Bionic Robot and Method for Detecting Ride Comfort in Intelligent Vehicles

Abstract

This invention relates to the field of automotive cabin comfort testing technology, and discloses an intelligent bionic robot and method for testing automotive driving comfort. The robot includes a support frame, a bionic eye, a vibration structure, a gimbal, and a central controller. The bionic eye is positioned on the upper part of the support frame, with an active vibration module connected below it. The vibration structure is connected to the bottom of the support frame, and the gimbal is positioned between the vibration structure and the support frame. The central controller is electrically connected to the bionic eye, the active vibration module, the vibration structure, and the gimbal, respectively, for unified control of the coordinated actions of each module. This invention achieves automated, high-precision, repeatable, and safe quantitative testing of automotive cabin driving comfort without human intervention.

Description

Technical Field

[0001] This invention relates to the field of automotive cabin comfort detection technology, specifically to an intelligent automotive driving comfort detection bionic robot and method. Background Technology

[0002] With the deepening research on intelligent automotive cockpits and driving experience, the testing of driving comfort under dynamic driving conditions has become a core part of cockpit design and verification. Among them, the visual fatigue comfort of users when looking at mobile phones, books, etc. in the car is an important dimension of current cockpit experience evaluation.

[0003] Among existing automotive ride comfort testing technologies, the closest prior art is a method and system scheme for testing and evaluating the comfort of autonomous vehicles based on an autonomous driving robot, disclosed in patent document CN113647956A. This method uses an autonomous driving robot to control the vehicle to execute preset movements, simultaneously collects the electrocardiographic data of the test driver, and combines this with the driver's subjective rating for cross-comparison between subjective and objective data to ultimately determine the vehicle's comfort motion parameters. While this scheme achieves high precision and repeatability in vehicle motion control, it has significant shortcomings in practical applications. Testing must rely on human participants, which is not only time-consuming and labor-intensive, but also costly. Furthermore, conducting tests under extreme conditions can easily cause physiological harm to the human body, making it impossible to carry out safe extreme value tests. The high proportion of subjective evaluation and large individual differences lead to a large range of fluctuations in test data, requiring a large number of samples for data calibration, which significantly increases the workload of testing. Evaluating comfort solely based on bodily physiological perceptions cannot simulate the visual fatigue experienced by users when using electronic devices or reading books inside the cabin. It also lacks automated detection capabilities for visual dimensions, making it difficult to meet the needs for accurate testing of cabin visual comfort.

[0004] Furthermore, the car driving simulation platform disclosed in patent document CN216249608U only enhances the haptic sensation of driving simulation through left and right shaking mechanism and support rod, without any comfort detection function, bionic structure, or visual perception module, and cannot solve the technical problem of cabin visual fatigue detection.

[0005] In summary, existing technologies cannot achieve automated, safe, and high-precision quantitative detection of visual fatigue comfort in automotive cabins without human intervention. The industry urgently needs a biomimetic detection technology that can replace human intervention, specifically detect visual fatigue, and adapt to dynamic cabin environments. Summary of the Invention

[0006] This invention aims to provide a bionic robot and method for detecting the driving comfort of intelligent vehicles, in order to solve the problems of existing vehicle cabin comfort detection relying on human testing, which is time-consuming and labor-intensive, extreme value detection is harmful to the human body, and the data is highly subjective and fluctuates greatly. At the same time, it also solves the problems of low efficiency and poor accuracy in detecting visual fatigue comfort in the cabin.

[0007] To solve the above problems, the present invention adopts the following technical solution: Option 1: A bionic robot for detecting the driving comfort of intelligent vehicles, including a support frame, bionic eyes, a vibration structure, a gimbal, and a central controller; the support frame simulates the human body, the bionic eyes are located on the upper part of the support frame, and an active vibration module is connected below the bionic eyes; the vibration structure is connected to the bottom of the support frame, and the gimbal is located between the vibration structure and the support frame; the central controller is electrically connected to the bionic eyes, the active vibration module, the vibration structure, and the gimbal, respectively, for unified control of the coordinated actions of each module; The bionic human eye includes a miniature camera, an eyelid structure, a convex lens, and a rotating mechanism. The miniature camera is located inside the eyelid structure, which is an upper and lower structure connected to a miniature servo motor via a connecting rod. The servo motor controls the opening and closing of the eyelid, performing periodic blinking at a frequency of 15-20 times per minute. A central controller controls the blinking frequency and closing duration to simulate eye fatigue. The convex lens is located in front of the camera, with a perforation at its edge through which a rotating shaft passes. The rotating shaft is connected to a connecting rod, and the other end of the connecting rod is connected to a miniature motor, which drives the connecting rod to rotate the convex lens. The lower eyelid achieves rapid closing through a flat gear transmission, simulating fast blinking. By varying the blinking speed, the blinking frequency can be simulated for different ages, road conditions, visual acuity, and levels of eye fatigue. The central controller has a built-in visual fatigue quantitative assessment unit, which is configured to perform the following visual fatigue quantitative assessment steps: The first step was to collect the blinking dynamics, half-closed eye state features, gaze stability features, and pupil diameter change rate (PL) of the bionic human eye in a 60fps video sequence. Among them, the blink dynamic characteristics include blink count per minute (BPM) and single blink duration (BK), with the normal blink duration set to 100-400 milliseconds; when the duration of multiple consecutive blinks exceeds 450ms, it is judged as a severe fatigue characteristic. Half-closed eye state characteristics: When the eye posture estimation EAR value decreases but the eyes are not completely closed and the duration exceeds 2 seconds, it is judged as a severe fatigue characteristic; Fixation stability characteristics: The root mean square error of the pupil center displacement in the two-dimensional plane was analyzed using optical flow. When the fixation point drift exceeds 1.2° of visual angle and lasts for more than 3 seconds under vibration, a "fixation instability event" is recorded. Pupil diameter change rate (PL): The pupil diameter change rate is calculated under fixed lighting conditions. When the diameter change rate decreases to <0.05 mm / s, it is determined that the visual accommodation ability has decreased. The second step involves the visual fatigue quantification assessment unit outputting a fatigue score of 0 to 10 based on the following fatigue quantification model: Score = 0.4 × (BPM_norm) + 0.3 × (BK_norm) + 0.2 × (FS_norm) + 0.1× (PL_norm) Wherein, BPM_norm is the normalized value of blink frequency, BK_norm is the normalized value of blink duration, FS_norm is the normalized value of fixation stability, and PL_norm is the normalized value of pupil diameter change rate. The third step is for the central controller to classify fatigue levels based on fatigue scores: 9-10 points indicate no fatigue, 6-8 points indicate mild fatigue, 3-5 points indicate moderate fatigue, and 0-2 points indicate severe fatigue. Based on these levels, the controller controls the active shaking module, vibration structure, and gimbal adjustment simulation strategy. Vehicle model simulation is divided into large, medium and small vehicles. Both vehicle model simulation and road simulation can change the vibration frequency of the vibrating structure.

[0008] Beneficial effects: Constructing a complete bionic detection hardware system, using bionic eyes to simulate human visual perception, replacing real people to complete cabin visual fatigue comfort detection, fundamentally eliminating the dependence on real people for testing.

[0009] Preferably, the vibration structure includes a high-response miniature cylinder, a U-shaped connecting rod, a pressure plate, a spring, and a sleeve; the cylinder piston rod is rigidly connected to the U-shaped connecting rod, the pressure plate is fixed to the end of the U-shaped connecting rod and abuts against the top of the spring, and the spring is fitted inside the sleeve; the cylinder air supply frequency is 5Hz to 30Hz, and the amplitude adjustment range is 0.1mm to 5mm.

[0010] Beneficial effects: By digitally quantifying vibration parameters, the vibration spectrum of different road conditions such as asphalt roads and gravel roads can be accurately simulated, improving the repeatability and simulation accuracy of the testing conditions.

[0011] Preferably, the gimbal is equipped with an electromechanical integrated locking mechanism. The locking mechanism is driven by an auxiliary cylinder, and the locking mechanism moves up and down along the gimbal connecting shaft by a stroke of 3mm to 10mm. When the locking mechanism moves up to the upper end of the connecting shaft, the gimbal is in an active anti-shake state. When the locking mechanism moves down to the bottom groove of the connecting shaft, the gimbal is in a rigid locking state. The gimbal is also equipped with a locking rod, which is connected to a cylinder. The cylinder enables extension and retraction. When extended, the locking rod passes through the locking mechanism to fix the gimbal.

[0012] Beneficial effects: The telescopic mechanism uses a cylinder for extension and retraction. When extended, the locking rod passes through the locking slot, fixing the gimbal in place and making it usable only as a connector. It allows for quick switching between two gimbal operating modes: an anti-shake mode that replicates the adaptive adjustment capabilities of the human eye, and a locking mode that meets the needs of precise testing from a fixed angle, adapting to various testing scenarios.

[0013] Preferably, the support is a three-axis moving structure with an X-axis travel of -600mm to 600mm, a Y-axis travel of -400mm to 400mm, and a Z-axis travel of -300mm to 300mm, used to simulate the natural displacement of a human body in a car cabin; the structure of the three-axis moving support is consistent in the X, Y, and Z directions, adopts a modular design, and uses standardized interfaces and dimensions. The length of the slide rail can be replaced with different lengths according to different situations, and both the slide rail and the slider use standard parts; the gimbal is also modularly designed, and the mechanical interface and electrical plug-in are all quick-change standard interfaces.

[0014] Beneficial effects: Enables rapid installation and replacement of different configurations. Accurately reproduces the limb movements of drivers and passengers within the cabin, making the testing scenario closer to real-world usage and improving the authenticity of the test results.

[0015] Preferably, the active shaking module of the bionic human eye and the vibration structure are synchronously controlled by a central controller, and the synchronization error between the two is ≤0.01s.

[0016] Beneficial effects: It achieves high-precision synchronization between eye tremors and vehicle vibration, completely restoring the visual perception environment under dynamic driving conditions, making visual fatigue detection more realistic.

[0017] Preferably, the central controller has a built-in scenario script module and a closed-loop feedback module, which supports preset continuous bump and emergency lane change test scenarios, and the sensor feedback response time is ≤0.05s.

[0018] Beneficial effects: Enables fully automated testing, closed-loop control ensures motion accuracy, eliminates the need for real-time manual intervention, and significantly improves testing efficiency.

[0019] Option 2: A bionic robot method for detecting the driving comfort of intelligent vehicles, based on any of the preceding bionic robots, including the following steps: S1. The central controller presets vibration parameters, pan-tilt working mode, support movement trajectory, and visual fatigue judgment thresholds; the visual fatigue judgment thresholds include: normal blinking frequency range [15, 20] times / minute, blinking frequency ≤10 times / minute or ≥25 times / minute correspond to "visual inhibitory fatigue" or "stress-induced increased blinking" respectively; normal blinking duration 100-400 milliseconds; gaze instability trigger threshold 1.2° visual angle and lasting for 3 seconds; pupil diameter change rate threshold 0.05 mm / s; S2. The vibration structure simulates the dynamic shaking of the cockpit. The gimbal switches to the corresponding mode, the three-axis support simulates the displacement of the human body in the cockpit, and the active shaking module moves synchronously with the vibration structure. S3. The bionic human eye acquires visual signals in real time, and the central controller executes the following sub-steps: Calculate blink rate (BPM). If BPM ≤ 10 or BPM ≥ 25, include it in the fatigue factor. The duration of a single blink is measured. If the duration of multiple consecutive blinks exceeds 450 ms or the half-closed state lasts for more than 2 seconds, it is judged as severe fatigue. The root mean square error of pupil center displacement in a 60fps video sequence was analyzed using optical flow method to detect the frequency of gaze instability events. Calculate the pupil diameter change rate PL under fixed illumination and compare it with the threshold of 0.05 mm / s; The text displayed in front of the camera is captured and compared with the input text to calculate the text recognition error rate C. S4. The central controller inputs the collected data into the fatigue quantification model to calculate the fatigue score, compares it with the preset threshold, outputs the visual fatigue comfort quantification detection result, and adaptively adjusts the simulation strategy according to the score level. Meanwhile, the central controller calculates the overall cabin comfort based on the following comfort index formula: S = 0.6 × (1+C) / S_text_max + 0.4 × A × [α×(Fx² + Fy² + Fz²)] Where S is the comfort index, C is the text recognition error rate, α is the number of experimental data groups, Fx, Fy, and Fz are the vibration frequencies of the X, Y, and Z axes, respectively, in Hz; the vibration frequency accounts for 0.4, and the text recognition error rate accounts for 0.6. S_text_max = 1 + C_max, which is the maximum (1+C) value calibrated in the preliminary experiment; A = 1 / S_freq_max, where A is the normalization coefficient for the vibration frequency term; S_freq_max is the sum of squares of the maximum vibration frequencies. Preliminary experimental calibration: S_freq_max = α × (Fx_max² + Fy_max² + Fz_max²); Fx_max, Fy_max, and Fz_max are the maximum vibration frequencies of each axis, respectively.

[0020] Beneficial effects: It forms a complete automated detection process that does not require human intervention, enabling safe, efficient, and repeatable detection of cabin visual fatigue comfort.

[0021] Preferably, the cylinder ventilation frequency of the vibration structure in S2 is set to 20Hz and the amplitude is set to 1mm to simulate the driving conditions of a standard asphalt road.

[0022] Beneficial effects: Standardizing typical road condition parameters facilitates the benchmarking, reproduction, and comparison of test data, reducing detection errors.

[0023] Preferably, in S2, the control gimbal bayonet moves down to the bottom groove of the coupling, the gimbal remains in a rigid locked state, and the bionic human eye continuously collects visual signals from a fixed angle.

[0024] Preferably, the bionic human eye in S3 collects data at a frequency of 30 frames per second, and the central controller outputs a quantified comfort score of 0 to 10 points, with a lower score indicating a higher degree of visual fatigue.

[0025] Beneficial effects: It transforms subjective visual comfort into objective numerical scores, with stable and unfluctuating data, facilitating cabin design optimization and effect verification.

[0026] (a) Advantages of the present invention 1. Completely replaces human testing, avoids personal injury from extreme value testing, and significantly reduces testing costs and time; 2. It integrates vibration simulation, gimbal switching, three-axis displacement, and bionic vision to achieve dual-dimensional detection of body perception and visual perception; 3. The entire process is digitalized, repeatable, and under closed-loop control, with data accuracy far exceeding that of subjective human testing, eliminating the need for extensive sample calibration.

[0027] (ii) Unexpected solutions to problems This invention breaks through the inherent understanding in the field that comfort testing must rely on the physiological and subjective feedback of real people. It can complete the cabin visual fatigue comfort test without the participation of real people, solving the three major pain points of safety, efficiency and data stability that cannot be overcome by real people testing, and exceeding the conventional expectations of those skilled in the art for comfort testing.

[0028] (iii) Unexpected Technological Means Conventional biomimetic testing in this field only simulates body movement, while this invention couples active eye tremors, dual-mode gimbal locking, three-axis displacement, and cockpit vibration in multiple degrees of freedom, and achieves synchronous closed-loop control through a central controller. This multi-module collaborative biomimetic design is not a common practice in this field, but it achieves a real driving environment reproduction effect far exceeding that of a single module, and the technical means are non-obvious. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the robot of the present invention.

[0030] Figure 2This is a partial structural diagram of the active shaking module in the robot of the present invention.

[0031] Figure 3 This is a schematic diagram of the bionic human eyelid structure of the present invention.

[0032] Figure 4 for Figure 3 Enlarged view inside the dashed circle.

[0033] Figure 5 for Figure 4 A schematic diagram of the internal transmission structure after the casing has been removed.

[0034] The reference numerals in the accompanying drawings of the instruction manual include: motor 1, motor shaft 2, bevel gear one 3, bevel gear two 4, bevel gear three 5, fixed structure 6, cylinder 7, connecting rod one 8, double-sided bevel gear 9, telescopic shaft 10, bevel gear four 11, flat gear one 12, flat gear two 13, annular moving ring 14, sliding buckle 15, circular connecting rod two 16, fixing block 17, connecting rod three 18, upper eyelid 19, lower eyelid 20, connecting rod four 21, bevel gear five 22, pressure plate 23, U-shaped connecting rod 24, sleeve 25, spring 26, bayonet 27, three-axis moving bracket 111, and gimbal 112. Detailed Implementation

[0035] The following detailed description illustrates the specific implementation method: Example 1 This embodiment provides a testing device for dynamic simulation of an automotive cockpit, the specific structure of which is as follows: Figure 1 As shown, the device includes, from top to bottom, a bionic human eye module, an active shaking module, an adaptive gimbal module, and a three-axis motion support module.

[0036] The bionic eye module uses a Sony IMX582 1 / 2-inch CMOS image sensor, paired with a 6mm fixed-focus distortion-free industrial lens. It has a horizontal field of view of approximately 52° and a vertical field of view of approximately 39°, covering A4 paper or a 10-inch electronic screen at normal reading distances (30-50cm). Visual acquisition parameters are set as follows: resolution 1920×1080, frame rate 60fps, automatic white balance off, and fixed exposure time 1 / 120s to eliminate interference from sudden changes in ambient light.

[0037] like Figure 3-5As shown, the bionic human eye includes a miniature camera, an eyelid structure, a convex lens, and a rotating mechanism. The miniature camera is located within the eyelid structure, which is an upper and lower structure. The upper eyelid 19 and the lower eyelid 20 are connected to a miniature servo motor via a connecting rod. The servo motor controls the opening and closing of the eyelids, performing periodic blinking at a frequency of 15-20 times per minute. A central controller controls the blinking frequency and closing duration to simulate eye fatigue. The convex lens is positioned in front of the camera, with a perforation at its edge through which a rotating shaft passes. The rotating shaft is connected to a connecting rod, and the other end of the connecting rod is connected to a miniature motor, which drives the connecting rod to rotate the convex lens. The lower eyelid 20 achieves rapid closing via a flat gear 11 (or a bevel gear), simulating fast blinking. By varying the blinking speed, the blinking frequency can be simulated for different ages, road conditions, visual acuity, and levels of eye fatigue, resulting in more comprehensive and accurate test results.

[0038] like Figures 3 to 5 As shown, where Figure 4 for Figure 3 Enlarged view inside the dashed circle. Figure 5 for Figure 4 A schematic diagram of the internal transmission structure after the casing has been removed. Figure 5 The number 11 in the text refers to a spur gear (which can be understood as a bevel gear). Figure 4 The square between 19 and 20 is a miniature camera. If the illustration is not clear, you can open the CAD drawing file to view it.

[0039] The eyelid is divided into the upper eyelid 19 and the lower eyelid 20. The eyelid transmission mechanism includes a motor 1, a motor shaft 2, a bevel gear 1 3, a bevel gear 2 4, a bevel gear 3 5, a fixed structure 6, a cylinder 7, a connecting rod 1 8, a bidirectional bevel gear 9, a telescopic shaft 10, a bevel gear 4 11, a flat gear 1 12, a flat gear 2 13, an annular moving ring 14, a sliding buckle 15, a circular connecting rod 2 16, a fixed block 17, a connecting rod 3 18, a connecting rod 4 21, and a bevel gear 5 22.

[0040] Motor shaft 2 connects motor 1 and bevel gear 3. When motor 1 is running, it drives bevel gear 3 to rotate through motor shaft 2. Bevel gear 4 simultaneously meshes with bevel gear 3 and bevel gear 5, transmitting power to bevel gear 5. Bevel gear 5 drives bidirectional bevel gear 9 to rotate through telescopic shaft 10.

[0041] The upper conical surface of bevel gear 11 meshes with the upper conical surface of the double bevel gear 9. The first spur gear 12 shares the same shaft as bevel gear 11. The second spur gear 13 meshes with the first spur gear 12, and the shaft of the second spur gear 13 is fixed to the hub of the outer ring of bevel gear 11 (bevel gear 11 has a hub extending outward from its central axis, with an opening on the outer side of the hub, and the shaft of the second spur gear 13 is fixed within this opening). The shaft of the second spur gear 13 extends to the right and has a sliding buckle 15, which is fitted inside the annular moving ring 14. The annular moving ring 14 has two circular connecting rods 16 at both ends, and two fixing blocks 17 at both ends of the circular connecting rods 16. Each fixing block 17 has a through hole in the middle, with a diameter slightly larger than that of the circular connecting rods 16, allowing the circular connecting rods 16 to slide freely within the hole of the fixing block 17. The annular moving ring 14 also has a connecting rod 18, which connects to the upper eyelid 19.

[0042] The transmission path of the slow blinking action is as follows: the rotation of the bidirectional bevel gear 9 drives the bevel gear 4 11, the bevel gear 4 11 drives the spur gear 1 12, the spur gear 1 12 drives the spur gear 2 13 to rotate around the spur gear 1 12; the rotation of the spur gear 2 13 causes the sliding buckle 15 to move back and forth, which in turn drives the annular moving ring 14 to move up and down; the annular moving ring 14 drives the upper eyelid 19 to move up and down through the connecting rod 3 18, thereby realizing the slow blinking simulation of the upper eyelid.

[0043] The transmission path of the rapid blinking action is as follows: When the cylinder rod of cylinder 7 extends, it pushes the telescopic shaft 10 to extend through connecting rod 1 8, and the bidirectional bevel gear 9 moves downward accordingly. Its lower cone surface meshes with bevel gear 5 22. After bevel gear 5 22 rotates, it drives the lower eyelid 20 to move up and down through connecting rod 4 21, realizing the rapid closure of the lower eyelid and simulating the rapid blinking of the human eye.

[0044] By combining fast and slow blinking, the blinking frequency can be simulated under different ages, road conditions, visual acuity, and eye fatigue states, making the test results more comprehensive and accurate.

[0045] The bionic eye assesses fatigue by the number of blinks per minute and the duration of each blink. A normal eye blinks 15-20 times per minute. BPM ≤ 10 blinks / minute or BPM ≥ 25 blinks / minute are considered "visual inhibitory fatigue" or "reactive increased blinking," respectively, and are both included in the fatigue factor. The normal blink duration is 100-400 milliseconds. When multiple consecutive blinks last more than 450 ms or a semi-closed state (decreased eye elasticity but not complete closure) lasts more than 2 seconds, it is considered severe fatigue. Fixation stability is assessed using optical flow analysis of the root mean square error of pupil center displacement in a two-dimensional plane within a 60fps video sequence. A "fixation instability event" is recorded when the fixation point drift exceeds 1.2° of visual angle (approximately 6.3 mm at a distance of 30 cm) and lasts for more than 3 seconds under vibration. The pupil diameter change rate is also assessed under fixed illumination, where the pupil diameter fluctuates slowly due to autonomic nervous system regulation. A decrease in the diameter change rate to <0.05 mm / s indicates a decline in visual accommodation ability, aiding in the assessment of fatigue level. In summary, the fatigue quantification output is: 0.4 × (BPM_norm) + 0.3 × (BK_norm) + 0.2 × (FS_norm) + 0.1 × (PL_norm), where BPM is the blink frequency, BK is the word blink duration, FS is fixation stability, and PL is the pupil diameter change rate. The final output result is 0~10. Normalized values ​​were calculated using the Min-Max linear normalization method, mapping the original measurements of each feature to the [0,1] interval. BPM_norm was calculated based on the deviation of blink frequency from the median of the normal interval, BK_norm was calculated based on the difference between blink duration and the upper limit of normal, FS_norm was calculated based on the ratio of the frequency of gaze instability events to the maximum number that could be recorded within the test duration, and PL_norm was calculated based on the difference between the pupil diameter change rate and the lower limit of the normal change rate.

[0046] Wherein, BPM_norm (normalized blink frequency value) is 0 within the normal range, and BPM_norm=1 when the fatigue threshold is reached. The intermediate values ​​are calculated by linear interpolation. The calculation process is as follows: First, take the median value of the normal range: BPM_base = (1x1 + x2) / 2; Second, calculate the deviation: ΔBPM = |BPM BPM_base|; Third, determine the maximum deviation: ΔBPM_max = max(|x5) BPM_base|, |2x1 BPM_base|); Fourth, the normalized value: BPM_norm = ΔBPM / ΔBPM_max.

[0047] The calculation of BK_norm (the normalized value of blink duration) is as follows: If BK ≤ y1 (the normal upper limit), then BK_norm = 0; If BK ≥ y2 (the fatigue threshold), then BK_norm = 1; If y1 < BK < y2, then BK_norm = (BK y1) / (y2 y1). Here, y1 and y2 can be dynamically adjusted. The upper limit of blink time adopted in this invention is y1 = 400 milliseconds (the upper limit of normal blink duration), and y2 = 450 milliseconds (the judgment threshold for severe fatigue). This setting defines the interval from 400 ms to 450 ms as the transition zone from mild to moderate fatigue, and realizes normalization through linear interpolation, making the fatigue score continuously adjustable The calculation process of FS_norm (the normalized value of fixation stability) is as follows: First, within a fixed test duration T (such as 60 seconds), count the number of instability events N; Second, the maximum recordable number: N_max = T / 3 (each instability lasts at least 3 seconds); Third, the normalized value: FS_norm = N / N_max.

[0048] PL_norm (the normalized value of pupil diameter change rate), the lower the PL, the worse the accommodation ability. Therefore, the normalized value is inversely proportional to PL. The calculation process is as follows: If PL ≥ yx1 (the normal lower limit), then PL_norm = 0; If PL ≤ PL_min (the minimum value of severe fatigue, calibrated through pre-experiment), then PL_norm = 1; If PL_min < PL < yx1, then PL_norm = (yx1 PL) / (yx1 PL_min); Among them, yx1 = 0.05 mm / s (the normal lower limit), and PL_min = 0.01 mm / s (the minimum value of severe fatigue, calibrated through pre-experiment) The calculation process for yx1 = 0.05 mm / s and PL_min = 0.01 mm / s is as follows: As mentioned earlier, the threshold for the pupil diameter change rate is 0.05 mm / s. When the pupil diameter change rate PL decreases to <0.05 mm / s, visual accommodation ability is considered to be declining. Therefore, yx1 is the lower limit of normal, and the critical threshold is 0.05 mm / s. When PL ≥ yx1 (yx1 is 0.05 mm / s), accommodation ability is normal, and fatigue contribution is 0. Based on the pupil change rate of 0.05 mm / s, and considering that severe fatigue, under hypothetical conditions, occurs at one-fifth or one-third of the lower limit, in the actual test experiment, male youths aged 20-35 were used. After driving continuously for 4 hours at night, 20 sets of test data were collected before and after the experiment. The lower quartile of the pupil diameter change rate was 0.01. If the data value of PL_min = 0.01 mm / s is between one-third and one-fifth of 0.05, then the data meets the set value.

[0049] yx1 = 0.05 mm / s corresponds to the "threshold of pupil diameter change rate 0.05 mm / s". When the change rate is ≥0.05 mm / s, the visual accommodation ability is considered to be normal and there is no fatigue contribution. PL_min = 0.01 mm / s is the lower limit of severe fatigue calibrated in the pre-experiment. When the rate of change is ≤0.01 mm / s, the pupil has almost no spontaneous fluctuation, which is judged as severe fatigue. Linear interpolation between 0.01 and 0.05 mm / s reflects the continuous transition of fatigue level.

[0050] 9-10 points: No fatigue, excellent visual experience; 6-8 points: Mild fatigue, can be read normally; 3-5 points: Moderate fatigue, prone to errors in reading; 0-2 points: Severe fatigue, not suitable for continued viewing.

[0051] like Figure 2 As shown, the active vibration module includes a sleeve 25, a spring 26 housed within the sleeve 25, and a pressure plate 23 covering the upper end of the spring 26. One end of a U-shaped connecting rod 24 is hinged to the pressure plate 23, and the other end is connected to the piston rod of a miniature linear cylinder. The cylinder's on / off frequency and stroke are controlled by a controller (not shown). During operation, the controller drives the cylinder piston rod to extend, compressing the spring 26 by pressing down on the pressure plate 23 via the connecting rod; subsequently, the cylinder rapidly depressurizes, and the spring 26 releases its elastic potential energy, causing the pressure plate 23 and its upper structure to rebound instantaneously, generating a transient impact in the vertical direction. By programming and controlling the cylinder's operating frequency (e.g., 5-30Hz) and compression amount, vertical vibrations caused by different road surfaces can be accurately simulated.

[0052] The cylinder working air pressure is 0.3~0.6 MPa (adjustable), corresponding to a theoretical output force of 23.6~47.1 N; spring material: piano wire, wire diameter 1.2mm, mean diameter 12mm, effective number of coils 36, free length 180mm, spring constant k = 8.2 N / mm; spring pre-compression is 12mm, preload is 16.4N.

[0053] As shown in Figure 1, the adaptive gimbal module is located below the jitter module and includes a two-degree-of-freedom pivot axis for pitch and yaw. A mechanical latch 27 is located near the pivot axis, connected to a small cylinder 7. When the cylinder 7 drives the latch 27 upwards to separate from the pivot axis, the gimbal 112, driven by a servo motor, actively compensates for its attitude, maintaining bionic human eye visual stability. When the cylinder 7 drives the latch 27 downwards and embeds it into the annular groove at the bottom of the pivot axis, the pivot axis is mechanically locked, and the gimbal 112 becomes a rigid connector. This design allows for dynamic switching between "anti-shake mode" and "fixed-view mode" as needed during testing. The gimbal 112 also has a locking rod connected to a cylinder, which extends and retracts. When extended, the locking rod passes through the latch, fixing the gimbal and thus ensuring it is fixed and used only as a connector.

[0054] The adaptive gimbal module is located below the jitter module. It includes two-degree-of-freedom axes for pitch and yaw, each driven by a brushless DC servo motor (model: DJI GM6020, torque 2.5 N·m, speed 300° / s), with an encoder resolution of 0.088°. The gimbal 112 incorporates a six-axis inertial measurement unit (IMU, model: ICM-20948), including a three-axis gyroscope (range ±250 / ±500 / ±1000 / ±2000° / s, zero-bias stability ±0.5° / h) and a three-axis accelerometer (range ±2 / ±4 / ±8 / ±16g). A mechanical bayonet is located near the pivot, connecting to a small cylinder (model: Festo ADN-8-5-IPA, 8mm bore, 5mm stroke). When the cylinder drives the bayonet upwards to separate from the pivot, the gimbal, driven by a servo motor, actively compensates for its posture (based on angular velocity feedback from an IMU, using PID control with a bandwidth of 100Hz), maintaining biomimetic human eye visual stability (anti-shake mode). When the cylinder drives the bayonet downwards and engages with the annular groove at the bottom of the pivot, the pivot is mechanically locked, and the gimbal becomes a rigid connector (fixed viewing angle mode). This design allows for dynamic switching between "anti-shake mode" and "fixed viewing angle mode" as needed during testing. The three-axis moving support module uses a high-precision linear slide driven by a servo motor, specifically... Figure 1The three-axis moving bracket 111 shown features independent control in three axes (X / Y / Z), with strokes of ±600mm, ±400mm, and ±300mm respectively, and a maximum acceleration of 0.5g. The X, Y, and Z axes of the three-axis moving bracket 111 are structurally identical, employing a modular design and standardized interfaces and dimensions. The slide rail length can be replaced with different lengths depending on the situation, and both the slide rail and the slider use standard parts. The entire system, coordinated by a controller, can synchronously execute jitter, gimbal state switching, and three-axis displacement, achieving a composite simulation of human motion and visual perception within the cockpit. The central controller uses an STM32H743 main control chip and communicates with each execution unit via a CAN bus, achieving a synchronization error of ≤5ms, meeting the requirements for high dynamic coordination.

[0055] The gimbal is also modularly designed, with both mechanical and electrical interfaces set as quick-change standard interfaces, enabling rapid installation and replacement of different configurations to meet testing needs of varying precision.

[0056] Vehicle model simulation is divided into large, medium and small vehicles. Both vehicle model simulation and road simulation can change the vibration frequency of the vibrating structure to adapt to different test scenarios.

[0057] Gimbal dual-mode switching: The anti-shake mode of this invention mimics this physiological mechanism: the gimbal actively drives the motor to rotate in the opposite direction based on the vehicle vibration detected by the IMU, so that the bionic human eye's line of sight is always focused on the target (such as a mobile phone screen or a book), thereby simulating the ability of the human eye to actively stabilize its gaze in a dynamic cockpit. In this mode, the visual fatigue detection results reflect the true level of comfort that the human eye can achieve through its own adjustment; In certain test scenarios (such as verifying the extreme values ​​of image shift of the HUD under extreme turbulence, and measuring the limiting performance of the anti-shake algorithm), it is necessary to eliminate the interference of the human eye's adaptive ability and directly measure the disturbance of the visual target caused by pure physical vibration. In the locking mode, the gimbal becomes a rigid connection, and the bionic eye moves rigidly along with the vehicle body. Under these conditions, the image jitter is entirely caused by external vibration, without any compensation. This is equivalent to simulating a "pathological" or "extreme" condition without vestibular-ocular reflexes, used to evaluate the anti-shake capability of the vision device itself or to determine safety boundaries. The two modes can be dynamically switched during testing via controller presets or external signal triggers. For example, visual fatigue under normal driving conditions can be measured first in anti-shake mode, and then instantly switched to lock mode to measure the physical vibration amplitude under the same vibration conditions. Comparing the differences between the two modes quantifies the contribution of human eye's adaptive ability to comfort. This is something a single-mode gimbal cannot achieve.

[0058] Vibration, eye tremors, and three-axis motion synchronization: The controller's internal hardware timer generates a 1ms reference clock. All motion commands are timestamped. According to the preset synchronous motion curve, the vibration structure is started at t=0ms, the eye tremor is started at t=5ms, and the three-axis X-axis displacement begins at t=10ms. The commands are triggered sequentially or synchronously. The position / frequency sensor inside each execution unit transmits the actual motion state back to the controller in real time. The controller uses a PID + feedforward algorithm to correct the commands, ensuring that the synchronization error between the actual motion and the set curve is ≤5ms. A complete cycle is 20ms, and a synchronization error of 5ms is equivalent to 1 / 4 of a cycle (90° phase difference). Experiments have shown that the human eye is not sensitive to visual-motor asynchrony of less than 1 / 4 of a cycle, therefore this error range can ensure the realism of the simulation; Process: ① t=0ms: Send a CAN command to the vibration structure cylinder to start vertical vibration at 15Hz and 3mm amplitude; ② t=0ms (synchronization): Send a command to the active vibration module to start high-frequency micro-vibration at the same 15Hz frequency but with a smaller amplitude (e.g., 1mm), simulating the head's transfer function to the vehicle body vibration; ③ t=10ms: Send a command to the X-axis servo motor to start moving forward with an acceleration of 0.3g, simulating braking forward thrust; ④ t=500ms: Send a command to the Y-axis servo motor to start moving laterally at a speed of 20mm / s, simulating slow steering; ⑤ During the movement, the controller reads the data from each sensor every 5ms and compares it with the preset trajectory. If phase lag is found (e.g., the vibration structure is delayed by 1ms due to air pressure fluctuations), the solenoid valve conduction time or motor torque is immediately adjusted for compensation.

[0059] Traditional cockpit simulators often use independently controlled vibration tables and visual displays, resulting in asynchrony errors of over 50ms, which can cause motion sickness (sensory conflict) in test subjects. The 5ms high-precision synchronization of this invention effectively eliminates visual-vestibular conflict, allowing the comfort measured by the bionic eye to truly reflect the overall fatigue after multimodal fusion, rather than the abnormal response of a single modality; The bionic human eye vibrates in sync with the vehicle's movement, while its spatial position slowly changes under the influence of the three-axis moving bracket 111. The gimbal actively stabilizes the line of sight in anti-shake mode. These three components are coupled yet each has its own independent control channel. Through the unified timing and closed-loop feedback of the central controller, a high-fidelity simulation of human-vehicle-vision composite motion is achieved.

[0060] The gimbal's dual-mode switching enables rapid conversion between mechanical locking and active image stabilization via a movable bayonet, serving two detection needs: "simulated human eye adaptation" and "purely physical fixed viewing angle." Furthermore, the synchronization of vibration, eye tremors, and three-axis movement relies on the unified timing (synchronization error ≤5ms) and closed-loop feedback of the central controller, achieving precise coordination of multi-degree-of-freedom motion—one of the core technical features of this invention.

[0061] Manual testing and calibration process: Before bionic robots are put into use, comfort assessment benchmarks need to be established through manual testing experiments. The experimental principle is as follows: a vibration device is used to simulate the vibration of a car in operation, and the vibration frequency is changed for different road conditions and different types of vehicles by controlling the vibration frequency; comfort is verified by testers reading the device while it is running, and the level of comfort is determined by the accuracy of the reading.

[0062] The test parameters were calibrated using vibration structure two, which is a triaxial (XYZ) vibration structure. A seat was placed on the vibration structure, and the three axes operated unidirectionally. When the X-axis vibrated, the Y and Z axes were stationary; when the Y and Z axes were moving, the other two axes were stationary. The three axes were defined as follows: the X-axis was parallel to the sitting posture, the Y-axis was perpendicular to the sitting posture, and the Z-axis was parallel to the vertical direction of the person sitting.

[0063] The testing method was as follows: a person sat in the chair and a single-axis test was conducted, with 50 sets of data collected for each person per axis; when the second vibration structure was opened, the tester read the text aloud, and staff members checked the accuracy of the reading.

[0064] Based on the fact that the vibration frequency range of a typical automobile is 1–80 Hz, and multiple experiments have shown that the comfort vibration frequency is 1–1.5 Hz, with 20–50 Hz being a low-frequency resonance, further experiments were conducted in the 20–50 Hz range. The ranges were divided into 20–30, 30–40, and 40–50 Hz regions, with 10 experiments performed in each region. The 10 data sets were then depolarized, and 4 sets were selected. Finally, the experimental results showed that the frequency reading error rates were highest at 10 Hz (Z-axis), 13 Hz (X-axis), and 15 Hz (Y-axis).

[0065] The bionic human eye uses a camera to capture text displayed in front of it and compares it with input text, calculating the error rate C. This error rate, combined with vibration frequency, leads to a comfort level assessment. The comfort level assessment is calculated using a formula, where vibration frequency accounts for 40% and the text recognition error rate accounts for 60%. The formula is as follows: S = 0.6 × (1+C) / S_text_max + 0.4 × A × [α×(Fx² + Fy² + Fz²)] where S is the comfort index, C is the text recognition error rate, A is the normalization coefficient, and α is the number of experimental data sets. Fx, Fy, and Fz are the vibration frequencies of the X, Y, and Z axes, respectively, in Hz; the vibration frequency accounts for 0.4, and the text recognition error rate accounts for 0.6. In the formula for the comfort index, Fx, Fy, and Fz are the vibration frequencies of the X, Y, and Z axes, respectively, with units of Hz. After multiplying by the coefficients, the units are kept consistent with the units on the left side of the formula addition, allowing them to be added together.

[0066] The formula here has been abbreviated for the sake of document brevity. The original formula expanded is:

[0067] The unit is Hz 2 Its definition is A = 1 /

[0068] ① In the original format, A is defined as: A = 1 / S freq_max , ; F x,max It is the calibrated maximum vibration frequency, measured in Hz; therefore, S freq_max The unit is H. 2 Since α is dimensionless, the unit of the product is Hz. 2 ; The unit of A in the original formula is 1 / Hz. 2 .

[0069] ②A The formula for calculation is [A].

α

[0070] S_text_max = 1 + C_max, which is the maximum (1+C) value calibrated in the preliminary experiment; A = 1 / S_freq_max, where A is the normalization coefficient for the vibration frequency term.

[0071] S_freq_max is the sum of squares of the maximum vibration frequencies, calibrated in the preliminary experiment: S_freq_max = α × (Fx_max² + Fy_max² + Fz_max²); Fx_max, Fy_max, and Fz_max are the maximum vibration frequencies of each axis under the worst operating conditions, which were measured through the preliminary experiment of the vibration structure. In the results, the larger the value, the lower the comfort level.

[0072] The worst operating condition refers to the combination of vibration frequencies that causes the tester the greatest discomfort during the triaxial vibration test of the bionic robot in the pre-experimental calibration phase using Vibration Structure 2. In the triaxial single-axis vibration test, with 50 data sets per axis, the frequencies with the highest reading error rates (Z-axis 10Hz, X-axis 13Hz, Y-axis 15Hz) represent the lowest levels of comfort.

[0073] The X-axis Fx_max is calculated by gradually increasing the frequency of acceleration / braking and recording the error rate. The peak value of the error rate is Fx_max.

[0074] Fy_max is the Y-axis value. The rotation direction is changed, and the frequency is gradually increased. The error rate is recorded, and the peak value of the error rate is Fy_max.

[0075] The Z-axis Fz_max is determined by vertical turbulence, with the frequency gradually increased and the error rate recorded. The peak value of the error rate is Fz_max.

[0076] S_freq_max is based on the sum of the squares of the maximum vibration frequencies of each axis under the worst operating conditions, used as the normalization denominator to normalize the range of vibration frequency comfort components to [0,1]. When the actual frequency equals Fx_max / Fy_max / Fz_max, the component reaches its maximum value of 1, indicating that it is under the worst operating conditions.

[0077] Example 2 This embodiment provides a working method based on the device of Embodiment 1, specifically applied to vibration environment testing of vehicle head-up displays (HUDs). The steps are as follows: First, the HUD display unit under test was fixed in the simulated windshield position, and the bionic human eye lens was aligned with its projection area. At the start of the test, the controller switched the adaptive gimbal module to "anti-shake mode". The gimbal dynamically adjusted according to the built-in gyroscope data to keep the bionic human eye's line of sight perpendicular to the HUD imaging plane.

[0078] Subsequently, a composite motion test procedure was executed: the controller synchronously activated the active vibration module and the three-axis moving support module. The cylinder of the vibration module reciprocated at a frequency of 15Hz and a stroke of 3mm to simulate the mid-frequency vibration generated by a vehicle driving on a rough asphalt road surface; at the same time, the three-axis support moved along a preset trajectory, simulating acceleration and braking of ±0.3g in the X-axis direction and slow steering sway of ±200mm in the Y-axis direction.

[0079] During testing, the bionic eye continuously acquires HUD images at a rate of 60fps, and image analysis software calculates the jitter amplitude and clarity of the displayed content in real time. When it is necessary to evaluate the display stability under extreme turbulence, the controller sends a command to the gimbal module to drive the bayonet cylinder to switch the gimbal to "fixed viewing angle mode". At this time, the bionic eye is rigidly connected to the support, and the extreme values ​​of the HUI image offset under the state of no posture compensation can be recorded.

[0080] The critical flicker frequency of the human eye is approximately 50–60 Hz (i.e., at 50–60 frames per second or higher, the human eye cannot distinguish discrete images and perceive continuous motion). This solution uses 60 fps, which is close to the physiological upper limit of the human eye's temporal resolution. This makes the dynamic blur and jitter trajectory captured by the bionic eye highly consistent with the visual experience subjectively felt by real passengers. Conventional vibration measurement or image stabilization tests often use 100 fps or even higher frame rates to pursue accurate reproduction of physical parameters. However, this invention takes the opposite approach, abandoning excessively high physical precision and actively matching the perceptual bandwidth of the human eye, thereby significantly improving the correlation between the "comfort score" measured by the robot and the subjective score of real people (experimental data can prove this). This choice breaks away from the technical bias that "the more precise the measuring instrument, the better."

[0081] This embodiment achieves comprehensive performance testing of the visual stabilization system in a real driving environment by dynamically coordinating jitter, displacement, and gimbal status, and is especially suitable for verifying the environmental adaptability of visual devices in smart cockpits.

[0082] The above are merely embodiments of the present invention. Commonly known technical solutions and / or characteristics are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the technical solution of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A bionic robot for detecting the driving comfort of intelligent vehicles, characterized in that, It includes a support frame, a bionic eye, a vibration structure, a gimbal, and a central controller. The support frame is used to simulate the human body, the bionic eye is set on the upper part of the support frame, and the active shaking module is connected below the bionic eye. The vibration structure is connected to the bottom of the support frame, and the gimbal is set between the vibration structure and the support frame. The central controller is electrically connected to the bionic eye, the active shaking module, the vibration structure, and the gimbal, respectively, and is used to uniformly control the coordinated action of each module. The bionic human eye includes a miniature camera, an eyelid structure, a convex lens, and a rotating mechanism. The miniature camera is located inside the eyelid structure, which is an upper and lower structure connected to a miniature servo motor via a connecting rod. The servo motor controls the opening and closing of the eyelid, performing periodic blinking at a frequency of 15-20 times per minute. A central controller controls the blinking frequency and closing duration to simulate eye fatigue. The convex lens is located in front of the camera, with a perforation at its edge through which a rotating shaft passes. The rotating shaft is connected to a connecting rod, and the other end of the connecting rod is connected to a miniature motor, which drives the connecting rod to rotate the convex lens. The lower eyelid achieves rapid closing through a flat gear transmission, simulating fast blinking. By varying the blinking speed, the blinking frequency can be simulated for different ages, road conditions, visual acuity, and levels of eye fatigue. The central controller has a built-in visual fatigue quantification assessment unit, which is configured to perform the following visual fatigue quantification assessment steps: The first step was to collect the blinking dynamics, half-closed eye state features, gaze stability features, and pupil diameter change rate (PL) of the bionic human eye in a 60fps video sequence. Among them, the blink dynamic characteristics include blink count per minute (BPM) and single blink duration (BK), with the normal blink duration set to 100-400 milliseconds; when the duration of multiple consecutive blinks exceeds 450ms, it is judged as a severe fatigue characteristic. Half-closed eye state characteristics: When the eye posture estimation EAR value decreases but the eyes are not completely closed and the duration exceeds 2 seconds, it is judged as a severe fatigue characteristic; Fixation stability characteristics: The root mean square error of the pupil center displacement in the two-dimensional plane was analyzed using optical flow. When the fixation point drift exceeds 1.2° of visual angle and lasts for more than 3 seconds under vibration, a "fixation instability event" is recorded. Pupil diameter change rate (PL): The pupil diameter change rate is calculated under fixed lighting conditions. When the diameter change rate decreases to <0.05 mm / s, it is determined that the visual accommodation ability has decreased. The second step involves the visual fatigue quantification assessment unit outputting a fatigue score of 0 to 10 based on the following fatigue quantification model: Score = 0.4 × (BPM_norm) + 0.3 × (BK_norm) + 0.2 × (FS_norm) + 0.1 × (PL_norm) Wherein, BPM_norm is the normalized value of blink frequency, BK_norm is the normalized value of blink duration, FS_norm is the normalized value of fixation stability, and PL_norm is the normalized value of pupil diameter change rate. The third step is for the central controller to classify fatigue levels based on fatigue scores: 9-10 points indicate no fatigue, 6-8 points indicate mild fatigue, 3-5 points indicate moderate fatigue, and 0-2 points indicate severe fatigue. Based on these levels, the controller controls the active shaking module, vibration structure, and gimbal adjustment simulation strategy. Vehicle model simulation is divided into large, medium and small vehicles. Both vehicle model simulation and road simulation can change the vibration frequency of the vibrating structure.

2. The biomimetic robot of claim 1, wherein, The vibration structure includes a high-response miniature cylinder, a U-shaped connecting rod, a pressure plate, a spring, and a sleeve; the cylinder piston rod is rigidly connected to the U-shaped connecting rod, the pressure plate is fixed to the end of the U-shaped connecting rod and abuts against the top of the spring, and the spring is fitted inside the sleeve; the cylinder air supply frequency is 5Hz to 30Hz, and the amplitude adjustment range is 0.1mm to 5mm.

3. The biomimetic robot of claim 1, wherein, The gimbal has an integrated electromechanical locking mechanism inside. The locking mechanism is driven by an auxiliary cylinder. The locking mechanism moves up and down along the gimbal's connecting shaft by 3mm to 10mm. When the locking mechanism moves up to the top of the connecting shaft, the gimbal is in active anti-shake mode. When the locking mechanism moves down to the bottom groove of the connecting shaft, the gimbal is in rigid locking mode. The gimbal also has a locking rod, which is connected to a cylinder. The cylinder extends and retracts the rod. When extended, the locking rod passes through the locking mechanism to fix the gimbal.

4. The biomimetic robot of claim 1, wherein, The support frame features a three-axis movement structure with an X-axis travel of -600mm to 600mm, a Y-axis travel of -400mm to 400mm, and a Z-axis travel of -300mm to 300mm, designed to simulate the natural displacement of a human body within a car cabin. The three-axis movement support frame has identical structures in all three directions (X, Y, and Z), employing a modular design with standardized interfaces and dimensions. The slide rail length can be changed to suit different needs, and both the slide rail and slider utilize standard parts. The gimbal is also modularly designed, with quick-change standard interfaces for both mechanical and electrical connectors.

5. The biomimetic robot of claim 2, wherein, The active shaking module of the bionic human eye and the vibration structure are synchronously controlled by the central controller, and the synchronization error between the two is ≤0.01s.

6. The biomimetic robot of claim 1, wherein, The central controller has a built-in scenario script module and closed-loop feedback module, which supports preset continuous bump and emergency lane change test scenarios, and the sensor feedback response time is ≤0.05s.

7. The intelligent automobile driving comfort detection bionic robot working method is characterized in that, The implementation of the bionic robot based on any one of claims 1-6 includes the following steps: S1. The central controller presets vibration parameters, pan-tilt working mode, support movement trajectory, and visual fatigue judgment thresholds; the visual fatigue judgment thresholds include: normal blinking frequency range [15, 20] times / minute, blinking frequency ≤10 times / minute or ≥25 times / minute correspond to "visual inhibitory fatigue" or "stress-induced increased blinking" respectively; normal blinking duration 100-400 milliseconds; gaze instability trigger threshold 1.2° visual angle and lasting for 3 seconds; pupil diameter change rate threshold 0.05 mm / s; S2. The vibration structure simulates the dynamic shaking of the cockpit. The gimbal switches to the corresponding mode, the three-axis support simulates the displacement of the human body in the cockpit, and the active shaking module moves synchronously with the vibration structure. S3. The bionic human eye acquires visual signals in real time, and the central controller executes the following sub-steps: Calculate blink rate (BPM). If BPM ≤ 10 or BPM ≥ 25, include it in the fatigue factor. The duration of a single blink is measured. If the duration of multiple consecutive blinks exceeds 450 ms or the half-closed state lasts for more than 2 seconds, it is judged as severe fatigue. The root mean square error of pupil center displacement in a 60fps video sequence was analyzed using optical flow method to detect the frequency of gaze instability events. Calculate the pupil diameter change rate PL under fixed illumination and compare it with the threshold of 0.05 mm / s; The text displayed in front of the camera is captured and compared with the input text to calculate the text recognition error rate C. S4. The central controller inputs the collected data into the fatigue quantification model to calculate the fatigue score, compares it with the preset threshold, outputs the visual fatigue comfort quantification detection result, and adaptively adjusts the simulation strategy according to the score level. Meanwhile, the central controller calculates the overall cabin comfort based on the following comfort index formula: S = 0.6 × (1+C) / S_text_max + 0.4 × A × [α×(Fx² + Fy² + Fz²)] Where S is the comfort index, C is the text recognition error rate, α is the number of experimental data groups, Fx, Fy, and Fz are the vibration frequencies of the X, Y, and Z axes, respectively, in Hz; the vibration frequency accounts for 0.4, and the text recognition error rate accounts for 0.

6. S_text_max = 1 + C_max, which is the maximum (1+C) value calibrated in the preliminary experiment; A = 1 / S_freq_max, where A is the normalization coefficient for the vibration frequency term; S_freq_max is the sum of squares of the maximum vibration frequencies. Preliminary experimental calibration: S_freq_max = α × (Fx_max² + Fy_max² + Fz_max²); Fx_max, Fy_max, and Fz_max are the maximum vibration frequencies of each axis, respectively.

8. The working method according to claim 7, characterized in that, The cylinder ventilation frequency of the vibration structure in S2 is set to 20Hz and the amplitude is set to 1mm to simulate the driving conditions of a standard asphalt road.

9. The method of claim 7, wherein, In S2, the control gimbal bayonet moves down to the bottom groove of the coupling, the gimbal remains in a rigid locked state, and the bionic human eye continuously collects visual signals from a fixed angle.

10. The method of claim 7, wherein, The S3's bionic eye captures data at a frequency of 30 frames per second, and the central controller outputs a 0-10 point quantitative comfort score, with a lower score indicating a higher degree of visual fatigue.

Images