A trackable electric spotlight and a control system thereof

Abstract

This invention discloses a track-mounted electric spotlight and its control system, belonging to the field of lighting control technology. It aims to solve the technical problem of existing track-mounted electric spotlights being unable to actively sense and predict user gaze movement, resulting in asynchronous light spot and gaze, and a lack of differentiated lighting response based on eye movement. The invention includes: a spotlight housing with an opening for illumination, and a lens arranged within the opening; an LED light source module disposed within the spotlight housing; a drive circuit board disposed within the spotlight housing and electrically connected to the LED light source module; and an electric rotating bracket connected to the spotlight housing, wherein a rotation drive motor is disposed within the electric rotating bracket for driving the spotlight housing to rotate horizontally. This invention achieves precise synchronous tracking of the light spot position and the gaze focus through automatic classification and differentiated predictive control of eye movement, effectively solving the technical problem of "vision-light asynchrony" in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of lighting control technology, and more specifically, to a track-mounted electric spotlight and its control system. Background Technology

[0002] Track lighting is widely used in museums, art galleries, commercial exhibition halls, retail stores, and other locations requiring focused lighting due to its advantages such as flexible installation, adjustable beam direction, and controllable beam pattern. Traditional track lighting typically requires manual adjustment, necessitating users to climb to a higher position using ladders or similar tools to adjust the horizontal rotation and tilt angles of each spotlight individually to align the beam with the target exhibit or area. This adjustment method is not only cumbersome and inefficient, but also requires repeated manual calibration, especially in scenarios where exhibits frequently change, significantly increasing maintenance costs.

[0003] In recent years, with the development of smart lighting technology, motorized track spotlights have emerged that can be controlled by remote control or mobile application to rotate the motor, allowing users to remotely adjust the direction of the light spot without climbing. However, these motorized spotlights still rely on manual operation—users need to observe the illuminated area in real time and gradually adjust the position of the light spot by repeatedly clicking the direction button or dragging the control lever on the screen until the light spot coincides with the target area. This closed-loop control method of "human eye observation - brain judgment - manual operation" has significant operation delays and positioning errors, especially when it is necessary to quickly switch the illumination target or continuously track moving objects, manual operation can hardly meet the real-time requirements.

[0004] More importantly, existing motorized track lighting technology completely lacks the ability to perceive the user's visual attention. In practical applications, the user's gaze often reaches the target location before manual operation—for example, when a visitor quickly scans different exhibits in an exhibition hall, their gaze has already moved to the next exhibit, while the spotlight's beam remains on the previous exhibit, creating a disconnected experience of "view-light asynchrony." This asynchrony not only reduces the immersiveness of the lighting experience but also results in a lack of effective lighting during the user's gaze shift, requiring additional ambient light to compensate and causing unnecessary energy waste. Therefore, we propose a track-mounted motorized spotlight and its control system. Summary of the Invention

[0005] The purpose of this invention is to provide a track-mounted electric spotlight and its control system to solve the technical problem that existing track-mounted electric spotlights cannot actively sense and predict the movement of the user's line of sight, resulting in the light spot being out of sync with the line of sight and lacking differentiated lighting response based on eye movement behavior.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a track-mounted electric spotlight, comprising:

[0007] The spotlight housing has an opening for illumination, and a lens is arranged in the opening;

[0008] An LED light source module is disposed inside the spotlight housing;

[0009] A drive circuit board is disposed inside the spotlight housing and is electrically connected to the LED light source module;

[0010] An electric rotating bracket is connected to the spotlight housing, and a rotary drive motor is installed inside the electric rotating bracket to drive the spotlight housing to rotate in the horizontal direction;

[0011] A pitch adjustment base is connected between the spotlight housing and the electric rotating bracket. A pitch drive motor is installed inside the pitch adjustment base to drive the spotlight housing to swing in the vertical direction.

[0012] An eye-tracking acquisition module is installed on the housing of the spotlight and is used to acquire the coordinates of the gaze focus of the target person.

[0013] The drive circuit board includes a processor and a memory. When the processor executes a computer program stored in the memory, it performs the following steps:

[0014] Obtain the coordinates of the gaze focus in multiple consecutive frames acquired by the eye-tracking acquisition module;

[0015] Eye movement feature parameters are calculated based on the gaze focus coordinates of multiple consecutive frames. The eye movement feature parameters include angular velocity, angular acceleration, and gaze duration.

[0016] Based on the eye movement feature parameters, eye movement behavior is classified into saccades, smooth tracking, or fixation.

[0017] When classified as saccade, the coordinates of the saccade endpoint are predicted based on the coordinates of the line of sight at the saccade starting point and the saccade direction.

[0018] When classified as smooth tracking, the coordinates of future trajectory points are predicted based on the gaze focus coordinates of multiple consecutive frames.

[0019] When classified as fixation, maintain the current spot position;

[0020] Based on the prediction results, the rotary drive motor and the pitch drive motor are controlled to move the center of the spotlight to the predicted position.

[0021] This invention acquires the user's gaze focus coordinates in real time through an eye-tracking module. Combined with a drive circuit board, it automatically classifies and differentially predicts three eye-tracking behaviors: saccades, smooth tracking, and fixation. This achieves precise synchronous tracking of the light spot position and the gaze focus, effectively solving the technical problem of "vision-light asynchrony" in existing technologies. Specifically, when the user's gaze is in a saccade state (i.e., rapidly shifting from one fixation point to another), traditional lighting equipment cannot predict the saccade endpoint, causing the light spot to lag behind the gaze movement during the saccade, leaving the user in a dark area when reaching the new fixation point. This invention calculates the saccade amplitude in real time based on the peak angular velocity during the saccade process using the Main sequence formula and predicts the saccade endpoint coordinates based on the saccade direction. At the moment the saccade ends, the light spot is precisely projected to the endpoint position, achieving a "light ahead of the eye" pre-illumination effect. When the user's gaze is in a smooth tracking state (i.e., continuously following a moving target), this invention uses polynomial fitting and extrapolation formulas to predict future trajectory points, allowing the light spot to smoothly "attach" to the gaze trajectory. When the user's gaze is fixed, this invention uses low-pass filtering to suppress high-frequency jitter caused by micro-eye movements, maintaining a stable light spot. Differential control across three behavior modes ensures that the light spot remains synchronized with the user's gaze regardless of how the user moves their eyes, completely eliminating the delay and error of manual control and significantly improving the real-time performance and accuracy of illumination tracking.

[0022] Preferably, the eye-tracking acquisition module is a binocular eye-tracking camera, which is disposed on the front end face of the spotlight housing.

[0023] Preferably, an inertial measurement unit is also provided inside the spotlight housing. The inertial measurement unit is electrically connected to the drive circuit board and is used to collect the head posture angle of the target person.

[0024] When the processor executes the computer program, it also performs the following steps: acquiring the head posture angle collected by the inertial measurement unit; when the confidence level of the gaze focus coordinates collected by the eye-tracking acquisition module is lower than a preset threshold, calculating the gaze direction based on the head posture angle, and switching to head posture following mode.

[0025] Preferably, an ambient light sensor is also provided inside the spotlight housing. The ambient light sensor is electrically connected to the drive circuit board and is used to collect ambient illuminance.

[0026] When the processor executes the computer program, it also performs the following steps: acquiring the ambient illuminance collected by the ambient light sensor; when the rate of change of ambient illuminance exceeds a preset threshold, adjusting the exposure parameters of the eye-tracking acquisition module and recalibrating the gaze focus coordinates.

[0027] Preferably, the step of predicting the coordinates of the scanning endpoint specifically includes: calculating the scanning amplitude based on the scanning start point and the peak angular velocity during the scanning process; determining the scanning direction based on the direction of the line connecting the scanning start point and the scanning end point; and generating the coordinates of the scanning endpoint based on the coordinates of the scanning start point, the scanning amplitude, and the scanning direction.

[0028] Among them, the scanning range With peak angular velocity The relationship between them is given by the Main sequence formula. Express, scan the destination coordinates From the coordinates of the starting point of the scan Sweep range The line of sight between the target personnel and the projection plane of their line of sight and the unit vector of the scanning direction Through formula The calculation yields the following result: Indicates the personalized slope coefficient; This indicates the peak angular velocity during the scanning process; This represents the personalized intercept coefficient.

[0029] A track-mounted electric spotlight control system for controlling at least one track-mounted electric spotlight, comprising:

[0030] The control unit is communicatively connected to both the eye-tracking acquisition module and the drive circuit board.

[0031] The control unit includes:

[0032] The eye-tracking feature extraction module is used to calculate angular velocity, angular acceleration, and gaze duration based on the gaze focus coordinates of multiple consecutive frames acquired by the eye-tracking acquisition module.

[0033] The behavior classification module is used to classify the eye movement behavior of the target person into saccades, smooth tracking, or fixation based on the angular velocity, angular acceleration, and fixation duration.

[0034] The saccade prediction module is used to predict the coordinates of the saccade endpoint based on the coordinates of the line of sight at the saccade starting point and the saccade direction when the saccade is classified as a saccade.

[0035] The smooth tracking prediction module is used to predict the coordinates of future trajectory points based on the gaze focus coordinates of multiple consecutive frames when the tracking is classified as smooth.

[0036] A gaze-holding module is used to maintain the current spot position when the image is classified as gazed.

[0037] The motor control module is used to control the rotation drive motor and the pitch drive motor according to the output results of the saccade prediction module, the smooth tracking prediction module or the gaze holding module, so that the center of the spotlight moves to the predicted position.

[0038] Preferably, the saccade prediction module includes:

[0039] The scanning amplitude calculation unit is used to calculate the scanning amplitude based on the scanning start point and the peak angular velocity during the scanning process;

[0040] The scanning direction determination unit is used to determine the scanning direction based on the direction of the line connecting the scanning start point and the scanning end point.

[0041] The endpoint coordinate generation unit is used to generate the endpoint coordinates of the scan based on the coordinates of the scan start point, scan amplitude, and scan direction.

[0042] Preferably, the smooth tracking prediction module includes:

[0043] The trajectory fitting unit is used to fit motion trajectory curves based on the coordinates of the gaze focus in multiple consecutive frames.

[0044] The extrapolation prediction unit is used to extend the motion trajectory curve in the positive direction of the time axis to generate the coordinates of future trajectory points.

[0045] Among them, trajectory fitting and prediction adopt The order polynomial with respect to the nearest Frame view focus coordinates The fitting curve equation is as follows: , Future moments Coordinates of the predicted trajectory points From extrapolation formula , We obtain the following formula: , A fitting function representing the change of the coordinates of the line of sight focus over time; , express coordinates and Fitting coefficients of the coordinate polynomial; It is a time variable.

[0046] Preferably, the gaze-maintaining module includes:

[0047] The micro-eye-tracking filter unit is used to perform low-pass filtering on the coordinates of the gaze focus during fixation to suppress high-frequency jitter.

[0048] The spot locking unit is used to lock the center position of the spot based on the filtered line-of-sight focus coordinates;

[0049] The low-pass filter uses a first-order low-pass digital filter, and the filtering formula is as follows:

[0050] ;

[0051] The spot locking unit will filter the coordinates The target location as the center of the light spot;

[0052] In the formula, For the first The coordinates of the gaze focus after frame filtering; These are the filter coefficients; For the first The raw gaze focus coordinates are directly output by the frame eye-tracking acquisition module; These are the filtered coordinates from the previous frame.

[0053] Preferably, the control unit further includes:

[0054] The head pose fusion module is used to obtain the head pose angle of the target person;

[0055] The modality switching module is used to switch to head posture following mode when the confidence level of the gaze focus coordinates acquired by the eye-tracking acquisition module is lower than a preset threshold, and to calculate the gaze direction based on the head posture angle.

[0056] Compared with the prior art, the beneficial effects of the present invention are:

[0057] 1. This invention acquires the user's gaze focus coordinates in real time through an eye-tracking acquisition module. Combined with the automatic classification and differentiated predictive control of three eye-tracking behaviors—saccades, smooth tracking, and fixation—by a drive circuit board, it achieves precise synchronous tracking of the light spot position and the gaze focus, effectively solving the technical problem of "vision-light asynchrony" in existing technologies. Specifically, when the user's gaze is in a saccade state (i.e., rapidly shifting from one fixation point to another), traditional lighting equipment cannot predict the saccade endpoint, causing the light spot to lag behind the gaze movement during the saccade process, leaving the user in a dark area when reaching the new fixation point. This invention calculates the saccade amplitude in real time based on the peak angular velocity during the saccade process using the Main sequence formula, and predicts the coordinates of the saccade endpoint by combining the saccade direction. At the moment the saccade ends, the light spot is precisely projected to the endpoint position, achieving a "light ahead of the eye" pre-illumination effect. When the user's gaze is in a smooth tracking state (i.e., continuously following a moving target), this invention uses polynomial fitting and extrapolation formulas to predict future trajectory points, allowing the light spot to smoothly "attach" to the gaze trajectory. When the user's gaze is fixed, this invention uses low-pass filtering to suppress high-frequency jitter caused by micro-eye movements, maintaining a stable light spot. Differential control across three behavior modes ensures that the light spot remains synchronized with the user's gaze regardless of how the user moves their eyes, completely eliminating the delay and error of manual control and significantly improving the real-time performance and accuracy of illumination tracking.

[0058] 2. This invention also adaptively adjusts the beam angle and output power of the spotlight under different eye-tracking behavior modes, significantly reducing energy consumption in unnecessary areas while ensuring illumination quality in the focal area. This further solves the energy waste problem caused by ineffective spot movement and wide-angle illumination during traditional tracking lighting in the scanning process. Specifically, when the invention detects that the user is in a scanning state, it automatically narrows the beam angle to its minimum (focusing mode) and reduces the output power to a low power threshold; while in fixation or smooth tracking states, it widens the beam angle and restores high power output. Narrowing the beam angle during scanning ensures that the spot only covers a very small area of ​​the line of sight during rapid movement, avoiding glare interference and ineffective illumination caused by the spot sweeping across a large irrelevant area under a wide beam angle. More importantly, this invention utilizes the physiological characteristic that the human eye is insensitive to changes in brightness during scanning, actively reducing the output power during scanning—the greater the scanning amplitude, the greater the power reduction. Because the user's visual sensitivity decreases during scanning, this power reduction operation does not cause a significant change in subjective brightness perception, thus achieving a deep energy-saving effect of "saving light without the user's knowledge." During the fixation and smooth tracking phases, users require high-precision, high-brightness illumination. At this point, full power output is restored to ensure the lighting quality for critical visual tasks remains unaffected. This asymmetric power control strategy allows the lighting system's energy consumption to precisely match the user's visual needs, significantly reducing overall energy consumption over long-term use compared to traditional continuous full-power tracking solutions.

[0059] 3. This invention further addresses the issues of single-eye tracking failing under interference scenarios such as blinking and sudden changes in lighting, as well as the decrease in prediction accuracy caused by differences in eye movement characteristics among different users, by introducing an inertial measurement unit to achieve head posture following mode switching when eye tracking fails, and by updating user eye movement feature parameters online through a personalized learning module. Specifically, when the confidence level of the gaze focus coordinates of the eye movement acquisition module falls below a preset threshold due to blinking, drastic changes in ambient illumination, or glare from user-worn glasses, this invention automatically switches to head posture following mode: the inertial measurement unit acquires the user's head posture angle, and the gaze direction is inferred based on the head rotation direction to maintain coarse illumination following. This multimodal redundant perception mechanism ensures that the system can still provide continuous illumination services when eye tracking fails briefly, and smoothly switches back to the main mode after eye tracking is restored, significantly improving the robustness and reliability of the system. At the same time, this invention identifies user identity through a personalized learning module and independently stores the eye movement feature parameters (including Main sequence coefficients, polynomial fitting order, prediction time step, etc.) for each user. After each saccade, the system uses the error between the actual measured saccade amplitude and the predicted amplitude to iteratively correct the personalized coefficients through an online update formula. With increased usage, the prediction model gradually converges to the user's true eye movement characteristics, continuously improving the accuracy of saccade endpoint prediction and achieving adaptive lighting that becomes more accurate with use. This personalized learning mechanism, which requires no offline training and is completed entirely automatically in the background, allows the same lighting system to adapt to diverse user groups of different ages and eye movement habits, fundamentally solving the problem of insufficient prediction accuracy in general-purpose models. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the overall structure of the present invention;

[0061] Figure 2 This is a schematic diagram of the overall structure of the present invention from a downward viewing angle;

[0062] Figure 3 This is a schematic diagram of the internal structure of the present invention;

[0063] Figure 4 This is a schematic diagram of the main method for controlling the track-mounted electric spotlight of the present invention;

[0064] Figure 5 This is a schematic diagram of the eye-movement behavior classification and prediction sub-process of the present invention;

[0065] Figure 6 This is a schematic diagram of the multimodal sensing fusion sub-process of the present invention;

[0066] Figure 7 This is a schematic diagram of the software framework of the track-type electric spotlight control system of the present invention.

[0067] Explanation of the labels in the diagram:

[0068] 1. Spotlight housing; 2. LED light source module; 3. Electric rotating bracket; 4. Rotation drive motor; 5. Pitch adjustment base; 6. Pitch drive motor; 7. Eye-tracking acquisition module; 8. Ambient light sensor; 9. Lens. Detailed Implementation

[0069] Example 1: As Figures 1 to 6 As shown, the present invention relates to a track-mounted electric spotlight, comprising:

[0070] The spotlight housing 1 has an opening for illumination, and a lens 9 is arranged in the opening;

[0071] In an embodiment of the present invention, an inertial measurement unit is further provided inside the spotlight housing 1. The inertial measurement unit is electrically connected to the drive circuit board and is used to collect the head posture angle of the target person.

[0072] The processor, when executing a computer program, also performs the following steps:

[0073] Obtain the head attitude angles acquired by the inertial measurement unit;

[0074] When the confidence level of the gaze focus coordinates acquired by the eye-tracking acquisition module 7 is lower than a preset threshold, the gaze direction is calculated based on the head posture angle, and the system switches to head posture following mode.

[0075] In an embodiment of the present invention, when the gaze tracking confidence level is lower than a preset threshold, the modality switching module switches to head posture following mode. In this mode, the processor uses the head pitch angle collected by the inertial measurement unit... and yaw angle The line-of-sight direction is calculated using a preset linear mapping function: line-of-sight pitch angle. line of sight yaw angle ,in and This is a scaling factor, ranging from 0.6 to 1.0. The specific value can be quickly calibrated during the initialization phase based on individual user differences: the user is required to look directly in front and to the left and right side calibration points, and the scaling factor is calculated based on the correspondence between the head rotation angle and the actual line of sight. and .

[0076] In an embodiment of the present invention, an ambient light sensor 8 is further provided inside the spotlight housing 1. The ambient light sensor 8 is electrically connected to the drive circuit board and is used to collect ambient illuminance.

[0077] The processor, when executing a computer program, also performs the following steps:

[0078] The ambient light sensor 8 collects the ambient illuminance.

[0079] When the rate of change of ambient illuminance exceeds a preset threshold, the exposure parameters of the eye-tracking acquisition module 7 are adjusted, and the gaze focus coordinates are recalibrated.

[0080] LED light source module 2 is disposed inside the spotlight housing 1;

[0081] A drive circuit board is disposed inside the spotlight housing 1 and is electrically connected to the LED light source module 2;

[0082] An electric rotating bracket 3 is connected to the spotlight housing 1. The electric rotating bracket 3 is equipped with a rotary drive motor 4, which is used to drive the spotlight housing 1 to rotate in the horizontal direction.

[0083] A pitch adjustment base 5 is connected between the spotlight housing 1 and the electric rotating bracket 3. A pitch drive motor 6 is provided inside the pitch adjustment base 5 to drive the spotlight housing 1 to swing in the vertical direction.

[0084] An eye-tracking acquisition module 7 is mounted on the spotlight housing 1 and is used to acquire the coordinates of the gaze focus of the target person.

[0085] In an embodiment of the present invention, the eye-tracking acquisition module 7 is a binocular eye-tracking camera, which is disposed on the front end face of the spotlight housing 1.

[0086] In an embodiment of the present invention, the binocular eye-tracking camera includes two infrared cameras and multiple infrared LED light sources. The infrared LED light sources project infrared light onto the user's eyes, forming a Purkinje spot on the corneal surface. When the processor executes the computer program, it calculates the coordinates of the gaze focus through the following steps:

[0087] First, the center position of the pupil and the center position of the corneal reflective spot are detected from the eye images captured by the infrared camera;

[0088] Secondly, calculate the offset vector between the center of the pupil and the center of the corneal reflective spot;

[0089] Then, according to a preset gaze mapping function (pre-calibrated via polynomial regression), the offset vector is mapped to the gaze focus coordinates on the screen or in space. When a binocular structure is used, the viewing distance between the user and the gaze projection plane is calculated using the binocular parallax principle. .

[0090] The calibration process of the gaze mapping function is as follows: During system initialization, calibration points in the control screen or space are lit sequentially (e.g., a 9-point or 16-point array), and the user gazes at each calibration point in turn. The processor records the pupil-spot offset vector corresponding to each calibration point. and calibration point spatial coordinates A second-order polynomial regression model is used: , The coefficients are solved using the least squares method. , Complete the mapping function calibration.

[0091] The drive circuit board includes a processor and a memory. When the processor executes a computer program stored in the memory, it performs the following steps:

[0092] The eye-tracking acquisition module 7 acquires the coordinates of the gaze focus in multiple consecutive frames.

[0093] Eye movement feature parameters are calculated based on the gaze focus coordinates of multiple consecutive frames. The eye movement feature parameters include angular velocity, angular acceleration, and gaze duration.

[0094] Based on the eye movement feature parameters, eye movement behavior is classified into saccades, smooth tracking, or fixation.

[0095] When classified as saccade, the coordinates of the saccade endpoint are predicted based on the coordinates of the line of sight at the saccade starting point and the saccade direction.

[0096] In embodiments of the present invention, the step of predicting the coordinates of the scanning endpoint specifically includes:

[0097] The scanning amplitude is calculated based on the starting point of the scanning and the peak angular velocity during the scanning process;

[0098] Determine the scanning direction based on the direction of the line connecting the starting and ending points of the scanning.

[0099] The coordinates of the scanning endpoint are generated based on the coordinates of the scanning start point, scanning range, and scanning direction.

[0100] Among them, the scanning range With peak angular velocity The relationship between them is expressed by the Main sequence formula:

[0101] ;

[0102] Among them, scanning the endpoint coordinates From the coordinates of the starting point of the scan Sweep range The line of sight between the target personnel and the projection plane of their line of sight and the unit vector of the scanning direction The calculation yielded:

[0103] ;

[0104] In the formula:

[0105] It indicates the scanning range, representing the angle or spatial distance traversed by the line of sight from the starting point to the ending point;

[0106] This represents the personalized slope coefficient, which characterizes the proportional relationship between the user's scanning speed and amplitude.

[0107] This represents the peak angular velocity during the saccade process, i.e., the maximum angular velocity of the line of sight movement;

[0108] This represents the personalized intercept coefficient, used to compensate for the velocity-amplitude offset at the beginning of the saccade.

[0109] Indicates the coordinates of the focal point of the line of sight at the starting point of the scan;

[0110] This indicates the predicted coordinates of the scan endpoint;

[0111] The line-of-sight distance between the target person and the projection plane (can be pre-calibrated or estimated in real time).

[0112] The unit vector representing the scanning direction points roughly in the direction from the starting point to the ending point.

[0113] Explanation of the operational logic: First, based on the peak angular velocity during the scanning process... The amplitude of saccades is calculated using a linear relationship (Main sequence). This linear relationship is determined by the individualized coefficient. and The decision reflects the inherent ratio between the eye movement characteristics of different users (such as the ratio between saccade speed and amplitude). Then, the coordinates of the saccade initiation point are used. Calculated saccadic amplitude and the unit vector of the scanning direction The predicted coordinates of the scan endpoint are obtained through vector addition. The entire calculation maps eye-tracking parameters to spatial coordinates. Using the formula above, the system can determine the coordinates based on the user's unique eye-tracking characteristics (coefficients). and This system predicts the endpoint of the scanning process in real time, ensuring the light spot precisely reaches the target position the instant the scanning ends, eliminating the lag characteristic of traditional follow-lighting. The computational logic combines biomechanical principles (Main sequence) with personalized parameters, ensuring both the biological rationality of the prediction and individualized adaptation. This significantly improves the continuity of lighting and visual comfort during the scanning process, while avoiding energy waste caused by ineffective light spot movement during scanning.

[0114] It should be noted that the scanning range in this embodiment... Viewpoint (radians) is used, and the coordinates of the line-of-sight focal point are in spatial coordinates (millimeters or pixels). Both are determined by the viewing distance. Transformation: Spatial displacement = (Approximate at a small angle). View distance Depth data can be obtained through the depth sensor in the eye-tracking acquisition module or through pre-calibration.

[0115] In an embodiment of the present invention, the line of sight... The acquisition methods include any of the following:

[0116] Method 1: The distance between the user and the projection plane of the gaze is calculated in real time using the binocular parallax principle of the binocular eye-tracking camera. Specifically, based on the parallax of the same feature point in the eye images captured by the left and right cameras, combined with the baseline distance and focal length of the two cameras, the viewing distance is calculated using the trigonometric ranging formula. ;

[0117] Method 2: During the system initialization phase, the user is required to gaze at a preset calibration point. The viewing distance is then calculated based on the spatial coordinates of the calibration point and the detected line-of-sight direction. And store this value in memory as a fixed line of sight;

[0118] Method 3: Integrate a time-of-flight (ToF) depth sensor into the spotlight housing to directly measure the distance between the user and the spotlight as an approximation of the line-of-sight L.

[0119] When classified as smooth tracking, the coordinates of future trajectory points are predicted based on the gaze focus coordinates of multiple consecutive frames.

[0120] In embodiments of the present invention, the step of predicting the coordinates of future trajectory points specifically includes:

[0121] The motion trajectory curve is fitted based on the gaze focus coordinates of multiple consecutive frames;

[0122] Extend the motion trajectory curve forward along the time axis to generate the coordinates of future trajectory points.

[0123] Among them, adopt The order polynomial with respect to the nearest Frame view focus coordinates The fitting curve equation is as follows:

[0124] , ;

[0125] Among them, the future moment Coordinates of the predicted trajectory points The extrapolation formula yields:

[0126] , ;

[0127] In the formula:

[0128] , A fitting function representing the change of the coordinates of the line of sight focus over time;

[0129] The order of the polynomial fitting determines the complexity of the trajectory curve;

[0130] , express coordinates and Fitting coefficients of the coordinate polynomial;

[0131] It is a time variable;

[0132] Indicates future time The coordinates of the predicted trajectory points;

[0133] The predicted time is equal to the latest current time. Add prediction time step .

[0134] Explanation of operational logic: First, in terms of time... As the independent variable, respectively, for the focal point of the line of sight coordinates and Coordinate Construction A polynomial fitting model of order X. Using the least squares method, utilizing the nearest... Historical data of frames Solving for polynomial coefficients and After fitting, the future times will be... Substituting the two polynomials, the coordinates of the predicted trajectory points are calculated. This operation essentially extends the discrete gaze trajectory into a continuous time function and uses function extrapolation to achieve advanced prediction. Polynomial fitting and extrapolation formulas enable the system to continuously predict future trajectories based on historical patterns of gaze movement during smooth tracking, achieving synchronized and smooth movement of the light spot and the gaze focus. Compared to simple linear extrapolation, higher-order polynomials can better fit complex curvilinear motions (such as arc scanning or variable-speed tracking), thus significantly improving prediction accuracy. This operational logic ensures that the illumination spot remains "attached" to the moving target being observed by the user, enhancing illumination tracking performance in dynamic scenes while reducing energy consumption from repeated adjustments due to trajectory prediction errors.

[0135] When classified as fixation, maintain the current spot position;

[0136] In embodiments of the present invention, the step of maintaining the current light spot position specifically includes:

[0137] Low-pass filtering is applied to the coordinates of the gaze focus during fixation to suppress high-frequency jitter;

[0138] The center position of the light spot is determined based on the filtered line-of-sight focus coordinates.

[0139] Specifically, a first-order low-pass digital filter is used to filter the coordinates of the line-of-sight focus. The filtering formula is as follows:

[0140] ;

[0141] The spot locking unit will filter the coordinates The target location is the center of the light spot.

[0142] In the formula:

[0143] For the first The coordinates of the gaze focus after frame filtering;

[0144] The filter coefficients, ranging from 0 to 1, determine the degree of trust in the original data. The smaller the value, the stronger the smoothing effect.

[0145] For the first The raw gaze focus coordinates are directly output by the frame eye-tracking acquisition module;

[0146] These are the filtered coordinates from the previous frame.

[0147] Operational logic explanation: A first-order recursive low-pass filter is used to smooth the original gaze focus coordinates. The filtered output of the current frame... From the original input of the current frame Compared with the filtered output of the previous frame The weighted average is obtained by using the filter coefficients. Control. This operation essentially attenuates high-frequency noise (such as micro-eye movements or sensor jitter) while preserving the low-frequency trend of the gaze focus, thus achieving a stable spot lock position. This low-pass filter formula effectively suppresses spot jitter caused by physiological micro-eye movements or sensor noise during fixation, keeping the spot highly stable when the user is gazing at a static target, eliminating visual flickering or drift. Through recursive computation, the filtering process does not require storing large amounts of historical data, resulting in high computational efficiency, making it suitable for embedded real-time systems. Stable spot locking not only improves user visual comfort but also avoids additional motor energy consumption and light source power fluctuations caused by unnecessary spot oscillations.

[0148] Based on the prediction results, the rotary drive motor 4 and the pitch drive motor 6 are controlled to move the center of the spotlight to the predicted position.

[0149] Example 2: Figures 1 to 7 As shown, a track-mounted electric spotlight control system includes:

[0150] At least one track-mounted electric spotlight, the track-mounted electric spotlight including a spotlight housing 1, an LED light source module 2 and a drive circuit board disposed in the spotlight housing 1, an electric rotating bracket 3 connected to the spotlight housing 1 and a pitch adjustment base 5, wherein a rotation drive motor 4 is disposed in the electric rotating bracket 3 and a pitch adjustment base 5 is disposed in the pitch adjustment base 5.

[0151] An eye-tracking acquisition module 7 is mounted on the spotlight housing 1 or set independently of the spotlight, and is used to acquire the coordinates of the gaze focus of the target person.

[0152] The control unit is communicatively connected to the eye-tracking acquisition module 7 and the drive circuit board, respectively.

[0153] The control unit includes:

[0154] The eye-tracking feature extraction module is used to calculate angular velocity, angular acceleration and gaze duration based on the gaze focus coordinates of multiple consecutive frames acquired by the eye-tracking acquisition module 7.

[0155] The behavior classification module is used to classify the eye movement behavior of the target person into saccades, smooth tracking, or fixation based on the angular velocity, angular acceleration, and fixation duration.

[0156] The saccade prediction module is used to predict the coordinates of the saccade endpoint based on the coordinates of the line of sight at the saccade starting point and the saccade direction when the saccade is classified as a saccade.

[0157] In another embodiment of the present invention, the saccade prediction module includes:

[0158] The scanning amplitude calculation unit is used to calculate the scanning amplitude based on the scanning start point and the peak angular velocity during the scanning process;

[0159] The scanning direction determination unit is used to determine the scanning direction based on the direction of the line connecting the scanning start point and the scanning end point.

[0160] The endpoint coordinate generation unit is used to generate the endpoint coordinates of the scan based on the coordinates of the scan start point, scan amplitude, and scan direction.

[0161] Among them, the scanning range With peak angular velocity The relationships between them satisfy the aforementioned Main sequence relationship; scan the endpoint coordinates. It is also calculated using the aforementioned vector addition formula.

[0162] The smooth tracking prediction module is used to predict the coordinates of future trajectory points based on the gaze focus coordinates of multiple consecutive frames when the tracking is classified as smooth.

[0163] In another embodiment of the present invention, the smooth tracking prediction module includes:

[0164] The trajectory fitting unit is used to fit motion trajectory curves based on the coordinates of the gaze focus in multiple consecutive frames.

[0165] The extrapolation prediction unit is used to extend the motion trajectory curve in the positive direction of the time axis to generate the coordinates of future trajectory points.

[0166] The trajectory fitting and prediction employ the aforementioned polynomial fitting and extrapolation formulas.

[0167] A gaze-holding module is used to maintain the current spot position when the image is classified as gazed.

[0168] In another embodiment of the present invention, the gaze-maintaining module includes:

[0169] The micro-eye-tracking filter unit is used to perform low-pass filtering on the coordinates of the gaze focus during fixation to suppress high-frequency jitter.

[0170] The spot locking unit is used to lock the center position of the spot based on the filtered line-of-sight focus coordinates.

[0171] The low-pass filter uses the aforementioned first-order digital filter formula.

[0172] The motor control module is used to control the rotation drive motor 4 and the pitch drive motor 6 according to the output results of the saccade prediction module, the smooth tracking prediction module or the gaze holding module, so that the center of the spotlight spot moves to the predicted position.

[0173] In another embodiment of the present invention, the control unit further includes:

[0174] The head pose fusion module is used to obtain the head pose angle of the target person;

[0175] The modality switching module is used to switch to head posture following mode when the confidence level of the gaze focus coordinates acquired by the eye-tracking acquisition module 7 is lower than a preset threshold, and to calculate the gaze direction based on the head posture angle.

[0176] The confidence score is a quantitative assessment of the processor's reliability of the gaze focus coordinates in the current frame. The confidence score is calculated using at least one of the following dimensions: completeness of pupil detection (the ratio of the number of detected pupil edge points to the expected number), clarity of corneal reflective spots (spot area contrast), and the magnitude of gaze focus jumps between consecutive frames (if the jump exceeds a preset range, the confidence score is reduced). The scores of each dimension are weighted, summed, and normalized to the 0-1 range to obtain the final confidence score. When the confidence score is below 0.5, the system is deemed unreliable, triggering a mode switch.

[0177] In another embodiment of the present invention, the control unit further includes:

[0178] A personalized learning module is used to identify the identity of the target person and store eye movement feature parameters associated with the identity.

[0179] The model update unit is used to update the eye movement feature parameters based on the currently acquired eye movement data;

[0180] The behavior classification module adjusts the classification threshold based on the eye movement feature parameters stored in the personalized learning module.

[0181] The personalized feature parameters include the aforementioned Main sequence coefficients. and and the order of the aforementioned fitting polynomial. and prediction time step The model update unit uses the recursive least squares method to update the coefficients. , Perform online updates:

[0182] ;

[0183] ;

[0184] In the formula:

[0185] , Personalized slope coefficients before and after the update;

[0186] , The personalized intercept coefficients before and after the update;

[0187] The learning rate, measured in units of T, controls the magnitude of each update.

[0188] This refers to the actual measured scanning amplitude.

[0189] To use the old coefficients to predict the scan amplitude;

[0190] To maintain dimensional consistency during the personalization coefficient update process, a reference angular velocity is introduced, denoted as a unit angular velocity (e.g., 1 rad / s). The peak angular velocity was normalized.

[0191] This represents the peak angular velocity during the scanning process.

[0192] Explanation of the operational logic: The recursive least squares method is used to update the personalized Main sequence coefficients through online gradient descent. and After each scan, the system obtains the actual scan range. And the magnitude predicted using the old coefficients (Depend on (Calculation) and comparison yield the prediction error. Then, using the learning rate Let be the step size, and then... and Make corrections: The correction term is additionally multiplied by the peak angular velocity. This is to demonstrate the physical meaning of the effect of velocity on slope. Updated coefficients. , Used for the next saccade prediction. Through the aforementioned online update formula, the system can continuously learn the user's eye movement habits, gradually converging the Main sequence coefficients to the individual's true values. With increased usage, the accuracy of saccade endpoint prediction continuously improves, achieving adaptive lighting that becomes more accurate with use. This computational logic requires no offline training or manual calibration; it is completed entirely automatically in the background, providing a seamless and natural user experience. Precise personalized predictions further reduce the deviation between the light spot and the gaze focus, lowering the energy consumption of repeated adjustments due to deviation corrections, while simultaneously increasing the user's trust and reliance on smart lighting.

[0193] The model update unit triggers coefficient updates after each saccade. The learning rate... The value ranges from 0.01 to 0.1, and its specific value can be dynamically adjusted according to the user's usage frequency: the higher the usage frequency, the lower the learning rate, to avoid overfitting. The initial value of the Main sequence coefficients is the population statistical average, that is, by collecting scan data from multiple sample users, a general Main sequence coefficient is obtained through fitting. and The convergence criterion for the model is: continuity. Second-rate( If the absolute value of the prediction error during the scan is less than a preset threshold (e.g., 0.05 radians), the coefficient update is stopped and the current personalized coefficients are locked. If the prediction error is subsequently detected to exceed the threshold multiple times consecutively, online learning is restarted.

[0194] In another embodiment of the present invention, the control unit further includes:

[0195] The error detection module is used to calculate the prediction error vector between the actual position of the line-of-sight focus and the center position of the light spot in real time;

[0196] The error compensation module is used to superimpose the prediction error vector onto the prediction value of the next frame to correct the prediction result.

[0197] Among them, the prediction error vector Defined as:

[0198] ;

[0199] Compensated Predicted Location It is given by the following formula:

[0200] ;

[0201] In the formula:

[0202] For a moment The prediction error vector (two-dimensional vector);

[0203] For a moment The actual coordinates of the focal point of the line of sight;

[0204] For a moment The actual coordinates of the center of the light spot;

[0205] The predicted position for the next frame after error compensation;

[0206] The original, uncompensated predicted location (as calculated above).

[0207] Explanation of the operational logic: First, calculate the prediction error vector at the current moment. That is, the actual position of the focal point of the line of sight. The actual position of the light spot center The deviation between them. Then, this error vector is directly superimposed onto the original predicted position in the next frame. Above, the predicted position after compensation. This operation essentially uses past errors as feedforward correction terms, introducing them into future predictions to form a closed-loop negative feedback system. This closed-loop compensation formula can eliminate prediction biases caused by model inaccuracies, motor delays, or external interference in real time, ensuring that the light spot position always closely follows the gaze focus. The frame-by-frame accumulation effect of the error vector quickly suppresses any systematic shifts without relying on complex system identification. This "measure-and-complement" mechanism significantly enhances the robustness of the pre-illumination system, ensuring illumination tracking accuracy even if model parameters do not converge or the user suddenly changes their gaze pattern, while avoiding ineffective illumination and energy waste caused by long-term deviations.

[0208] In another embodiment of the present invention, the control unit further includes:

[0209] A beam angle adjustment module is used to adjust the beam angle of the track-mounted motorized spotlight based on the classification results of the behavior classification module.

[0210] When classified as scan, narrow the beam angle;

[0211] When classified as gaze, the beam angle is widened.

[0212] Among them, beam angle With line-of-sight angular velocity The relationship is represented by a piecewise function:

[0213] ;

[0214] In the formula:

[0215] This refers to the output beam angle;

[0216] This is the minimum beam angle (focused state);

[0217] This is the maximum beam angle (floodlight mode);

[0218] The middle beam angle;

[0219] For a moment line-of-sight angular velocity;

[0220] This is the threshold for the scanning angle speed; anything exceeding this value is considered a rapid scanning.

[0221] This is the gaze velocity threshold; anything below this value is considered a gaze state.

[0222] Explanation of computational logic: Based on the output of the behavior classification module and the real-time angular velocity A suitable beam angle is selected using a piecewise function. When a scan is detected and the angular velocity exceeds the scan threshold... At that time, the minimum beam angle is used. (Focus mode); when it is identified as gaze and the angular velocity is below the gaze threshold At that time, the maximum beam angle is used. (Floodlight mode); other transitional states (such as smooth tracking or slow scanning) use the intermediate beam angle. This logic enables automatic matching of the beam angle with eye movement. By dynamically adjusting the beam angle, the system narrows the beam spot during saccades to reduce visual interference and increase light concentration, while widening the beam spot during fixation to cover a larger, more comfortable field of vision. This eye-motion adaptive light pattern switching significantly improves the ergonomics of lighting: during saccades, the narrow beam avoids glare caused by the beam spot sweeping across irrelevant areas, while during fixation, the wide beam reduces fatigue caused by contrast in the peripheral field of vision. Simultaneously, the narrow beam mode precisely projects light energy onto a very small area, reducing useless lighting energy consumption per unit area, meeting the needs of industrial vision, visual intelligence, AI vision, visual inspection, and visual measurement.

[0223] In this embodiment, the track-mounted electric spotlight is equipped with a beam angle adjustment mechanism, which includes a stepper motor, a transmission gear set, and a movable lens group. The processor controls the stepper motor to drive the transmission gear set, causing the lens group to move along the optical axis, thereby achieving continuous adjustment of the beam angle: when the lens group moves towards the LED light source module, the beam angle narrows; when the lens group moves away from the LED light source module, the beam angle widens. The processor has a pre-stored mapping table between the beam angle and the lens group position. Based on the output result of the behavior classification module, the target position is determined by looking up the table, and the stepper motor is controlled to drive the lens group to move to the target position.

[0224] In another embodiment of the present invention, the control unit further includes:

[0225] A power adjustment module is used to adjust the output power of the LED light source module according to the classification results of the behavior classification module.

[0226] When classified as scan, reduce the output power to the first threshold.

[0227] When classified as smooth tracking or gaze, the output power is increased to the second threshold.

[0228] Among them, output power The correlation with eye movement behavior type is as follows:

[0229] ;

[0230] Furthermore, during the scanning process, the power reduction factor... With the sweep amplitude Related:

[0231] ;

[0232] In the formula:

[0233] This refers to the output power of the LED light source module;

[0234] This is the low power value during scanning mode;

[0235] This represents the high power value in the non-scanning state;

[0236] The power reduction factor is... and The ratio;

[0237] The scanning range (from the aforementioned calculations);

[0238] The reference amplitude constant is used to control the decay rate.

[0239] Explanation of operational logic: First, the output power is directly set based on the classification of eye movement behavior: low power is used during saccades. High power is used for smooth tracking or gaze. Furthermore, the power reduction factor during the scanning process. With the sweep amplitude Correlation, calculated using the exponential decay function: the greater the magnitude, the more significant the power reduction. (The smaller the value). This formula enables the system to proactively reduce light output when the user quickly shifts their gaze, and restore full light output during fixation or tracking. This power adjustment formula utilizes the physiological characteristic that the human eye is insensitive to changes in brightness during saccades, proactively reducing power during rapid gaze shifts to achieve deep energy savings without the user noticing. The greater the saccade amplitude, the greater the power reduction, further matching the energy-saving potential during visual inhibition periods. High power is restored during smooth tracking or fixation phases, ensuring illumination quality for critical visual tasks. This asymmetric power strategy significantly reduces the system's total energy consumption without sacrificing user experience, making it particularly suitable for scenarios requiring prolonged continuous use.

[0240] In this embodiment, the first threshold is set as follows: the output power in the saccade state is set to 20% to 50% of the output power in the normal fixation state, with the specific ratio depending on the saccade amplitude. Relatedly, the greater the scan amplitude, the higher the power reduction ratio, following an exponential decay relationship. ,in This serves as a reference amplitude constant. The second threshold, representing the output power under normal gaze conditions, can be dynamically adjusted based on ambient light levels: the higher the ambient light, the lower the second threshold, in order to maintain a constant visual illuminance in the focal area of ​​the gaze.

[0241] In another embodiment of the present invention, the number of track-mounted electric spotlights is multiple, and the multiple track-mounted electric spotlights are arranged along a track; the control unit further includes:

[0242] The spotlight scheduling module is used to select the target spotlight closest to the predicted location from multiple track-mounted motorized spotlights based on the spatial coordinates of the predicted location, and drive the target spotlight to perform the lighting task.

[0243] The control unit further includes:

[0244] The light field coordination module is used to drive the two or more track-mounted motorized spotlights to simultaneously output illumination light when the predicted position is located in the overlapping area covered by the light spots of two or more track-mounted motorized spotlights, and to allocate the output power ratio according to the distance between the predicted position and each spotlight.

[0245] Among them, the Power distribution of each spotlight Determined by the distance-weighted formula:

[0246] ;

[0247] In the formula:

[0248] To be assigned to the The power of each spotlight;

[0249] This represents the total output power of the overlapping region;

[0250] To predict the position to the th The Euclidean distance between the centers of the spotlights;

[0251] This is the spatial decay constant, which controls the rate at which the weight decreases with distance;

[0252] This represents the total number of spotlights within the overlapping area covered by the light spot.

[0253] It is a natural exponential function.

[0254] Explanation of computational logic: When the predicted position falls within the overlapping area of ​​multiple spotlight beams, a spatial soft allocation strategy is used to calculate the power that each spotlight should share. First, the power distribution from the predicted position to the [missing information] spotlight is calculated. The distance between the centers of the spotlights Then through the negative exponential function Obtain the weight of each spotlight (the closer the spotlight, the higher the weight). Sum the weights of all spotlights, normalize the result using the sum of the weights, and then multiply by the total power. , obtained the Power distribution of each spotlight This calculation achieves a smooth transition of the light field, avoiding abrupt changes in illuminance caused by hard switching. The distance-weighted power allocation formula ensures that the light field generated by multiple spotlights is spatially continuous and smooth, with seamless power transfer between spotlights during predicted positional movement, eliminating flicker or dark areas that occur with traditional single-lamp switching. The exponential decay characteristic ensures that the spotlight closest to the focal point provides primary illumination, while surrounding spotlights provide supplementary lighting, guaranteeing high illuminance in the focal area while avoiding unnecessary energy consumption from distant spotlights. The overall light field remains uniform and stable as it follows the line of sight, improving visual comfort while achieving overall energy savings in the multi-lamp system.

[0255] The saccade prediction module is further used for:

[0256] When the coordinates of the scanning endpoint are outside the reachable range of the current spotlight, the coordinates of the scanning endpoint are sent to the spotlight scheduling module, which then pre-activates the spotlight closest to the scanning endpoint.

[0257] In another embodiment of the present invention, the control unit further includes:

[0258] The rest reminder module is used to generate a rest reminder signal when the duration of fixation exceeds a preset threshold and the rate of change of eye movement feature parameters during fixation is lower than a preset threshold.

[0259] In another embodiment of the present invention, the control unit further includes:

[0260] The wireless communication module is used to communicate with mobile terminals or voice-activated audio devices and receive external control commands.

[0261] The motor control module is also used to control the rotary drive motor and the pitch drive motor according to the external control command.

[0262] In another embodiment of the present invention, the control unit is integrated into the drive circuit board of the track-mounted electric spotlight, or is set independently outside the track.

[0263] In another embodiment of the present invention, the control unit further includes:

[0264] The color temperature adjustment module is used to adjust the color temperature of the LED light source module according to the duration of gaze when the behavior classification module classifies it as gaze: the longer the gaze duration, the lower the color temperature gradually becomes.

[0265] Among them, color temperature With gaze duration The relationship is an exponential decay function:

[0266] ;

[0267] In the formula:

[0268] For gaze duration Output color temperature at that time;

[0269] For the cumulative fixation duration (timed from the start of fixation);

[0270] Minimum color temperature (warm light limit);

[0271] This is the maximum color temperature (initial value of cool light).

[0272] It is a time constant that controls the rate at which the color temperature decays.

[0273] Explanation of the operation logic: When the system determines that the user is in a gaze state, it calculates based on the cumulative gaze duration. Dynamically adjust color temperature. Initial moment ( The color temperature is As fixation time increases, the exponential decay term... As color temperature gradually approaches 0 from 1, from Towards Smooth transition. Time constant. By controlling the decay rate, the color temperature change rate is made to match the human eye's sensitivity curve to color temperature changes. This calculation achieves a physiologically adaptive adjustment where the color temperature automatically decreases as eye strain accumulates. The exponential decay color temperature adjustment formula simulates the physiological laws of human circadian rhythms and visual fatigue: gradually decreasing the color temperature (warming up) during prolonged eye contact reduces the stimulation of the retina by short-wavelength blue light, alleviating visual fatigue; simultaneously, a warm-toned environment helps relax eye muscles. The smooth exponential transition avoids visual discomfort caused by abrupt color temperature changes. This design elevates lighting from simply "illuminating" to "actively protecting the eyes," protecting the user's vision while lower color temperatures typically correspond to lower luminous flux requirements (at the same visual brightness), thus indirectly achieving energy savings.

[0274] In this manual, all physical quantities (such as angle, time, length, power, etc.) in the algorithm are represented by SI values ​​in floating-point operations. The calculations in the formulas only involve numerical calculations, and the dimensions are kept equivalent in the formula structure, so there will be no logical errors.

[0275] The embodiments disclosed in this invention are preferred embodiments, but are not limited thereto. Those skilled in the art can easily understand the spirit of this invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of this invention, they are all within the protection scope of this invention.

Claims

1. A track-mounted electric spotlight, characterized in that, include: The spotlight housing (1) has an opening for illumination, and a lens (9) is arranged in the opening. The LED light source module (2) is disposed inside the spotlight housing (1); The driving circuit board is disposed inside the spotlight housing (1) and is electrically connected to the LED light source module (2); An electric rotating bracket (3) is connected to the spotlight housing (1). The electric rotating bracket (3) is equipped with a rotating drive motor (4) for driving the spotlight housing (1) to rotate in the horizontal direction. The pitch adjustment base (5) is connected between the spotlight housing (1) and the electric rotating bracket (3). The pitch adjustment base (5) is equipped with a pitch drive motor (6) for driving the spotlight housing (1) to swing in the vertical direction. An eye-tracking acquisition module (7) is installed on the spotlight housing (1) and is used to acquire the coordinates of the gaze focus of the target person; The drive circuit board includes a processor and a memory. When the processor executes a computer program stored in the memory, it performs the following steps: Obtain the coordinates of the gaze focus in multiple consecutive frames acquired by the eye-tracking acquisition module (7); Eye movement feature parameters are calculated based on the gaze focus coordinates of multiple consecutive frames. The eye movement feature parameters include angular velocity, angular acceleration, and gaze duration. Based on the eye movement feature parameters, eye movement behavior is classified into saccades, smooth tracking, or fixation. When classified as saccade, the coordinates of the saccade endpoint are predicted based on the coordinates of the line of sight at the saccade starting point and the saccade direction. When classified as smooth tracking, the coordinates of future trajectory points are predicted based on the gaze focus coordinates of multiple consecutive frames. When classified as fixation, maintain the current spot position; Based on the prediction results, control the rotary drive motor (4) and the pitch drive motor (6) to move the center of the spotlight to the predicted position.

2. A track-mounted electric spotlight according to claim 1, characterized in that, The eye-tracking acquisition module (7) is a binocular eye-tracking camera, which is located on the front end of the spotlight housing (1).

3. A track-mounted electric spotlight according to claim 2, characterized in that, An inertial measurement unit is also provided inside the spotlight housing (1). The inertial measurement unit is electrically connected to the drive circuit board and is used to collect the head posture angle of the target person. When the processor executes the computer program, it also performs the following steps: acquiring the head posture angle collected by the inertial measurement unit; when the confidence level of the gaze focus coordinates collected by the eye-tracking acquisition module (7) is lower than a preset threshold, calculating the gaze direction based on the head posture angle, and switching to the head posture following mode.

4. A track-mounted electric spotlight according to claim 3, characterized in that, An ambient light sensor (8) is also provided inside the spotlight housing (1). The ambient light sensor (8) is electrically connected to the drive circuit board and is used to collect ambient illuminance. When the processor executes the computer program, it also performs the following steps: acquiring the ambient illuminance collected by the ambient light sensor (8); when the rate of change of ambient illuminance exceeds a preset threshold, adjusting the exposure parameters of the eye-tracking acquisition module (7) and recalibrating the gaze focus coordinates.

5. A track-mounted electric spotlight according to claim 4, characterized in that, The steps for predicting the coordinates of the end point of the scan specifically include: calculating the scan amplitude based on the scan start point and the peak angular velocity during the scan process; determining the scan direction based on the direction of the line connecting the scan start point and the scan end point; and generating the coordinates of the scan end point based on the scan start point coordinates, scan amplitude, and scan direction. Among them, the scanning range With peak angular velocity The relationship between them is given by the Main sequence formula. Express, scan the destination coordinates From the coordinates of the starting point of the scan Sweep range The line of sight between the target personnel and the projection plane of their line of sight and the unit vector of the scanning direction Through formula The calculation yields the following result: Indicates the personalized slope coefficient; This indicates the peak angular velocity during the scanning process; This represents the personalized intercept coefficient.

6. A track-mounted electric spotlight control system for controlling at least one track-mounted electric spotlight as described in claim 5, characterized in that, include; The control unit is communicatively connected to the eye-tracking acquisition module (7) and the drive circuit board, respectively; The control unit includes: The eye movement feature extraction module is used to calculate angular velocity, angular acceleration and gaze duration based on the gaze focus coordinates of multiple consecutive frames acquired by the eye movement acquisition module (7); The behavior classification module is used to classify the eye movement behavior of the target person into saccades, smooth tracking, or fixation based on the angular velocity, angular acceleration, and fixation duration. The saccade prediction module is used to predict the coordinates of the saccade endpoint based on the coordinates of the line of sight at the saccade starting point and the saccade direction when the saccade is classified as a saccade. The smooth tracking prediction module is used to predict the coordinates of future trajectory points based on the gaze focus coordinates of multiple consecutive frames when the tracking is classified as smooth. A gaze-holding module is used to maintain the current spot position when the image is classified as gazed. The motor control module is used to control the rotation drive motor (4) and the pitch drive motor (6) according to the output results of the saccade prediction module, the smooth tracking prediction module or the gaze holding module, so that the center of the spotlight moves to the predicted position.

7. A track-mounted electric spotlight control system according to claim 6, characterized in that, The saccade prediction module includes: The scanning amplitude calculation unit is used to calculate the scanning amplitude based on the scanning start point and the peak angular velocity during the scanning process; The scanning direction determination unit is used to determine the scanning direction based on the direction of the line connecting the scanning start point and the scanning end point. The endpoint coordinate generation unit is used to generate the endpoint coordinates of the scan based on the coordinates of the scan start point, scan amplitude, and scan direction.

8. A track-mounted electric spotlight control system according to claim 6, characterized in that, The smooth tracking prediction module includes: The trajectory fitting unit is used to fit motion trajectory curves based on the coordinates of the gaze focus in multiple consecutive frames. The extrapolation prediction unit is used to extend the motion trajectory curve in the positive direction of the time axis to generate the coordinates of future trajectory points. Among them, trajectory fitting and prediction adopt The order polynomial with respect to the nearest Frame view focus coordinates The fitting curve equation is as follows: , Future moments Coordinates of the predicted trajectory points From extrapolation formula , We obtain the following formula: , A fitting function representing the change of the coordinates of the line of sight focus over time; , express coordinates and Fitting coefficients of the coordinate polynomial; It is a time variable.

9. A track-mounted electric spotlight control system according to claim 6, characterized in that, The gaze-maintaining module includes: The micro-eye-tracking filter unit is used to perform low-pass filtering on the coordinates of the gaze focus during fixation to suppress high-frequency jitter. The spot locking unit is used to lock the center position of the spot based on the filtered line-of-sight focus coordinates; The low-pass filter uses a first-order low-pass digital filter, and the filtering formula is as follows: ； The spot locking unit will filter the coordinates The target location as the center of the light spot; In the formula, For the first The coordinates of the gaze focus after frame filtering; These are the filter coefficients; For the first The raw gaze focus coordinates are directly output by the frame eye-tracking acquisition module; These are the filtered coordinates from the previous frame.

10. A track-mounted electric spotlight control system according to claim 6, characterized in that, The control unit further includes: The head pose fusion module is used to obtain the head pose angle of the target person; The modality switching module is used to switch to head posture following mode when the confidence level of the gaze focus coordinates acquired by the eye-tracking acquisition module (7) is lower than a preset threshold, and to calculate the gaze direction based on the head posture angle.

Images