CN117076816A

CN117076816A - Response prediction method, response prediction apparatus, computer device, storage medium, and program product

Info

Publication number: CN117076816A
Application number: CN202310890538.3A
Authority: CN
Inventors: 聂冰冰; 秦德通; 王情帆; 李泉; 卢天乐; 刘斯源; 周青
Original assignee: Tsinghua University
Current assignee: Tsinghua University
Priority date: 2023-07-19
Filing date: 2023-07-19
Publication date: 2023-11-17
Anticipated expiration: 2043-07-19
Also published as: CN117076816B

Abstract

The present application relates to a response prediction method, apparatus, computer device, storage medium, and program product. The method comprises the following steps: for a target dangerous state traffic scene, starting from a preset stage of the target dangerous state traffic scene, calculating the accumulation amount of target evidence information at intervals of preset time, wherein the target evidence information is evidence information on which a target decision is made; after the accumulated quantity is obtained through each calculation, determining whether a decision response condition is met or not according to the accumulated quantity corresponding to the current calculation moment; if the decision response condition is met, determining the response time required by the driver to make the target decision according to the current calculation time, wherein in the process, a deep learning model is not required to be driven by big data in the traditional technology, so that the difficulty of predicting the response time of the driver is reduced.

Description

Response prediction method, response prediction apparatus, computer device, storage medium, and program product

Technical Field

The present application relates to the field of prediction technology, and in particular, to a response prediction method, apparatus, computer device, storage medium, and program product.

Background

Along with the development of automatic driving technology, the driving assistance system is more and more intelligent, and the requirements of users on the driving assistance system are also higher and higher, wherein in dangerous state traffic scenes, the decision process of a driver and the response time of various decisions are predicted, and more variable inputs can be provided for the driving assistance system to predict the motion planning and the path of a vehicle.

In the traditional technology, the response time of a driver in a normal traffic scene is predicted through a deep learning model, and the prediction in a dangerous state traffic scene is lacking, because the deep learning model needs large data driving and the data in the dangerous state traffic scene is difficult to acquire, the traditional technology has the problem that the difficulty of predicting the decision response of the driver in the dangerous state traffic scene is large.

Disclosure of Invention

In view of the foregoing, it is desirable to provide a response prediction method, apparatus, computer device, storage medium, and program product that can reduce the difficulty in predicting a driver's decision response.

In a first aspect, the present application provides a response prediction method. The method comprises the following steps: for the target dangerous state traffic scene, starting from a preset stage of the target dangerous state traffic scene, calculating the accumulation amount of target evidence information every other preset time length, wherein the target evidence information is evidence information on which a target decision is made; after the accumulated quantity is obtained through each calculation, determining whether a decision response condition is met or not according to the accumulated quantity corresponding to the current calculation moment; and if the decision response condition is met, determining the response time required by the driver for making the target decision according to the current calculation time.

In one embodiment, calculating the accumulated amount of the target evidence information at intervals of a preset time period includes: and acquiring the accumulated quantity of the target evidence information obtained by calculation at the previous calculation time at intervals of preset time, determining the first evidence information quantity supporting the target decision currently and the second evidence information quantity opposite to the target decision according to the road information at the current time in the target dangerous state traffic scene, and acquiring the accumulated quantity of the target evidence information at the current calculation time according to the accumulated quantity, the first evidence information quantity and the second evidence information quantity corresponding to the previous calculation time.

In one embodiment, obtaining the accumulated amount of the target evidence information at the current computing time according to the accumulated amount, the first evidence information amount and the second evidence information amount corresponding to the previous computing time includes: acquiring the noise amount at the current calculation time based on the wiener process; and acquiring the accumulated quantity of the target evidence information at the current calculation moment according to the accumulated quantity, the first evidence information quantity, the second evidence information quantity and the noise quantity corresponding to the last calculation moment.

In one embodiment, determining whether the decision response condition is satisfied according to the accumulated amount corresponding to the current calculation time includes: determining whether a decision response condition is met according to the accumulated quantity, the driver parameter and the vehicle parameter corresponding to the current calculation moment; wherein the driver parameter is related to information of a driver in the target dangerous state traffic scene and the vehicle parameter is related to a state of a vehicle in the target dangerous state traffic scene.

In one embodiment, determining whether the decision response condition is satisfied according to the accumulated amount, the driver parameter, and the vehicle parameter corresponding to the current calculation time includes: determining whether the sum of the accumulated quantity corresponding to the current calculation time and the driver parameter is greater than or equal to the vehicle parameter; and if the vehicle parameter is greater than or equal to the vehicle parameter, determining that the decision response condition is met.

In one embodiment, the vehicle parameter is calculated according to a state of the vehicle in the target dangerous state traffic scene and an objective function, wherein the objective function is a function with gradually decreasing function value along with time.

In one embodiment, the target decision is a decision in a decision pool, the method further comprising: determining the response time length corresponding to each target decision in the decision pool; and sequencing a plurality of target decisions included in the decision pool according to the response time length corresponding to each target decision, and determining the response process of the driver in the target dangerous state traffic scene.

In a second aspect, the application further provides a response prediction device. The device comprises: the calculation module is used for calculating the accumulation amount of target evidence information for the target dangerous state traffic scene from the preset stage of the target dangerous state traffic scene at intervals of preset time, wherein the target evidence information is evidence information on which a target decision is made; the first determining module is used for determining whether the decision response condition is met or not according to the accumulated quantity corresponding to the current calculation moment after the accumulated quantity is obtained through each calculation; and the second determining module is used for determining the response time required by the driver for making the target decision according to the current calculation moment if the decision response condition is met.

In one embodiment, the calculating module is specifically configured to obtain, at intervals of a preset time, an accumulated amount of the target evidence information obtained by calculating at a previous calculating time, determine, according to road information at a current time in the target dangerous state traffic scene, a first evidence information amount supporting the target decision currently and a second evidence information amount opposing the target decision, and obtain, according to the accumulated amount, the first evidence information amount and the second evidence information amount corresponding to the previous calculating time, the accumulated amount of the target evidence information at the current calculating time.

In one embodiment, the calculating module is specifically configured to obtain the noise amount at the current calculating time based on the wiener process; and acquiring the accumulated quantity of the target evidence information at the current calculation moment according to the accumulated quantity, the first evidence information quantity, the second evidence information quantity and the noise quantity corresponding to the last calculation moment.

In one embodiment, the first determining module is specifically configured to determine whether a decision response condition is met according to an accumulated amount, a driver parameter, and a vehicle parameter corresponding to a current calculation time; wherein the driver parameter is related to information of a driver in the target dangerous state traffic scene and the vehicle parameter is related to a state of a vehicle in the target dangerous state traffic scene.

In one embodiment, the first determining module is specifically configured to determine whether a sum of the accumulated amount corresponding to the current calculation time and the driver parameter is greater than or equal to a vehicle parameter, and if the sum is greater than or equal to the vehicle parameter, determine that the decision response condition is satisfied.

In one embodiment, the target decisions are decisions in a decision pool, and the device further comprises a third determining module, configured to determine a response duration corresponding to each target decision in the decision pool; and sequencing a plurality of target decisions included in the decision pool according to the response time length corresponding to each target decision, and determining the response process of the driver in the target dangerous state traffic scene.

In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the steps of the method according to any of the first aspects above when the processor executes the computer program.

In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method of any of the first aspects above.

In a fifth aspect, the present application also provides a computer program product. The computer program product comprising a computer program which, when executed by a processor, implements the steps of the method according to any of the first aspects above.

According to the response prediction method, the response prediction device, the computer equipment, the storage medium and the program product, for the target dangerous state traffic scene, the accumulated amount of the target evidence information is calculated from the preset stage of the target dangerous state traffic scene every other preset time period, wherein the target evidence information is evidence information on which a target decision is made, after the accumulated amount is obtained through calculation, whether a decision response condition is met or not is determined according to the accumulated amount corresponding to the current calculation time, if the decision response condition is met, the response time required by a driver for making the target decision is determined according to the current calculation time, and a big data driving deep learning model in the traditional technology is not needed in the process, so that the difficulty of predicting the decision response of the driver is reduced.

Drawings

FIG. 1 is a flow chart of a response prediction method in one embodiment;

FIG. 2 is a schematic diagram of an evidence accumulation model in one embodiment;

FIG. 3 is a diagram of another response prediction method in one embodiment;

FIG. 4 is a schematic diagram of a Bayesian framework in one embodiment;

FIG. 5 is a block diagram of a response predicting apparatus according to an embodiment;

fig. 6 is an internal structural diagram of a computer device in one embodiment.

Detailed Description

The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.

The dangerous state traffic scene refers to a scene working condition that a dangerous source suddenly appears or a traffic participant who normally runs is converted into the dangerous source in a road traffic environment, and the normal running of a vehicle is influenced. Dangerous traffic scenarios typically require active responses (e.g., throttle release, braking, steering, etc.) by the driver to avoid risk and collision avoidance.

Vehicles in the road traffic environment can be divided into four stages from normal running to collision, the vehicles in the first stage are normally driven, the vehicles face the complex traffic environment, the vehicles continuously interact with the environment through sensors or drivers to realize environment monitoring, and the normal running of the vehicles is ensured. When dangerous working conditions occur, the vehicle enters a second stage, a vehicle sensor or a driver senses a dangerous source, and the vehicle active safety system intervenes in the second stage, wherein the period usually lasts from 6 seconds before collision to 3 seconds before collision, and the probability of the active safety system successfully intervening in and avoiding collision is high. Once collision is not avoided, the vehicle enters a critical dangerous state in a third stage, namely enters a dangerous state traffic scene, a driver takes an active response action (also called active action) to intervene in the maximum capability of the vehicle, meanwhile, an intelligent restraint system is started, the vehicle intervention can successfully avoid the collision and can be unsuccessful but reduce the collision contact strength, and the decision action and the response time of the driver in the third stage directly influence the state of the driver at the moment of collision zero, so that the damage risk is influenced. Finally, the vehicle enters a fourth-stage collision stage, and targeted protection is provided for passengers in the vehicle under the action of the active and passive safety integrated protection equipment of the vehicle.

In the dangerous state traffic scene, if a driver does not take active control over the vehicle, the dangerous state traffic scene can evolve into serious traffic accidents, and the safety of traffic participants is threatened. The driver is required to actively respond during the vehicle entering the third-stage critical dangerous state to avoid the collision or reduce the collision intensity. The active response mainly focuses on decision types and response durations (i.e. response time) of different types, and different decision processes in time history form the active response history of the driver. The reaction time of a driver in a dangerous traffic scene is generally defined as the time from the occurrence of a dangerous source or dangerous information (such as the start of braking of a leading vehicle in a following scene) to the decision of making an active response by the driver and taking collision avoidance actions (such as throttle release, braking, steering and the like).

In the conventional technology, the research of a driver decision type and a reaction time prediction algorithm is concentrated under the conventional cruising tasks (such as lane changing, steering, car following, fatigue detection, and the identification of a second task of a driver on an automatic driving vehicle) of a vehicle, but the research of the driver decision type inference and the reaction time prediction in a dangerous state traffic scene is lacking. The traditional driver decision type inference and reaction time prediction model is divided into two technical schemes of a decomposition type and an end-to-end type, wherein the decomposition type scheme is further divided into three modules of perception, control and decision, the division is clear, the interpretability is strong, but the system complexity is high, and the calculated amount is large; the end-to-end scheme simulates anthropomorphic driving behavior decision based on a machine learning model or a deep learning model, has small calculated amount and low system complexity, but has high algorithm requirement, low safety and poor interpretability and reliability, and the machine learning model and the deep learning model need a large amount of data to drive, which is a great limitation of dangerous traffic scene research.

In addition, the decision type inference and the reaction time prediction of the driver are carried out through machine learning models such as decision trees, fuzzy logic control and the like or deep learning models, and most of the decision type inference and the reaction time prediction are based on a data-driven black box process, and although the decision tree has good fitting precision, the logic relationship in the decision process cannot be explained, and the reaction time distribution and the change form of the driver are difficult to explain.

In addition, the driver decision type estimation and the reaction Time prediction are also performed with reference to physical quantities describing the degree of scene urgency such as TTC (Time to Collision ), THW (Time to Headway) and the like. However, a person-vehicle-road constitutes an abnormally complex road traffic system, personnel are core elements of the system, a driver can sense key elements in a dangerous state traffic scene, understand potential hazards in the traffic scene, predict motion forms of dangerous sources according to priori knowledge and scene characteristics, make collision avoidance decisions and finally take collision avoidance operation behaviors, and therefore in a real dangerous state traffic scene, the driver's sense decisions and active responses are simultaneously influenced by various factors of the driver (age, style, fatigue state and the like), vehicles (dynamics state, control mode and the like), traffic scenes (accident types, emergency degree, dangerous source types and the like). Therefore, the decision type and reaction time predicted with TTC and THW have a problem of low prediction accuracy.

Based on the above, it is necessary to integrate multiparty factors to predict the decision type and the response time of the driver in the dangerous traffic scene.

The response time is an important random variable affecting the dangerous state development, and in the field of psychology or cognition, the SDT (Signal Detection Theory ) considers that in the process of receiving stimulation information and making a decision, the receptor can generate noise in the process of sensing stimulation signals and random signal processing of human brain neurons, so that uncertainty of human decision behaviors is caused. In addition, when the driver processes most road conflicts, the driver does not need to mobilize a high-level cognitive function area of the brain to carry out long-time deliberate consideration, and only needs to quickly respond according to perceived information and priori knowledge.

In the field of theoretical neuroscience, decision models under different scales are established for decision processes of human brain under external stimulus. The NSP (Neuron Spiking Model, neuron pulse model) of the mesoscopic layer can simulate the physiological process that the accumulated potential reaches a threshold value to generate release after the neuron receives the stimulation signal; RRM (Reduced Rate Model, simplified burst firing rate model) of mesoscopic layer is a simplified multi-neuron burst firing model built based on dynamic stability characteristics of functional neuron groups; DDM (Drift Diffusion Model, drift diffusion decision model) and late (Linear Approach to Threshold with Ergodic Rate, linear threshold decision of rate of traversal) models of the macroscopic behavior layer can characterize behavior characteristics exhibited by humans in the behavioral cognitive layer. The DDM and LATER models of the behavior layer can then characterize the relationship between human decision results, response time and stimulus signal variation. However, compared to a single static decision task in traditional cognitive behavioral research, driver decisions in dynamic traffic scenarios are affected by numerous potential factors.

Most of SDT-based models focus on constructing threshold model detection and reaction thresholds, namely, signals of interest are reacted when the signals exceed the thresholds at a certain moment, and the time history of dangerous traffic scene development is ignored. Evidence accumulation processes emphasize the evolution and accumulation of different signals with noise over time. Evidence accumulation models, such as DDM or late models, have been demonstrated to be able to fit cognitive decision tasks of varying task and complexity, and studies have demonstrated the corresponding neurophysiologic mechanisms. But has no application in real traffic scenarios.

Therefore, three influencing factors of human-vehicle-road can be fused into a model by means of an evidence accumulation process to accurately predict the decision type and the reaction time of a driver, and the specific process is as follows.

In one embodiment, as shown in fig. 1, a response prediction method is provided, and the method is applied to a terminal for illustration, and includes the following steps:

step 101, for a target dangerous state traffic scene, starting from a preset stage of the target dangerous state traffic scene, calculating an accumulated amount of target evidence information every preset time, wherein the target evidence information is evidence information on which a target decision is made.

The target dangerous state traffic scene refers to the third stage, and the driver is in active response and collision avoidance stage. Active response refers to the process of the driver making a decision and performing the decision based on dangerous conditions, e.g., the driver seeing the lead vehicle to start braking, the driver deciding to turn right after a short thought, and steering the steering wheel to the right.

The preset stage may be a start time of a target dangerous state traffic scene input in advance.

The target decision includes at least one of throttle release, braking, steering, and the like. The evidence information refers to visual change stimulus (visual angle θ, visual angle change rate or a combination thereof), visual prompt information such as a brake lamp, three-dimensional space information such as distance and speed, and the like.

Optionally, the accumulated amount of the target evidence information at the current moment is calculated according to the accumulated amount of the target evidence information calculated at the previous calculation moment, the first evidence information amount supporting the target decision and the second evidence information amount opposite to the target decision.

Step 102, after the accumulated quantity is obtained through each calculation, determining whether a decision response condition is met according to the accumulated quantity corresponding to the current calculation time.

Optionally, determining whether the accumulated amount corresponding to the current calculation time is greater than or equal to a decision threshold, and if so, determining that a decision response condition is met.

In another alternative embodiment, it is determined whether the sum of the accumulated amount corresponding to the current calculation time and the driver parameter is greater than or equal to a vehicle parameter, and if the sum is greater than or equal to the vehicle parameter, it is determined that the decision response condition is satisfied, where the driver parameter is related to information of the driver in the target dangerous state traffic scene, and the vehicle parameter is related to a state of the vehicle in the target dangerous state traffic scene.

And step 103, if the decision response condition is met, determining the response time required by the driver to make the target decision according to the current calculation time.

Optionally, the accumulated amount of the target evidence information for making the target decision is calculated every preset time, and whether the accumulated amount meets the decision response condition is judged, if not, the accumulated amount of the target evidence information is continuously calculated at the next calculation time until the accumulated amount calculated at a certain calculation time meets the decision response condition, and the time is taken as the response time required by the driver for making the target decision, namely the response time of the driver.

For example, the accumulated amount of target evidence information is calculated every 0.01 seconds, and when the calculated accumulated amount satisfies the decision condition at 1.15 seconds, the response time period required for the driver to make the target decision is 1.15 seconds.

In summary, for the target dangerous state traffic scene, from the preset stage of the target dangerous state traffic scene, the accumulated amount of the target evidence information is calculated at intervals of preset time, wherein the target evidence information is evidence information according to which a target decision is made, after the accumulated amount is calculated each time, whether a decision response condition is met or not is determined according to the accumulated amount corresponding to the current calculation time, if the decision response condition is met, the response time required by a driver for making a target decision is determined according to the current calculation time, and in the process, a deep learning model is not required to be driven by big data in the traditional technology, so that the difficulty of predicting the decision response of the driver is reduced.

The road information is road environment information, and includes dangerous scene types (such as rear-end collisions, traffic intersection collisions, oncoming vehicles, etc.), dangerous source types (such as cars, trucks, pedestrians, riders, etc.), collision angles, and collision speeds.

Optionally, the terminal may query, from the decision database, first evidence information supporting the target decision and second evidence information opposing the target decision according to the road information, and count the number of the first evidence information and the number of the second evidence information, respectively, to obtain a first evidence information amount and a second evidence information amount. The decision database stores a plurality of groups of corresponding relations of support and anti-evidence information corresponding to the road information and the target decision information.

The above embodiments can be expressed by the following mathematical formulas:

E(t)＝E(t-Δt)+β(t)Δt+ρW(t)Δt (1)

wherein E (t) represents the accumulated amount at the time t, E (t- Δt) represents the accumulated amount corresponding to the last calculation time, and Δt represents the preset time period. S (t) represents a first evidence information amount supporting the target decision at the time t, O (t) represents a second evidence information amount opposing the target decision at the time t, and R representsRoad information, g (y _i ) Representing information corresponding to the ith item included in the road information, e.g., g (y ₁ ) Represents the dangerous scene type, g (y ₂ ) Represents the type of dangerous source, g (y ₃ ) Represents the angle of impact, g (y ₄ ) The difference beta (t) between S (t) and O (t) representing the collision speed is related to road information and can also be related to the interaction process (stimulus intensity and evidence quality) of dangerous traffic scenes. W (t) represents the Wiener Process or Brownian Motion, i.e. the amount of noise at time t, ρ represents the scaling factor of the Wiener Process, and the distribution of W (t) -W (t- Δt) in the Wiener Process is independent of time t, desirably 0, and can be represented by equation (3).

In addition, in the case of the optical fiber,represents the amount of noise evidence accumulation within a preset duration of deltat,and represents the accumulation amount of the evidence information in the deltat preset time period.

Wherein, the information of the driver comprises age, sex, driving style, driving experience, fatigue state and distraction state; the states of the vehicle include position, heading angle, speed, acceleration, control state (automatic driving, manual driving, take over critical state, etc.).

Alternatively, the above embodiments may be expressed by the following mathematical formulas:

S ₀ +E(t)≥α(t) (4)

E ₀ (t)＝α(t)-S ₀ (5)

wherein, the formula (4) is a decision response condition, S ₀ Representing driver parameters, i.e. initial values of evidence accumulation, relating to different individual/group driver characteristics, i.e. information D relating to the driver, f (x) _i ) Information corresponding to the ith item included in the information indicating the driver, e.g. f (x ₁ ) Represents age, f (x) ₂ ) Represents sex, f (x) ₃ ) Represents driving style, f (x ₄ ) Represents driving experience, f (x) ₅ ) Indicating fatigue state, f (x) ₆ ) Represent the state of distraction, S ₀ Suppose that normal distribution N (α, σ) is obeyed. Alpha (t) represents a vehicle parameter at time t, i.e. a decision threshold at time t, which is related to the state V of the vehicle and to the time pressure t, G (V, t) represents an objective function, h (z) _i ) Information representing the correspondence of the ith item included in the state of the vehicle, e.g. h (z ₁ ) Represents the position, h (z ₂ ) Represents heading angle, h (z ₃ ) Indicating the speed, h (z ₄ ) Indicating acceleration, h (z ₅ ) Indicating the control state. E (E) ₀ And (t) represents an evidence accumulation threshold which is required to be accumulated when the decision threshold is reached at the moment t, and is determined by the difference value between the decision threshold at the moment t and the initial value.

The above formulas (1) - (7) can be regarded as the constructed evidence accumulation model. As shown in FIG. 2, a schematic diagram of an evidence accumulation model is provided, it can be seen that the initial value S in the evidence accumulation model ₀ Together with the decision threshold (response boundary) a determines the evidence accumulation threshold E required for making a decision reaction ₀ (t) either the initial value increase or the response boundary decrease is beneficial to react with little evidence to shorten the reaction time.

Initial value S ₀ The method has the advantages that the method is related to information of drivers, driving experience is rich, dangerous scenes can be predicted and responded in advance, the drivers with conserved driving styles can make decisions to respond as soon as possible to keep safer distances, therefore, initial values of different individuals or different groups can be obtained through six variables in the information of the drivers, the larger the initial values are, the drivers can respond faster to dangerous traffic scenes with the same stimulus, and normal distribution represents random uncertainties of different individual distribution forms and individual responses in the groups.

The evidence accumulation rate v corresponds to the intensity of the stimulus signal in the dangerous state traffic scene, and the more urgent the dangerous state traffic scene is, the stronger the stimulus signal (such as visual change stimulus, distance, speed and the like) is, the larger the evidence accumulation rate is, and the faster the evidence accumulation is. In addition, the evidence accumulation rate v is also related to the road information, and the larger the first evidence information quantity supporting the target decision is obtained according to the road information, the smaller the second evidence information quantity opposing the target decision is, the larger the evidence accumulation rate is.

The response boundary alpha is related to the state of the vehicle and the time pressure, the initial response boundary of the dangerous state traffic scene is represented as a constant boundary, the driver can deteriorate the dangerous state traffic scene if not taking an active response in the initial stage, and is forced to make a response decision, the response boundary is represented as a decrease (collapse boundary alpha') with time in the evidence accumulation model, namely, the evidence quantity required for triggering the decision becomes smaller with the increase of the decision time, and the response time is macroscopically represented as an insufficient decision with longer response time. The state of the vehicle such as the control mode of the vehicle (automatic driving, manual driving, taking over critical state, etc.) or whether there is a brake light, an in-vehicle prompt tone, etc. can also change the response boundary, thereby helping the driver to make a quick decision to respond. E represents the cumulative amount of evidence information accumulated to make the target decision, i.e., the cumulative posterior probability of the evidence of making the decision.

Optionally, the decision pool includes target decisions such as throttle release, braking and steering, for each target decision, a response time corresponding to each target decision can be obtained according to the method, for example, the response time corresponding to the throttle release is 0.70 seconds, the response time corresponding to the braking is 0.92 seconds, the response time corresponding to the steering is 1.13 seconds, and then the response history of the driver in the target dangerous state traffic scene is that the driver performs the throttle release operation at 0.70 seconds, the brake operation at 0.90 seconds and the steering operation at 1.13 seconds.

In summary, as shown in fig. 3, another response prediction method is provided, which includes the following steps:

step 301, for the target dangerous state traffic scene, starting from a preset stage of the target dangerous state traffic scene, acquiring an accumulated amount of the target evidence information calculated at the previous calculation time at intervals of a preset time period, and determining a first evidence information amount supporting the target decision currently and a second evidence information amount opposing the target decision according to the road information at the current time in the target dangerous state traffic scene.

Step 302, acquiring the noise amount at the current calculation time based on the wiener process; and acquiring the accumulated quantity of the target evidence information at the current calculation moment according to the accumulated quantity, the first evidence information quantity, the second evidence information quantity and the noise quantity corresponding to the last calculation moment.

Step 303, determining whether the sum of the accumulated amount corresponding to the current calculation time and the driver parameter is greater than or equal to the vehicle parameter; if the vehicle parameter is greater than or equal to the vehicle parameter, determining that a decision response condition is met, wherein the driver parameter is related to the information of a driver in the target dangerous state traffic scene, the vehicle parameter is related to the state of a vehicle in the target dangerous state traffic scene, the vehicle parameter is calculated according to the state of the vehicle in the target dangerous state traffic scene and an objective function, and the objective function is a function with gradually reduced function value along with time.

Step 304, if the sum of the accumulated amount corresponding to the current calculation time and the driver parameter meets the decision response condition, determining the response time required by the driver to make the target decision according to the current calculation time.

Step 305, the target decisions are decisions in the decision pool, and the response time length corresponding to each target decision in the decision pool is determined; and sequencing a plurality of target decisions included in the decision pool according to the response time length corresponding to each target decision, and determining the response process of the driver in the target dangerous state traffic scene.

The application also has the following advantages:

(1) With the help of the current state of leading edge research in the fields of cognitive psychology, cognitive neuroscience and the like, an evidence accumulation model is established based on an evidence accumulation process, and the evidence accumulation model is a driver active response prediction model capable of reflecting a driver perception decision mechanism. The model can comprehensively consider the driver factors (corresponding to the information of the driver), the road environment factors (corresponding to the road information) and the vehicle state factors (corresponding to the state of the vehicle) to infer the decision type and the response time of the driver, can explain the influence of different factors and parameter changes on the perception decision of the driver, improves the interpretation and reliability of an algorithm, and realizes the white-box prediction of the active response of the driver.

(2) The model can obtain relevant model parameters by using a proper amount of data, so that a large amount of data required by a data driving method is avoided, and the model has more applicability to dangerous traffic scenes with larger data acquisition difficulty.

(3) The model can construct parameterized models for representing different groups and individuals through relevant characterization of the factors of the drivers so as to reflect individual variability of the drivers, and meanwhile, the response process (such as braking moment, steering moment and the like) of the drivers in dangerous state traffic scenes can be deduced and predicted, so that personalized driving assistance is provided for different drivers, and the understandability in the man-machine interaction process is improved.

In addition, the model verification is carried out on the constructed evidence accumulation model, and the model verification is concretely as follows: for the accuracy of the model, two verification methods are provided. According to the first method, data of driving simulator tests and test data taking real traffic video images as stimulus are respectively obtained through volunteers, and data information of driving response and decision behaviors of the drivers under dangerous working conditions is obtained. The accuracy and effectiveness of the established model are measured by taking simulator tests and video stimulus tests as true values and comparing the true values with the response time errors of model predictions and the accuracy of decision type deduction, and attention should be paid to distinguishing the data training set from the verification set at the moment. The reason for using real traffic video images as stimulus is to strive to restore the stimulus level of real traffic scenes in laboratory environments. And secondly, carrying out decision type inference and reaction time prediction of a driver by using published test data, comparing the result with an existing model established by corresponding test data, and verifying the advancement of the model of the invention.

The above evidence accumulation model is constructed based on a bayesian framework, and as shown in fig. 4, a bayesian framework schematic diagram is provided, and the idea of the bayesian framework is as follows:

The bayesian framework is a proper mathematical framework for making decisions based on uncertain data, and bayesian decision refers to that a test result before the decision can influence the decision process, namely, the test result has the comprehensive influence of priori knowledge and posterior knowledge. A decision must be made to sample sequentially and accumulate evidence gradually until the evidence is sufficiently convincing. The gradual sampling and evidence accumulation process is a process of continuously updating the result by using the posterior in the Bayesian framework.

If there is a hypothesis H, and an event E is observed, it is available according to Bayesian law:

P(E)P(H|E)＝P(H)P(E|H) (8)

the rearrangement may result in the estimated probability being updated from the old value P (H) (a priori) to the new value P (h|e) (a posteriori) after the event E is observed:

to the left of the equal sign is a posterior probability (possesrondds), and to the right of the equal sign is a prior probability (priords) and likelihood ratio (likelihoodratio). The likelihood ratio occurs continuously in a multiplicative manner at the time of repeated sampling, and thus after converting the above equation into a logarithmic form:

log(posteriorodds)＝log(priorodds)+log(likelihoodratio) (10)

the form of multiplication translates into addition, so that the likelihood ratio supporting evidence of an option after each sample is increasing, the logarithmic probability supporting the correct hypothesis is increasing to represent the rate of the amount of information obtained each time. It can be seen from the Bayesian framework that what is needed is a decision signal S that represents a logarithmic probability, the initial value of the signal being S ₀ Representing a logarithmic prior probability; when the number of times occurs, S starts to rise gradually at a rate v, which is determined by a log-likelihood ratio, and the information representing the sampling provides supporting evidence for a certain option; the final logarithmic probability will reach the decision making level S _T . It can be seen that the decision process derived from the bayesian framework is consistent with the evidence accumulation framework of the reaction time response process.

The modeling process of the driver factor D, the road environment factor R, and the vehicle state factor V is as follows:

(1) Acquisition of data

The perception decision of the driver influences the interaction process under the dangerous state working condition, directly influences whether collision, collision time and collision angle can be avoided, and further influences the damage risk of traffic participants in the collision process. Part of information in the driver factors can be obtained through a driver behavior questionnaire or inquiry through (such as age, sex, driving style and driving experience), but the fatigue state and distraction state of the driver, road environment factors and vehicle state factors are closely related to the dangerous state scene, and specific analysis is needed through the dangerous state scene process information and the acquired physiological signals. The dangerous state scene data acquisition path is mainly obtained by means of volunteer experiments. Recording volunteer related information for driver factor analysis of perceived decision influence through driver behavior questionnaire survey before test and test task design and state evaluation; the road environment parameter data and the vehicle state data of the vehicle in the test process are recorded in real time, and the influence factors of the perception decision of the driver can be quantitatively analyzed from different layers.

The volunteer test is usually performed on a driving simulator, and different dangerous situations and perception decision tasks can be designed. Although the real vehicle test can provide the truest driving experience for the tested person, the real vehicle test is in best line with the tested driving habit. However, dangerous traffic scenes cannot be designed manually when the vehicle is driven on a real road, the efficiency is low, the vehicle is not provided with repeatability, and serious dangerous state scene tests cannot be carried out in consideration of safety. The related dangerous state scene data can also be analyzed by a natural driving data set or an accident analysis data set, and accident scene information is reappeared by combining an accident investigation report to complete statistics of driver perception decision influence factors. However, due to the limitation of information recording and the protection of the privacy of the inspector, it is difficult to obtain all the required level information by the method, and sufficient data support cannot be provided for the construction of the model.

(2) Building a model

The modeling is based on the premise that enough data is supported, and the basic theoretical method is controlled variable method research. Next, description will be made taking a driver factor as an example. The driver factor may subdivide age x ₁ Sex x ₂ Driving style x ₃ Driving experience x ₄ Fatigue state x ₅ State of distraction x ₆ Six aspects, where age may be approximated as a continuous variable and gender, driving style, driving experience, fatigue status, distraction status as discrete variables.

The classification of the variables is shown in Table 1. Further grading may be performed according to the situation of investigation and the degree of fineness, e.g. the distraction status may be graded differently according to different secondary tasks. The grading of each factor can be specifically determined through the questionnaire score and the state assessment of the driver, and the research basis is provided. And respectively researching and analyzing specific influences of each parameter on the factors of the driver by using a control variable method. The method comprises the steps of firstly determining positive and negative influences of various factors on the response time of a driver, determining a function trend line and further determining a function coefficient value.

TABLE 1

Variable name	Variable symbol	Variable classification
			Sex (sex)	x ₂	Female: 0, man 1
Driving style	x ₃	Conservation: 0, normal: 1, aggressive 2
			Driving experience	x ₄	Is not rich: 0, enrich: 1
Fatigue state	x ₅	Wakening: 0, fatigue: 1
			State of distraction	x ₆	Concentrate on: 0, distraction: 1

First study age x ₁ The influence on the factor D of the driver, the data statistics result and the accident data investigation report show that the accident data volume shows a trend of increasing and then decreasing along with the age of the driver, so that the factor D and the factor x can be approximately considered as ₁ The approximation of the relationship between them presents a quadratic function. Using the result of the known data set to calculate the parameters a, b, c of the quadratic function to obtain D and x ₁ Is a definite relation of (c).

In fatigue state x ₅ To illustrate the effect of discrete variables on driver factor D. Research and data statistics show that a driver in a wakeful state can sense dangers in a traffic scene and make decision judgment more quickly. The influence degree of the discrete variable on the factor is expressed by a coefficient, the size of the coefficient expresses the influence degree of the discrete variable on the factor, the positive and negative of the coefficient expresses the influence effect (the reaction time is prolonged or shortened) of the discrete variable on the factor, and the intercept expresses the drift of the overall influence of the discrete variable on the factor. Using the result of the known data set to calculate the influence factor D, f of the variable to obtain D and x ₅ Is a definite relation of (c).

D＝dx ₅ +f

Similarly, the driver factor D and the sex x can be obtained by a similar method ₂ Driving style x ₃ Driving experience x ₄ Fatigue state x ₅ State of distraction x ₆ Finally, the functional relation between D and age, sex, driving style, driving experience, fatigue state and distraction state can be quantized as follows:

the quantitative influence relation between the road environment factor R and the vehicle state factor V and the related variables thereof can be deduced through the above processes and steps, so that modeling of the driver perception decision influence factor is finally completed, and the influence factors and effects thereof are respectively represented from three layers:

(3) Model verification

The established theoretical model is to be verified experimentally, and two methods for verifying the established theoretical model are provided. In the first method, verification is performed by developing a data set collected by a volunteer behavior test, for example, in order to study the influence of the age or driving style of a driver on the perception decision characteristics of the driver, a typical dangerous state traffic scene can be designed in a targeted manner, a representative tested dangerous state scene driver risk perception test is screened, and model effect verification and model parameter revision are performed according to the test recorded result. In order to verify the application effect of the model in the real traffic scene, the revised model can be applied to the natural driving data set for validity verification, and the evaluation model can be verified in a targeted manner according to the acquired small amount of natural driving data types. It should be noted that, whether it be a volunteer behavioral trial data set or a natural driving data set, the training set and the verification set of the model need to be effectively divided, and the accuracy of the established model of the behavior mechanism of the driver is measured according to the accuracy of the model verified on the verification set.

It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.

Based on the same inventive concept, the embodiment of the application also provides a response prediction device for realizing the response prediction method. The implementation of the solution provided by the apparatus is similar to the implementation described in the above method, so the specific limitation in the embodiments of one or more response prediction apparatus provided below may refer to the limitation of the response prediction method hereinabove, and will not be repeated here.

In one embodiment, as shown in fig. 5, there is provided a response predicting apparatus, the response predicting apparatus 500 including: a calculation module 501, a first determination module 502 and a second determination module 503, wherein:

the calculation module 501 calculates the accumulation amount of target evidence information, which is evidence information on which a target decision is made, from a preset stage of the target dangerous state traffic scene every preset time period;

the first determining module 502 determines whether the decision response condition is satisfied according to the accumulated quantity corresponding to the current calculation time after the accumulated quantity is obtained by each calculation;

the second determining module 503 determines a response time period required by the driver to make the target decision according to the current calculation time if the decision response condition is satisfied.

In one embodiment, the calculating module 501 is specifically configured to obtain, at intervals of a preset time, an accumulated amount of the target evidence information obtained by calculating at a previous calculating time, determine, according to the road information at the current time in the target dangerous state traffic scene, a first evidence information amount supporting the target decision currently and a second evidence information amount opposing the target decision, and obtain, according to the accumulated amount, the first evidence information amount and the second evidence information amount corresponding to the previous calculating time, the accumulated amount of the target evidence information at the current calculating time.

In one embodiment, the calculating module 501 is specifically configured to obtain the noise amount at the current calculating time based on the wiener process; and acquiring the accumulated quantity of the target evidence information at the current calculation moment according to the accumulated quantity, the first evidence information quantity, the second evidence information quantity and the noise quantity corresponding to the last calculation moment.

In one embodiment, the first determining module 502 is specifically configured to determine whether a decision response condition is met according to an accumulated amount, a driver parameter, and a vehicle parameter corresponding to a current calculation time; wherein the driver parameter is related to information of a driver in the target dangerous state traffic scene and the vehicle parameter is related to a state of a vehicle in the target dangerous state traffic scene.

In one embodiment, the first determining module 502 is specifically configured to determine whether a sum of the accumulated amount corresponding to the current calculation time and the driver parameter is greater than or equal to a vehicle parameter, and if the sum is greater than or equal to the vehicle parameter, determine that the decision response condition is satisfied.

The respective modules in the above-described response predicting means may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In one embodiment, a computer device is provided, which may be a terminal, and the internal structure of which may be as shown in fig. 6. The computer device includes a processor, a memory, an input/output interface, a communication interface, a display unit, and an input means. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a response prediction method. The display unit of the computer device is used for forming a visual picture, and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be a key, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.

It will be appreciated by those skilled in the art that the structure shown in FIG. 6 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.

In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of any of the method embodiments described above when the computer program is executed.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when executed by a processor, implements the steps of any of the method embodiments described above.

In an embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of any of the method embodiments described above.

It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the embodiments provided herein may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processor referred to in the embodiments provided in the present application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, or the like, but is not limited thereto.

The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.

The foregoing examples illustrate only a few embodiments of the application and are described in detail herein without thereby limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.

Claims

1. A method of response prediction, the method comprising:

for a target dangerous state traffic scene, starting from a preset stage of the target dangerous state traffic scene, calculating the accumulation amount of target evidence information at intervals of preset time, wherein the target evidence information is evidence information on which a target decision is made;

after the accumulated quantity is obtained through each calculation, determining whether a decision response condition is met or not according to the accumulated quantity corresponding to the current calculation moment;

And if the decision response condition is met, determining the response time required by the driver to make the target decision according to the current calculation time.

2. The method according to claim 1, wherein calculating the accumulated amount of the target evidence information every predetermined time period includes:

and acquiring the accumulated quantity of the target evidence information obtained by calculation at the last calculation time every preset time, determining the first evidence information quantity supporting the target decision currently and the second evidence information quantity opposite to the target decision currently according to the road information at the current time in the target dangerous state traffic scene, and acquiring the accumulated quantity of the target evidence information at the current calculation time according to the accumulated quantity corresponding to the last calculation time, the first evidence information quantity and the second evidence information quantity.

3. The method according to claim 2, wherein the obtaining the accumulated amount of the target evidence information at the current calculation time from the accumulated amount corresponding to the previous calculation time, the first evidence information amount, and the second evidence information amount includes:

acquiring the noise amount at the current calculation time based on the wiener process;

And acquiring the accumulated quantity of the target evidence information at the current calculation time according to the accumulated quantity corresponding to the last calculation time, the first evidence information quantity, the second evidence information quantity and the noise quantity.

4. A method according to any one of claims 1 to 3, wherein determining whether the decision response condition is satisfied according to the accumulated amount corresponding to the current calculation time comprises:

determining whether the decision response condition is met according to the accumulated quantity, the driver parameter and the vehicle parameter corresponding to the current calculation moment;

wherein the driver parameter is related to information of a driver in the target dangerous state traffic scene, and the vehicle parameter is related to a state of a vehicle in the target dangerous state traffic scene.

5. The method of claim 4, wherein determining whether the decision response condition is satisfied based on the accumulated amount corresponding to the current calculation time, the driver parameter, and the vehicle parameter comprises:

determining whether the sum of the accumulated amount corresponding to the current calculation time and the driver parameter is greater than or equal to the vehicle parameter;

and if the vehicle parameter is greater than or equal to the vehicle parameter, determining that the decision response condition is met.

6. The method of claim 4, wherein the vehicle parameters are calculated from a state of a vehicle in the target dangerous state traffic scenario and an objective function, the objective function being a function of a function value that gradually decreases over time.

7. The method according to any one of claims 1 to 6, wherein the target decision is a decision in a decision pool, the method further comprising:

determining the response time length corresponding to each target decision in the decision pool;

and sequencing a plurality of target decisions included in the decision pool according to the response time length corresponding to each target decision, and determining the response process of the driver in the target dangerous state traffic scene.

8. A response predicting apparatus, the apparatus comprising:

the calculation module is used for calculating the accumulation of target evidence information for the target dangerous state traffic scene from the preset stage of the target dangerous state traffic scene at intervals of preset time, wherein the target evidence information is evidence information on which a target decision is made;

the first determining module is used for determining whether the decision response condition is met or not according to the accumulated quantity corresponding to the current calculation moment after the accumulated quantity is obtained through each calculation;

And the second determining module is used for determining the response time required by the driver for making the target decision according to the current calculation moment if the decision response condition is met.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 7 when the computer program is executed.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 7.