Driver takeover efficiency evaluation method for whole process of automatic driving takeover time sequence

Abstract

The invention discloses a driver takeover efficiency evaluation method for an automatic driving takeover time sequence whole process. The method comprises the following steps: dividing a takeover time sequence of a driver state behavior into a non-driving related behavior and state stage in an NDRT period and a takeover stage after TOR; setting a subjective scale score and objective physiological and behavior indexes to evaluate the comprehensive state of the driver in the non-driving related behavior and state stage during the NDRT period; the state behaviors in the takeover stage after TOR are divided into TOR man-machine interaction, takeover switching and control and workload penetrating through the whole takeover stage, and each state behavior in the takeover stage after TOR is evaluated. According to the method, the behavior state characteristics of the driver taking over each time sequence can be effectively represented, and technical support is provided for improving the experience feeling and the compliance of personnel.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving, and in particular to a method for evaluating driver takeover effectiveness for the entire process of autonomous driving takeover sequence. Background Art

[0002] The Society of Automotive Engineers defines six stages (0 to 5) based on the driving automation level of self-driving cars, where L0 means manual driving without any automation and L5 means fully autonomous driving. Conditional autonomous driving in the market is also called intelligent driving, and L3 is the main battlefield for current mainstream car manufacturers. Through continuous accumulation and continuous improvement in supervision, technology and commercialization, the implementation of high-level autonomous driving is about to enter the fast lane of development.

[0003] At present, there are many technical solutions for analyzing or evaluating the driver's takeover during the process of autonomous driving. For example, a human reliability analysis method for autonomous driving takeover combines the system theory process analysis method and the human reliability dynamic analysis method for qualitative analysis, and combines the Bayesian network and the success likelihood index method for quantitative calculation to obtain the evaluation index of the behavior formation factor that causes human errors in the takeover process, which can effectively promote risk assessment and management and reduce the probability of human errors. In addition, there is a method and system for predicting the performance of autonomous driving takeover based on human-computer interaction. This method collects human-computer interaction data for autonomous driving takeover, designs a variety of different traffic scenarios in the simulator, and collects vehicle and driver data at the time of takeover in different traffic scenarios, enriching the traffic scenarios where takeover occurs, so that the subsequent prediction model established by these data can better adapt to different traffic scenarios and obtain more accurate prediction results.

[0004] However, the above technical solutions are only aimed at human reliability analysis or automatic driving takeover performance prediction, and lack evaluation or analysis of the entire takeover process. Therefore, there is an urgent need for a driver takeover effectiveness evaluation method for the entire automatic driving takeover sequence, which can effectively characterize the driver's behavioral state characteristics at each takeover sequence and provide technical support for improving personnel experience and compliance. Summary of the invention

[0005] In view of this, the purpose of the present invention is to overcome the defects in the prior art and provide a driver takeover effectiveness evaluation method for the entire process of automatic driving takeover sequence, which can effectively characterize the behavioral state characteristics of the driver at each takeover sequence, and provide technical support for improving personnel experience and compliance.

[0006] The driver takeover efficiency evaluation method for the entire process of automatic driving takeover sequence of the present invention includes:

[0007] The takeover sequence of driver state behavior is divided into the non-driving related behavior and state stage during NDRT and the takeover stage after TOR;

[0008] Subjective scale scores and objective physiological and behavioral indicators were set to evaluate the comprehensive status of the driver during the non-driving related behavior and status phase during NDRT;

[0009] The state behaviors in the takeover phase after TOR are divided into TOR human-computer interaction, takeover switching and control, and workload throughout the entire takeover phase, and each state behavior in the takeover phase after TOR is evaluated separately.

[0010] Furthermore, the comprehensive status of the driver in the non-driving related behaviors and status phases during the NDRT period is evaluated, including:

[0011] The comprehensive state of the driver during the NDRT of autonomous driving is characterized by using the PAD emotion scale, NASA-TLX1 load scale and KSS fatigue sleepiness scale scores, and using ECG_1, ECG_2, ECG_3, ECG_4, ECG_5, EDA, Eye_1 and Eye_2 as physiological and behavioral indicators; the comprehensive state includes the driver's emotion, cognitive load and fatigue level;

[0012] Among them, ECG_1 represents heart rate, ECG_2 represents the average interval between adjacent heart beats, ECG_3 represents the variance of adjacent heart beat intervals, ECG_4 represents the percentage of consecutive heart beat intervals with a difference greater than 20ms in the total number of heart beat intervals, ECG_5 represents the ratio of the low-frequency component LF to the high-frequency component HF in the ECG signal; Eye_1 represents the average pupil diameter, and Eye_2 represents the number of glances.

[0013] Furthermore, the TOR human-computer interaction is evaluated, including:

[0014] Evaluate from two dimensions: easy to find and easy to understand:

[0015] It is easy to find that it is characterized by the subjective score Acs-s and the objective behavioral indicators Eye_8 and Eye_9;

[0016] Comprehension was characterized by the subjective score Cph-s and the objective behavioral indicators Eye_4 and Eye_5;

[0017] Among them, Acs-s represents the easy-to-find score, Cph-s represents the easy-to-understand score, Eye_4 represents the total duration and total number of dashboard visits, Eye_5 represents the total duration of visits to the projection information area, Eye_8 represents the first dashboard visit time, and Eye_9 represents the first projection information area visit time;

[0018] Evaluation is conducted from three aspects: satisfaction, acceptance and reliability:

[0019] Characterized by subjective scores Stf-s, Rcp-s, and Rel-s;

[0020] Among them, Stf-s represents the satisfaction score, Rcp-s represents the acceptance score, and Rel-s represents the reliability score.

[0021] Furthermore, the takeover switching and control are evaluated, including:

[0022] The evaluation is based on the driver's reaction, perception of vehicle and road conditions, and decision-making control after taking over:

[0023] The driver's reaction ability is characterized by the subjective score Rct-s and the objective index Eye_7;

[0024] The perception of vehicle and road conditions is characterized by the subjective score Awn-s and the objective indicators Eye_3, Eye_4, Eye_6, and Eye_10;

[0025] The decision control is characterized by the subjective score Sta-s and vehicle indicators Veh_1, Veh_2, Veh_3, Veh_4, Veh_5, and Veh_6;

[0026] Among them, Rct-s represents the reaction time score, Eye_7 represents the first road visit time, Awn-s represents the easy perception score, Eye_3 represents the total road visit duration and total number of times, Eye_4 represents the total instrument panel visit duration and total number of times, Eye_6 represents the total road situation area visit duration and total number of times, Eye_10 represents the first road situation visit time, Sta-s represents the stability score, Veh_1 represents the manipulation reaction time, Veh_2 represents the collision time of the vehicle relative to the obstacle in front when entering the new lane, Veh_3 represents the vehicle lateral crossing time, Veh_4 represents the mean lateral lane deviation, Veh_5 represents the longitudinal acceleration variance, and Veh_6 represents the lateral acceleration variance.

[0027] Furthermore, the workload throughout the entire takeover phase is evaluated, including:

[0028] Characterized by subjective scoring NASA-TLX2 load scale, eye movement index Eye_1 and Eye_2, skin electrode index EDA, and electrocardiogram index ECG_1, ECG_2, ECG_3, ECG_4, and ECG_5;

[0029] Wherein, EDA represents the average skin electrical activity.

[0030] Furthermore, it further includes: paying attention to the impact of the comprehensive state of the driver during non-driving related tasks on the situation awareness perception behavior;

[0031] Among them, the situation awareness refers to the driver's cognition of the vehicle condition and road condition, which is characterized by the objective indicators Eye_3 and Eye_4.

[0032] Furthermore, the following indicators are set to characterize the influence relationship existing between each stage in the takeover timing of the driver's state behavior:

[0033] H2: The influence of the driver's behavior state and situation awareness ability during automatic driving on each stage after TOR; If the subject shows a more fatigued state or a lower situation awareness ability during NDRT, it will directly affect the takeover efficiency of the subject and the control stability of the vehicle, increase the takeover load, and reduce the subject's experience evaluation of TOR, thereby affecting the entire process after takeover;

[0034] H3: The influence of the TOR strategy for automatic driving on the comprehensive behavior state during automatic driving; If TOR is replaced from the general modal strategy to the multi-modal hierarchical prompting strategy under typical scenario tasks, after the subject is familiar with it, it may increase the trust degree of the subject in the vehicle during automatic driving;

[0035] H4: The two-way influence of the overall experience of TOR human-machine interaction and the load during takeover; If the overall usability of TOR increases, the takeover load will be reduced, thereby improving the takeover efficiency, and the feedback will improve the subject's experience evaluation of TOR;

[0036] H5: The influence of the overall experience of TOR human-machine interaction on the takeover operation and short-term vehicle control; If the overall usability of TOR increases, the takeover efficiency will be directly improved, thereby increasing the switching efficiency and stability.

[0037] The beneficial effects of the present invention are as follows: A driver takeover efficiency evaluation method for the whole process of the automatic driving takeover timing disclosed by the present invention establishes a timing evaluation model for the driver's takeover behavior state, analyzes the driver's state, situation awareness, driver takeover control period load, takeover switching and control, and takeover request scheme during automatic driving through subjective and objective multi-dimensional data, effectively characterizes the behavior state characteristics of the driver during each takeover timing, and provides multi-faceted support for the implementation of future automatic driving in terms of clarifying the responsibility boundary, ensuring personnel safety, improving personnel experience and compliance. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further described below with reference to the drawings and embodiments:

[0039] Figure 1 It is a schematic diagram of the timing evaluation principle of the driver's takeover behavior state of the present invention;

[0040] Figure 2 It is a schematic diagram of the experimental task flow and data flow of the present invention. DETAILED DESCRIPTION

[0041] The present invention is further described below in conjunction with the accompanying drawings, as shown in the drawings:

[0042] This embodiment discloses a method for evaluating the driver's takeover effectiveness for the entire process of automatic driving takeover sequence, including the following steps:

[0043] The takeover sequence of driver state behavior is divided into the non-driving related behavior and state stage during NDRT and the takeover stage after TOR;

[0044] Subjective scale scores and objective physiological and behavioral indicators were set to evaluate the comprehensive status of the driver during the non-driving related behavior and status phase during NDRT;

[0045] The state behaviors in the takeover phase after TOR are divided into TOR human-computer interaction, takeover switching and control, and workload throughout the entire takeover phase, and each state behavior in the takeover phase after TOR is evaluated separately.

[0046] The present invention can handle high-level (such as L3) autonomous driving, evaluate the takeover capabilities under different tasks, comprehensively analyze whether the driver's status is out of the loop at each takeover stage under different tasks, and specifically designs a set of temporal multimodal ToRs mechanisms. This mechanism has temporal and urgency classification characteristics, and realizes personalized settings of ToRs based on driver characteristics and from the perspective of human-machine mutual trust, providing multi-faceted support for the implementation of autonomous driving in the future, such as clarifying the boundaries of responsibilities, ensuring personnel safety, and improving personnel experience and compliance.

[0047] Among them, NDRT (Non Driving Related Tasks), NDBC (Non driving Behavior & Condition), TOR (Takeover Request), TO (Takeover). NDBC refers to the driver's behavior and state that is not related to driving the car before the autonomous car issues a takeover request, such as reading news, watching videos, playing games, sending text messages, fatigue, anger, depression, etc. TOR refers to the takeover request issued by the autonomous car when it encounters a scene that it is not competent for, which is generally composed of a prompt sound issued by the car and the UI of the dashboard. TO refers to the driver's takeover process, and the research content includes TOR human-computer interaction, takeover switching and control, and the workload throughout the entire takeover stage.

[0048] In this embodiment, the comprehensive state of the driver during the non-driving related behaviors and status stages of NDRT is evaluated, specifically including:

[0049] By using the PAD emotion scale, the NASA-TLX1 workload scale, and the KSS fatigue and sleepiness scale for scoring, and taking ECG_1, ECG_2, ECG_3, ECG_4, ECG_5, EDA, Eye_1, and Eye_2 as physiological and behavioral indicators, the comprehensive state of the driver during autonomous driving NDRT is characterized; the comprehensive state includes the driver's emotion, cognitive load, and fatigue level;

[0050] Among them, ECG_1 represents heart rate, ECG_2 represents the mean adjacent R-R interval (IBI / MeanNN), ECG_3 represents the standard deviation of the R-R intervals (SDNN), ECG_4 represents the percentage of the number of consecutive R-R intervals with a difference greater than 20 ms in the total number of R-R intervals (PNN20), ECG_5 represents the ratio of the low-frequency component LF to the high-frequency component HF in the electrocardiogram signal; Eye_1 represents the average pupil diameter, and Eye_2 represents the number of saccades. The PAD emotion scale, the NASA-TLX1 workload scale, and the KSS fatigue and sleepiness scale are three different psychological assessment scales, which are used to evaluate emotional state, workload, and fatigue and sleepiness levels respectively.

[0051] In this embodiment, the human-machine interaction scheme for the takeover request (TOR) is evaluated from two dimensions: ease of use and the evaluation of the TOR scheme experience. That is, the TOR human-machine interaction is evaluated, specifically including:

[0052] The ease of use includes three aspects: visual, auditory, and tactile, and is evaluated from two dimensions: easy to discover and easy to understand:

[0053] Easy to discover is characterized by the subjective score Acs-s and the objective behavioral indicators Eye_8 and Eye_9;

[0054] Easy to understand is characterized by the subjective score Cph-s and the objective behavioral indicators Eye_4 and Eye_5;

[0055] Among them, Acs-s represents the easy-to-discover score, Cph-s represents the easy-to-understand score, Eye_4 represents the total duration and total number of dashboard accesses, Eye_5 represents the total duration of accessing the projected information area, Eye_8 represents the first dashboard access time (the first dashboard access moment after TOR - the TOR moment), and Eye_9 represents the first projected information area access time (the first AR-HUD access moment after TOR - the TOR moment);

[0056] The projection information area can be the AR-HUD area, which refers to the spatial range where the Augmented Reality Head-Up Display projects information. AR-HUD is a device that superimposes virtual information onto the real world. It projects virtual images and information onto the windshield through an optical projection system, enabling the driver to better understand the surrounding environment.

[0057] The driver's experience evaluation of the TOR solution is carried out from three aspects: satisfaction, acceptance, and reliability.

[0058] It is characterized by subjective scores Stf-s, Rcp-s, and Rel-s.

[0059] Among them, Stf-s represents the satisfaction score, Rcp-s represents the acceptance score, and Rel-s represents the reliability score. H6 represents the impact of the improvement in the easy discovery and easy understanding performance of TOR on the satisfaction, acceptance, and reliability of the TOR strategy. For example, if the usability performance of TOR is improved, the subjects will be more satisfied with the TOR strategy, more trusting, and more willing to accept its application in real scenarios.

[0060] In this embodiment, the evaluation of takeover switching and control is specifically as follows:

[0061] The evaluation is carried out from three aspects: the driver's reaction ability, the perception of vehicle conditions and road conditions, and the decision-making control after takeover.

[0062] The driver's reaction ability is characterized by the subjective score Rct-s and the objective index Eye_7.

[0063] The perception of vehicle conditions and road conditions is characterized by the subjective score Awn-s and the objective indexes Eye_3, Eye_4, Eye_6, and Eye_10.

[0064] The decision-making control is characterized by the subjective score Sta-s and the vehicle indexes Veh_1, Veh_2, Veh_3, Veh_4, Veh_5, and Veh_6.

[0065] Among them, Rct-s represents the reaction time score, Eye_7 represents the first road access time (the moment of the first road access after TOR - the moment of TOR), Awn-s represents the perceptibility score, Eye_3 represents the total duration and total number of road accesses, Eye_4 represents the total duration and total number of dashboard accesses, Eye_6 represents the total duration and total number of road situation area accesses, Eye_10 represents the first road situation access time (the moment of the first road situation access - the moment of TOR), Sta-s represents the stability score, Veh_1 represents the control reaction time (the moment when the button is pressed / the steering wheel is turned > ±2° or the process of stepping on the brake pedal > 1% - the moment of TOR), Veh_2 represents the collision time (ttcl) of the vehicle entering a new lane relative to the front obstacle, Veh_3 represents the vehicle's lateral crossing time (TLC), Veh_4 represents the mean value of the lateral lane offset (Lane_offset), Veh_5 represents the variance of the longitudinal acceleration, and Veh_6 represents the variance of the lateral acceleration.

[0066] In this embodiment, the workload throughout the takeover phase is evaluated, specifically including:

[0067] It is characterized by the subjective score NASA - TLX2 workload scale, eye movement indicators Eye_1 and Eye_2, galvanic skin response indicator EDA, electrocardiogram indicators ECG_1, ECG_2, ECG_3, ECG_4, and ECG_5; the NASA - TLX2 workload scale is a psychological assessment scale, belonging to the NASA - TLX workload scale together with the NASA - TLX1 workload scale, and is used to evaluate the workload.

[0068] Among them, EDA represents the mean value of the galvanic skin response.

[0069] In this embodiment, it also includes: paying attention to the impact of the comprehensive state of the driver during non - driving - related tasks on the situation awareness perception behavior;

[0070] Among them, the situation awareness refers to the driver's cognition of the vehicle condition and road condition, which is characterized by the objective indicators Eye_3 and Eye_4. Let H1 represent the impact of the comprehensive state of the driver during non - driving - related tasks on the situation awareness perception behavior.

[0071] In this embodiment, the following indicators are set to characterize the influence relationship existing between each stage in the takeover timing of the driver's state behavior:

[0072] H2: The influence of the driver's behavioral state and situation awareness ability during automated driving on each stage after TOR; for example, if the subjects show a more fatigued state or lower situation awareness ability during NDRT, it will directly affect the takeover efficiency of the subjects and the control stability of the vehicle, increase the takeover load, and reduce the subjects' experience evaluation of TOR, thus affecting the entire process after takeover.

[0073] H3: The influence of the TOR strategy for automated driving on the comprehensive behavioral state during automated driving; for example, if TOR is replaced from a general modal strategy to a multi-modal hierarchical prompting strategy under typical scenario tasks, it may increase the subjects' trust in the vehicle during automated driving after they are familiar with it.

[0074] H4: The two-way influence of the overall experience of TOR human-machine interaction and the load during takeover; for example, when the overall usability of TOR increases, the takeover load is reduced, thereby improving the takeover efficiency, and the feedback enhances the subjects' experience evaluation of TOR.

[0075] H5: The influence of the overall experience of TOR human-machine interaction on takeover operations and short-term vehicle control; for example, when the overall usability of TOR increases, the takeover efficiency is directly improved, thereby increasing the switching efficiency and stability.

[0076] To better understand the driver takeover efficiency evaluation method of the present invention, the proposed evaluation method is verified based on a simulated cockpit experiment. The experiment uses a simulation environment experiment method, that is, a driving simulator is used in the laboratory to build a scenario for simulated driving. On the simulated driving platform, non-invasive monitoring devices such as cameras and microphone arrays will be installed. A total of three groups of experimental variables are set, namely non-driving related tasks, takeover scenarios, and TOR, to verify the proposed multi-dimensional evaluation method of subjective and objective time series and the time series influence mechanism.

[0077] (1) Experimental design

[0078] Selection of non-driving related tasks: The interest groups of automated driving show diverse interests and high demand for entertainment (refer to the "Insight Report on Potential Users of Automated Driving (2021)"), so in this experiment, NDRT focusing on entertainment is selected, including reading, video, and games.

[0079] Selection of takeover scenarios: The scenarios are selected based on the team's research results, that is, scenarios that usually require human-machine co-driving or forced takeover, including urban road scenarios, highway ramp scenarios, and fog environment scenarios. They are built on the driving simulator, and the free flow traffic density is set to 6 pcu·km-1·lane-1, and the stable flow traffic density is set to 20 pcu·km-1·lane-1.

[0080] TOR Selection: Two types of takeover prompts are mainly selected. One is the general modal prompt strategy, and the other is the multi-modal hierarchical prompt strategy under typical scenario tasks, to verify the synergy mechanism of the multi-modal assisted takeover interaction design under typical scenario tasks in the theoretical model.

[0081] Each experimental scenario is as Figure 2 shown. The subjects start the automatic driving under the prompt and immerse themselves in the corresponding non-driving related tasks. During this process, emergencies may occur at any time, and the vehicle will send a takeover request, asking the subjects to take over as smoothly as possible and try to avoid collisions and accidents when taking over the vehicle.

[0082] During the experiment, a multi-channel physiological instrument and a head-mounted eye tracker are used to synchronously collect the subjects' galvanic skin data, electrocardiogram data, eye movement data, and the vehicle's various-dimensional driving data.

[0083] (2) Subject Screening

[0084] The data of this experiment is processed in accordance with the Helsinki Declaration. According to the description of the group portrait of those interested in autonomous driving (refer to the "Insight Report on Potential Users of Autonomous Driving (2021)"), 30 young participants are involved (18 males, 12 females), aged between 21 - 28 years old (M = 22.8, SD = 1.77). The driving mileage ranges from over 100,000 kilometers to less than 1,000 kilometers (4 levels, M = 1.4, SD = 0.67). All subjects have normal / corrected vision, a valid driver's license, and a driving history within the last year.

[0085] (3) Data Processing and Analysis

[0086] The eye movement, electrocardiogram, galvanic skin, and vehicle data are filtered to form the original data set, and then grouped and sliced together with the subjective data; based on the static data, the electrocardiogram and galvanic skin data of each subject are standardized to eliminate the influence of individual differences; non-parametric tests or analysis of variance are used to compare the subjective and objective data among the experimental variables, verify the various temporal influence mechanisms proposed by the evaluation model from the subjective dimension, assist in verifying this conclusion from the objective electrocardiogram, galvanic skin, and vehicle data dimensions, and verify the effectiveness of this subjective-objective multi-dimensional evaluation method.

[0087] The present invention comprehensively considers the physiological and psychological states and eye behaviors of the driver before, during, and after takeover and during autonomous driving, the experience evaluation of the takeover strategy during takeover, and the workload during the takeover process, and proposes a high-order autonomous driving driver takeover timing evaluation model, forming a technical solution for the test and evaluation of high-order autonomous driving takeover strategies, driver takeover effectiveness, and risk test and evaluation.

[0088] Through the driver takeover ability experiment, the multi-dimensional evaluation method and the time-series influence mechanism of the driver takeover state behavior under high-level autonomous driving are verified, making the subjective and objective comprehensive evaluation of the whole process of the driver takeover time series more scientific and effective.

[0089] The present invention proposes a systematic subjective and objective evaluation method for the driver takeover efficiency of the whole process of the high-level autonomous driving takeover time series. Through scientific analysis and practical verification, it can effectively characterize the behavioral state characteristics of the subjects at each time series of the takeover.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for evaluating driver takeover effectiveness for the entire process of automatic driving takeover sequence, characterized by: include: The takeover sequence of driver state behavior is divided into the non-driving related behavior and state stage during NDRT and the takeover stage after TOR; Subjective scale scores and objective physiological and behavioral indicators were set to evaluate the comprehensive status of the driver during the non-driving related behavior and status phase during NDRT; The state behaviors in the takeover phase after TOR are divided into TOR human-computer interaction, takeover switching and control, and workload throughout the entire takeover phase, and each state behavior in the takeover phase after TOR is evaluated separately.

2. The method for evaluating the driver's takeover effectiveness in the entire process of automatic driving takeover sequence according to claim 1, characterized in that: The comprehensive status of the driver during the non-driving related behaviors and status phases during NDRT is evaluated, including: The comprehensive state of the driver during the NDRT of autonomous driving is characterized by using the PAD emotion scale, NASA-TLX1 load scale and KSS fatigue sleepiness scale scores, and using ECG_1, ECG_2, ECG_3, ECG_4, ECG_5, EDA, Eye_1 and Eye_2 as physiological and behavioral indicators; the comprehensive state includes the driver's emotion, cognitive load and fatigue level; Among them, ECG_1 represents heart rate, ECG_2 represents the average interval between adjacent heart beats, ECG_3 represents the variance of adjacent heart beat intervals, ECG_4 represents the percentage of consecutive heart beat intervals with a difference greater than 20ms in the total number of heart beat intervals, ECG_5 represents the ratio of the low-frequency component LF to the high-frequency component HF in the ECG signal; Eye_1 represents the average pupil diameter, and Eye_2 represents the number of glances.

3. The driver takeover efficiency evaluation method for the entire process of automatic driving takeover sequence according to claim 1 is characterized by: Evaluate TOR human-computer interaction, including: Evaluate from two dimensions: easy to find and easy to understand: It is easy to find that it is characterized by the subjective score Acs-s and the objective behavioral indicators Eye_8 and Eye_9; Comprehension was characterized by the subjective score Cph-s and the objective behavioral indicators Eye_4 and Eye_5; Among them, Acs-s represents the easy-to-find score, Cph-s represents the easy-to-understand score, Eye_4 represents the total duration and total number of dashboard visits, Eye_5 represents the total duration of visits to the projection information area, Eye_8 represents the first dashboard visit time, and Eye_9 represents the first projection information area visit time; Evaluation is conducted from three aspects: satisfaction, acceptance and reliability: Characterized by subjective scores Stf-s, Rcp-s, and Rel-s; Among them, Stf-s represents the satisfaction score, Rcp-s represents the acceptance score, and Rel-s represents the reliability score.

4. The method for evaluating driver takeover effectiveness for the entire process of automatic driving takeover sequence according to claim 1, characterized in that: Evaluate the takeover and control, including: The evaluation is based on the driver's reaction, perception of vehicle and road conditions, and decision-making control after taking over: The driver's reaction ability is characterized by the subjective score Rct-s and the objective index Eye_7; The perception of vehicle and road conditions is characterized by the subjective score Awn-s and the objective indicators Eye_3, Eye_4, Eye_6, and Eye_10; The decision control is characterized by the subjective score Sta-s and vehicle indicators Veh_1, Veh_2, Veh_3, Veh_4, Veh_5, and Veh_6; Among them, Rct-s represents the reaction time score, Eye_7 represents the first road visit time, Awn-s represents the easy perception score, Eye_3 represents the total road visit duration and total number of times, Eye_4 represents the total instrument panel visit duration and total number of times, Eye_6 represents the total road situation area visit duration and total number of times, Eye_10 represents the first road situation visit time, Sta-s represents the stability score, Veh_1 represents the manipulation reaction time, Veh_2 represents the collision time of the vehicle relative to the obstacle in front when entering the new lane, Veh_3 represents the vehicle lateral crossing time, Veh_4 represents the mean lateral lane deviation, Veh_5 represents the longitudinal acceleration variance, and Veh_6 represents the lateral acceleration variance.

5. The method for evaluating driver takeover effectiveness for the entire process of automatic driving takeover sequence according to claim 1, characterized in that: Evaluate workload throughout the takeover phase, including: Characterized by subjective scoring NASA-TLX2 load scale, eye movement index Eye_1 and Eye_2, skin electrode index EDA, and electrocardiogram index ECG_1, ECG_2, ECG_3, ECG_4, and ECG_5; Wherein, EDA represents the average skin electrical activity.

6. The driver takeover efficiency evaluation method for the entire process of automatic driving takeover sequence according to claim 2 is characterized by: Also includes: Focus on the impact of the driver's comprehensive state on situational awareness perception behavior when performing non-driving related tasks; The situational awareness refers to the driver's cognition of the vehicle and road conditions, which is represented by the objective indicators Eye_3 and Eye_4.

7. The method for evaluating driver takeover effectiveness for the entire process of automatic driving takeover sequence according to claim 1, characterized in that: The following indicators are set to characterize the influence relationship between each stage in the takeover sequence of the driver's state behavior: H2: The impact of the driver's behavior and situational awareness during autonomous driving on the various stages after TOR; if the subject is more fatigued or has lower situational awareness during the NDRT, it will directly affect the subject's takeover efficiency and vehicle control stability, increase the takeover load, and reduce the subject's experience evaluation of TOR, thus affecting the entire post-takeover process; H3: The impact of the TOR strategy of autonomous driving on the comprehensive behavioral state during autonomous driving; if the TOR is replaced from a general modality strategy to a multimodal hierarchical prompt strategy under a typical scenario task, after the subjects are familiar with it, it may increase their trust in the vehicle during autonomous driving; H4: The overall experience of TOR human-computer interaction and the load during takeover have a two-way impact; if the overall availability of TOR increases, the takeover load will be reduced, thereby improving the takeover efficiency, and the feedback will improve the subjects' experience evaluation of TOR; H5: The impact of the overall experience of TOR human-computer interaction on takeover operations and short-term vehicle control; if the overall availability of TOR increases, the takeover efficiency will be directly improved, thereby increasing switching efficiency and stability.

Images