X, representing the causal effect of treatment T on the outcome Y, T is the treatment method, that is, the type of vehicle, which is AV or HV in the T-shaped intersection scenario, g(X) is the function used to control the effects of covariates and confounding factors; ∈ is the random error, which represents the randomness in Y that cannot be explained by the model; In equation (2), T is the treatment mentioned above, f(X) is the model that represents how T depends on the covariate X, and η is a random error similar to ∈ in equation (1); In formula (3), given X, the expected values of the error terms ∈ and η are zero, which means that X contains all relevant information that affects Y and T; X is the background data of the driver; The purpose of the process is to estimate the constant marginal CATE, that is, θ(X); the idea of estimating θ(X) is as follows, and the equation is rewritten as: YE[Y|X]=θ(X)·(TE[T|X])+∈#(4) In equation (4), E[Y|X] and E[T|X] are two conditional expectation functions, both of which are non-parametric regression tasks; Therefore, according to q(X) = E[Y|X] and f(X) = E[T|X], the residual is calculated: It is then related by the following equation: Subsequently, since (E[η·∈|X]=0), estimating θ(X) is the final regression problem, i.e., regressing Y on X, T (although the model is linear in T), i.e.: At this point, the value of CATE (i.e., θ(X)) is obtained.
6. The driver behavior analysis and prediction method based on causal inference according to claim 1 is characterized by: The speed and braking amplitude are indicators for measuring the driver's driving behavior and indicating his driving status; the driving status of the vehicle driven by the driver is clearly indicated by its speed; Driver braking behavior reflects the actions taken by drivers when approaching vehicles or objects in the surrounding area that are too close to them.
7. The driver behavior analysis and prediction method based on causal inference according to claim 1 is characterized by: The eye tracking information includes: gaze direction, vergence, eye opening, pupil diameter, pupil position, focus object position, focus object type and name, distance of the focus object from the eyeball, and gaze duration of the focus object.
Citation Information
Patent Citations
Method for evaluating immediate control ability of driver
CN115366890A
Driver-centric risk assessment: risk object identification via causal inference with intent-aware driving models
US20210261148A1
Cited By
Vehicle driving behavior monitoring method and device
CN120711236A
