Riding state monitoring method based on intelligent riding glasses

Abstract

The invention discloses a riding state monitoring method based on intelligent riding glasses, and relates to the technical field of intelligent riding. According to the riding state monitoring method based on the intelligent riding glasses, all-around safety monitoring is achieved, various high-precision sensors such as an IMU and a heart rate sensor are integrated, a complex data processing algorithm is combined, various safety related states such as emergency braking, falling down and head movement of a rider can be monitored in real time, and the riding state can be monitored in real time. By means of the multi-level judgment logic and lag processing mechanism, the accuracy and reliability of monitoring are greatly improved, potential risks are found in time, and more comprehensive safety guarantee is provided for a rider, for example, in emergency braking detection, the safety of the rider is improved. Through multi-step processing of double-stage low-pass filtering, deceleration threshold analysis, trend analysis and speed estimation of forward and backward linear acceleration, the braking condition can be accurately judged, and potential safety hazards caused by misjudgment are effectively avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent riding, and particularly relates to a riding state monitoring method based on intelligent riding glasses. BACKGROUND

[0002] With the aggravation of urban traffic congestion and the improvement of people's health awareness, bicycles, electric bicycles and other riding tools are increasingly becoming an important choice for people's travel. However, the safety problem in the process of riding is also increasingly prominent. According to the "2023 Global Road Safety Report" of the World Health Organization, tens of thousands of riders are injured or killed in traffic accidents every year, and a considerable part of the accidents are related to emergency braking, vehicle out of control, rider falling or distraction.

[0003] At present, there are various riding safety and state monitoring technologies on the market, each of which has its own characteristics but also has obvious limitations: 1. Vehicle-mounted sensor system: usually installed on the vehicle, including speed sensor, brake sensor, etc. Although this kind of system can provide relatively accurate vehicle motion information, it cannot directly perceive the body movements and posture of the rider, and the compatibility between different vehicles is poor.

[0004] 2. Wearable devices: such as heart rate monitoring bands, smart watches, etc., mainly monitor the physiological indicators of the rider, but it is difficult to accurately capture dangerous states such as emergency braking, body shaking and falling during riding.

[0005] 3. Data fusion and display system: such as the riding state display method described in patent CN119337310A, which captures real-time riding data and body state data from multiple terminal devices and performs fusion analysis. However, this system mainly relies on heart rate and riding speed to analyze user fatigue state, and does not use IMU (Inertial Measurement Unit) to provide posture data, which cannot detect head posture changes and micro-motion characteristics. The fatigue judgment standard is limited to "when the heart rate analysis result indicates that the heart rate value is greater than or equal to the heart rate threshold, it is determined that the user's riding state is tired", which lacks in-depth analysis of riding posture changes.

[0006] 4. Accident prediction and warning system: such as the riding warning method proposed in patent CN119107837A, which monitors user physiological parameters through intelligent wearable devices; relies on sensors installed on the riding tool to monitor riding speed, acceleration and riding posture; obtains riding map data and real-time road condition data through a smart terminal. The entire patent does not use IMU data and does not involve head posture monitoring and analysis methods. This system mainly focuses on external environmental factors and collision risks, and does not have the ability to analyze the head movements and riding state of the rider based on intelligent glasses.

[0007] In recent years, with the development of smart wearable device technology, smart riding glasses as a new type of riding equipment gradually enter the market. Such glasses usually integrate IMU, GPS and other sensors, among which IMU as an ideal riding state monitoring tool, the installation position is stable, and can accurately capture the six degrees of freedom (three-axis acceleration and three-axis angular velocity) of head movement, which provides a new technical possibility for riding posture, vehicle state and fatigue monitoring. The present application is based on the IMU, GPS and other sensor systems built in the smart riding glasses, and proposes a new type of riding state monitoring and fatigue monitoring method based on smart riding glasses. By analyzing the changes of head posture and combining the riding performance parameters, the riding state and fatigue state of the rider can be accurately identified, which fills the gap of the existing technology in the application of IMU data and riding posture monitoring, and provides an innovative technical solution to improve the safety of riding. SUMMARY

[0008] The present application aims to at least solve one of the technical problems existing in the prior art, and provides a riding state monitoring method based on smart riding glasses, which can solve the problems in the background art.

[0009] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a riding state monitoring method based on smart riding glasses, comprising the following specific steps: S1 data acquisition: The smart riding glasses are built-in with various sensors for collecting key data; Inertial Measurement Unit (IMU): contains three-axis accelerometer (range ±16g, sampling rate >=200Hz), three-axis gyroscope (range ±2000° / s, sampling rate >=200Hz) and three-axis magnetometer (range ±4000μT, sampling rate >=100Hz), which is used to obtain acceleration, angular velocity and magnetic field data; Heart rate sensor: uses photoplethysmography method to measure, sampling rate >=60Hz, and collects the heart rate data of the rider; S2 coordinate transformation: Determine the relationship between the actual installation direction of the smart riding glasses and the navigation coordinate system (such as NWU coordinate system), and convert the glasses coordinate system data into navigation coordinate system data; the commonly used conversion formula is as follows: Gyro data conversion: GyroNWU=[−Gyroz,Gyroy,Gyrox] Accelerometer data conversion: AccelNWU=[−Accelz,Accely,Accelx] S3 sensor fusion and attitude acquisition; S4 multi-state monitoring; S4 multi-state monitoring includes S401 emergency braking detection, S402 deceleration feature analysis, S403 head movement monitoring, S404 fall detection, S405 rocking state detection, and S406 riding posture monitoring; S5 comprehensive evaluation and early warning; S6 event recording and analysis; S7 system integration assistance.

[0010] Preferably, the S3 sensor fusion and posture acquisition: An improved complementary filter is used to realize posture estimation, and a quaternion is used to represent the posture; the specific steps are as follows: S301 gyroscope integration: integrating the angular velocity data of the gyroscope to obtain a predicted value of the posture; S302 accelerometer correction: using accelerometer data to correct the posture and eliminate the drift error of gyroscope integration; S303 complementary filter fusion: weighting and fusing the predicted value of the gyroscope and the corrected value of the accelerometer to obtain a more accurate posture estimation; S304 gyroscope bias estimation: real-time estimation of the gyroscope bias to further improve the accuracy of the posture estimation, based on the obtained posture quaternion, eliminating the gravity effect and extracting the linear acceleration.

[0011] Preferably, the S401 emergency braking detection; Data filtering: two-stage low-pass filtering of forward and backward linear acceleration, first-stage cutoff frequency 2.0Hz, second-stage cutoff frequency 1.0Hz.

[0012] Preferably, the S402 deceleration feature analysis: Threshold analysis: setting a deceleration threshold, when the deceleration exceeds the threshold, considering that an emergency braking may occur; Trend analysis: analyzing the change trend of deceleration to determine whether it is a sharp deceleration; Speed estimation: combining acceleration data and time information to estimate the change of riding speed.

[0013] Preferably, the S403 head movement monitoring: extracting and tracking the change of head yaw angle, and dynamically adjusting the braking detection threshold according to the head movement state; Multi-level decision logic: outputting the emergency braking detection result through multi-level decision logic and hysteresis processing.

[0014] Preferably, the S404 fall detection: Impact event detection: monitoring the peak value of the specific force, when ∣a∣>IMPACT_THRESHOLD (such as 3.5g), triggering an impact event; Attitude rate analysis: analyze attitude rate, record attitude sudden change event when Δθ / Δt>ATTITUDE_RATE_THRESHOLD; Free fall detection: detect near zero gravity state, identify free fall phase; Stationary state determination: determine whether in stationary state after falling down; Multi-feature fusion determination: fuse impact, attitude change, free fall, stationary state multi-feature sequence to determine fall event, and classify fall type (forward fall, lateral fall, backward fall).

[0015] Preferably, the S405 rocking state detection: Data filtering: extract vertical linear acceleration, and low-pass filter linear acceleration and pitch angle, vertical acceleration cutoff frequency 3.0Hz, pitch angle cutoff frequency 2.0Hz; Standard deviation analysis: analyze acceleration standard deviation based on long and short double time windows (long window 1.0s, short window 0.5s); Shaking frequency calculation: calculate shaking frequency through peak identification; State maintenance mechanism: apply asymmetric hysteresis mechanism to maintain shaking state; State determination: determine shaking state according to standard deviation threshold and duration.

[0016] Preferably, the S406 riding posture monitoring: Head inclination angle monitoring: perform long-term monitoring of head inclination angle trend; Head movement analysis: analyze head movement frequency and amplitude; Fatigue assessment: assess riding fatigue based on head posture long-term trend, head micro-motion frequency statistics multi-index comprehensive evaluation; Fatigue warning: generate graded fatigue warning according to fatigue degree.

[0017] Preferably, the S5 comprehensive assessment and warning: State correlation analysis: analyze correlation and conversion mode of multiple states such as emergency braking, falling, and rocking; Safety risk score: calculate safety risk comprehensive score; Graded warning mechanism: implement graded warning mechanism (level 1 prompt warning, level 2 attention warning, level 3 emergency warning); Intelligent feedback adaptation: provide intelligent feedback adaptation according to warning level, environmental conditions and user habits.

[0018] Preferably, the S6 event recording and analysis: Event logging: When various events such as braking, falling, swaying, and fatigue warning are detected, they are recorded in a specific format, including event type, start time, end time, duration, maximum intensity, confidence level, and related sensor data; Event filtering: Record corresponding events based on filtering logic (e.g., braking event duration > 0.3 seconds, fall event confidence > 0.7, rocking event duration > 2.0 seconds, severe fatigue state); Event statistics: Perform event statistics (total number of events of various types during the ride, time distribution, intensity distribution, correlation analysis), and generate a riding report (safe riding score, riding habit analysis, safety recommendations, long-term trend analysis); S7 System Integration Assistance: Integration with vision systems: When an emergency event is detected, the camera is triggered to record, and computer vision algorithms are used to identify obstacles ahead and assist in environmental perception to determine the status. Integration with alarm systems: Provides differentiated feedback based on event type and urgency; Integration with health monitoring systems: Combines heart rate monitoring to assess the cyclist's physical condition; Integration with navigation systems: Records the location of dangerous events, marks dangerous sections of road and provides advance warnings for future rides, and dynamically adjusts route recommendations based on the rider's condition; Integration with cloud services: Upload event data for long-term analysis.

[0019] Compared with the prior art, the beneficial effects of the present invention are: 1. This cycling status monitoring method based on smart cycling glasses achieves comprehensive safety monitoring: By integrating multiple high-precision sensors, such as IMU and heart rate sensors, and combining them with complex data processing algorithms, it can monitor various safety-related states of cyclists in real time, including emergency braking, falls, and head movements. Multi-level judgment logic and hysteresis processing mechanisms greatly improve the accuracy and reliability of monitoring, enabling timely detection of potential dangers and providing cyclists with more comprehensive safety protection. For example, in emergency braking detection, through multi-step processing of forward and backward linear acceleration using dual-stage low-pass filtering, deceleration threshold analysis, trend analysis, and speed estimation, the braking situation can be accurately judged, effectively avoiding safety hazards caused by misjudgment.

[0020] 2、The riding state monitoring method based on the intelligent riding glasses realizes deep analysis of the riding state: the rocking state and the riding posture can be deeply monitored and analyzed, the rocking state is accurately judged through filtering, standard deviation analysis and shaking frequency calculation of the vertical linear acceleration and the pitch angle, the riding fatigue degree is comprehensively evaluated based on head inclination monitoring and motion analysis, more targeted riding state feedback is provided for the rider, the rider's riding habit is optimized, the riding efficiency and experience are improved, for example, according to the rocking state detection result, the rider can better adjust the riding rhythm and reasonably distribute the physical strength.

[0021] 3、The riding state monitoring method based on the intelligent riding glasses realizes intelligent integration and cooperation: the intelligent riding glasses are deeply integrated with multiple systems such as vision, alarm, health monitoring, navigation and cloud service, when an emergency occurs, the camera can be automatically triggered to record and perform environment perception, differential feedback can be provided in combination with the alarm system, the rider's physical state can be evaluated according to the health monitoring data, the navigation system can adjust the route according to the rider's state, and the data can be uploaded to the cloud for long-term analysis, the multi-system integration and cooperation realizes the interconnection and comprehensive utilization of data, and provides more intelligent, convenient and personalized services for the rider.

[0022] 4、The riding state monitoring method based on the intelligent riding glasses realizes data recording and analysis: perfect event recording and analysis function is provided, key information of various events such as braking, falling, rocking and fatigue warning can be recorded in detail, and effective screening and statistics can be performed according to the screening logic, the generated riding report covers safety riding score, riding habit analysis and safety suggestion content, helps the rider to deeply understand his own riding behavior, finds potential problems, and takes corresponding improvement measures to promote riding safety and health. BRIEF DESCRIPTION OF DRAWINGS

[0023] The application will be further described below in combination with the drawings and embodiments: Figure 1 It is the overall system architecture of the application; Figure 2 It is the emergency braking detection algorithm flowchart of the application; Figure 3 It is the falling detection algorithm flowchart of the application; Figure 4 It is the rocking state recognition algorithm flowchart of the application; Figure 5 It is the riding fatigue state monitoring flowchart of the application. DETAILED DESCRIPTION

[0024] The specific embodiments of the present application will be described in detail in this part, the preferred embodiments of the present application are shown in the drawings, the role of the drawings is to supplement the description of the text part with figures, so that people can intuitively and visually understand each technical feature and the overall technical scheme of the present application, but it cannot be understood as a limitation on the protection scope of the present application.

[0025] In the description of the present application, it should be understood that the orientation description, such as the orientation or position relationship indicated by up, down, front, back, left, right and the like, is based on the orientation or position relationship shown in the drawings, only for the purpose of describing the present application and simplifying the description, and is not intended to indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore it cannot be understood as a limitation on the present application.

[0026] In the description of the present application, greater than, less than, more than and the like are understood as not including the number, above, below and the like are understood as including the number. If the first, second is described, it is only for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated or the sequence of technical features indicated.

[0027] In the description of the present application, unless otherwise explicitly limited, the words such as setting, installing, connecting and the like should be understood in a broad sense, and the person skilled in the art can reasonably determine the specific meaning of the above words in the present application in combination with the specific content of the technical scheme.

[0028] Please refer to Figures 1-5 The present application provides a technical scheme: a riding state monitoring method based on intelligent riding glasses, comprising the following specific steps: S1 data acquisition: The intelligent riding glasses are built-in with various sensors for collecting key data; Inertial measurement unit (IMU): containing three-axis accelerometer (range ± 16g, sampling rate >=200Hz), three-axis gyroscope (range ±2000° / s, sampling rate >=200Hz) and three-axis geomagnetic meter (range ±4000μT, sampling rate >=100Hz), used for acquiring acceleration, angular velocity and magnetic field data; Heart rate sensor: using photoplethysmography method for measurement, sampling rate >=60Hz, collecting the heart rate data of the rider; S2 coordinate transformation: Determine the relationship between the actual installation direction of the intelligent riding glasses and the navigation coordinate system (such as NWU coordinate system), and convert the glasses coordinate system data into navigation coordinate system data; the commonly used conversion formula is as follows: Gyroscopic data conversion: GyroNWU=[−Gyroz,Gyroy,Gyrox] Accelerometer data conversion: AccelNWU = [−Accelz, Accely, Accelx] S3 Sensor fusion and attitude acquisition: Attitude estimation is implemented using an improved complementary filter, with attitude represented by a quaternion; the specific steps are as follows: S301 Gyroscope integration: Integrate the angular velocity data from the gyroscope to obtain the predicted value of the attitude; S302 Accelerometer correction: Use accelerometer data to correct the attitude and eliminate the drift error of gyroscope integration; S303 Complementary filter fusion: Weighted fusion of the predicted value of the gyroscope and the corrected value of the accelerometer to obtain a more accurate attitude estimate; S304 Gyroscope bias estimation: Real-time estimation of the gyroscope bias to further improve the accuracy of attitude estimation, based on the obtained attitude quaternion, eliminate the influence of gravity, and extract linear acceleration; S4 Multi-state monitoring: S401 Emergency braking detection Data filtering: Two-stage low-pass filtering of forward and backward linear acceleration, with a first-stage cutoff frequency of 2.0 Hz and a second-stage cutoff frequency of 1.0 Hz; S402 Deceleration feature analysis: Threshold analysis: Set a deceleration threshold, when the deceleration exceeds the threshold, consider that an emergency braking may occur; Trend analysis: Analyze the change trend of deceleration to determine whether it is a sharp deceleration; Speed estimation: Estimate the change of riding speed based on acceleration data and time information; S403 Head movement monitoring: Extract and track the change of head yaw angle, and dynamically adjust the braking detection threshold according to the head movement state; Multi-level decision logic: Output the emergency braking detection result through multi-level decision logic and hysteresis processing; S404 Fall detection: Impact event detection: Monitor the peak value of the jerk, when |a| > IMPACT_THRESHOLD (e.g. 3.5g), trigger an impact event; Attitude rate analysis: Analyze the attitude rate, when Δθ / Δt > ATTITUDE_RATE_THRESHOLD, record the attitude sudden change event; Free fall detection: Detect the near-zero gravity state and identify the free fall phase; Still state determination: Determine whether it is in a still state after falling; Multi-feature fusion judgment: fusion of impact, posture change, free fall, and static state multi-feature sequence judgment of falling events, and falling type classification (forward falling, lateral falling, backward falling); S405 rocking state detection: Data filtering: extract linear acceleration in the vertical direction, and low-pass filter linear acceleration and pitch angle, with a vertical acceleration cutoff frequency of 3.0Hz and a pitch angle cutoff frequency of 2.0Hz; Standard deviation analysis: analyze acceleration standard deviation based on long and short double time windows (long window 1.0 second, short window 0.5 second); Shaking frequency calculation: calculate the shaking frequency by peak identification; State maintenance mechanism: apply an asymmetric hysteresis mechanism to maintain the shaking state; State judgment: determine the shaking state according to the standard deviation threshold and the duration; S406 riding posture monitoring: Head tilt angle monitoring: long-term monitoring of head tilt angle trend; Head movement analysis: analyze head movement frequency and amplitude; Fatigue assessment: based on long-term trend of head posture, head micro-motion frequency statistics, multi-index comprehensive evaluation of riding fatigue; Fatigue warning: generate graded fatigue warning according to fatigue; S5 comprehensive evaluation and warning: State correlation analysis: analyze the correlation and conversion mode of multiple states such as emergency braking, falling, and rocking; Safety risk score: calculate the safety risk comprehensive score; Graded warning mechanism: implement a graded warning mechanism (level 1 prompt warning, level 2 attention warning, level 3 emergency warning); Intelligent feedback adaptation: provide intelligent feedback adaptation according to warning level, environmental conditions and user habits; S6 event recording and analysis: Event recording: when detecting brake, fall, rocking, and fatigue warning events, record in a specific format, including event type, start time, end time, duration, maximum intensity, confidence, and related sensor data; Event screening: record corresponding events according to screening logic (such as brake event duration>0.3 seconds, fall event confidence>0.7, rocking event duration>2.0 seconds, severe fatigue state); Event statistics: perform event statistics (total number of events, time distribution, intensity distribution, correlation analysis during riding), generate riding report (safe riding score, riding habit analysis, safety suggestion generation, long-term trend analysis); S7 system integration assistance: Integration with vision system: trigger camera recording when emergency is detected, combine with computer vision algorithm for front obstacle recognition and environmental perception to assist in state judgment; Integration with alarm system: provide differentiated feedback according to event type and emergency level; Integration with health monitoring system: combine with heart rate monitoring to evaluate the physical state of the rider; Integration with navigation system: record the location of dangerous events, mark dangerous sections and provide early warning in future rides, dynamically adjust route recommendations according to the rider's state; Integration with cloud service: upload event data for long-term analysis; Further, the intelligent riding glasses used in the present application include the following hardware components: 1. Inertial Measurement Unit (IMU): Three-axis accelerometer: range ±16g, sampling rate >=200Hz; Three-axis gyroscope: range ±2000° / s, sampling rate >=200Hz; Three-axis magnetometer: range ±4000μT, sampling rate >=100Hz; 2. Processing unit: High-pass AR1 or wearable device dedicated chip; Support edge computing and AI acceleration; Low power design, suitable for long time endurance; 3. Positioning system: Built-in GPS module, supports accurate positioning; Support connecting mobile phone GPS module as backup or enhanced positioning scheme; Support AGPS assisted positioning, improve first positioning speed; 4. Communication module: Bluetooth 5.0, support BLE low power communication; Optional WiFi module, support data upload and firmware update; Support pairing with smart phone, realize data exchange and notification; 5. Power system: Lithium battery: capacity >=500mAh; Power management chip, support dynamic power consumption adjustment; Fast charging technology, support 80% charging in 1 hour; 6. Other components: High-definition camera: >=1080p resolution, field of view >=120°; Ambient light sensor: automatically adjust display brightness; Bone conduction speaker: provides auditory feedback and warnings; Embedded display: provides real-time display of key information; PPG heart rate sensor: integrated into the inner side of the glasses legs, sampling rate ≥ 60Hz; 7. Storage system: Built-in flash memory: ≥ 32GB, for data recording and application storage; Support external SD card expansion storage (optional); 8. Sensor interface: I2C, SPI, UART standard interface; Support extended external sensors; Further, the scheme realizes all-around safety monitoring: through the integration of multiple high-precision sensors such as IMU and heart rate sensor, combined with complex data processing algorithms, it can monitor the emergency braking, falling, head movement and other safety-related states of the rider in real time, multi-level judgment logic and hysteresis processing mechanism, greatly improving the accuracy and reliability of monitoring, discovering potential dangers in time, and providing more comprehensive safety protection for riders, for example, in emergency braking detection, through double-stage low-pass filtering of forward and backward linear acceleration, deceleration threshold analysis, trend analysis and speed estimation multi-step processing, it can accurately judge the braking situation, effectively avoid safety hazards caused by misjudgment; Further, the scheme realizes deep analysis of riding state: it can deeply monitor and analyze the rocking state and riding posture, accurately judge the rocking state through filtering, standard deviation analysis and rocking frequency calculation of vertical linear acceleration and pitch angle; based on head inclination monitoring and motion analysis, it comprehensively evaluates the riding fatigue degree, provides more targeted riding state feedback for riders, helps riders optimize riding habits, and improves riding efficiency and experience, for example, according to the rocking state detection result, the rider can better adjust the riding rhythm and reasonably allocate physical strength; Further, the scheme realizes intelligent integration and cooperation: the intelligent riding glasses are deeply integrated with vision, alarm, health monitoring, navigation and cloud service systems, when an emergency occurs, it can automatically trigger the camera to record and perform environment perception, provide differentiated feedback combined with the alarm system, evaluate the rider's physical state according to the health monitoring data, the navigation system can adjust the route according to the rider's state, and the data can be uploaded to the cloud for long-term analysis, this multi-system integration and cooperation realizes the interconnection and comprehensive utilization of data, and provides more intelligent, convenient and personalized services for riders; Further, the scheme realizes data recording and analysis: with perfect event recording and analysis function, key information of various events such as brake, fall, rocking and fatigue warning can be recorded in detail, and effective screening and statistics are carried out according to screening logic, and the generated cycling report covers safe cycling score, cycling habit analysis, safety suggestion content, helps the cyclist to deeply understand his cycling behavior, finds potential problems, and thus takes corresponding improvement measures, promotes cycling safety and health.

[0029] Further, coordinate system definition and conversion: The IMU sensor in the intelligent cycling glasses usually adopts a device local coordinate system (Device Frame), and when processing navigation, posture and the like, the sensor data needs to be converted to a standard global coordinate system. Common global coordinate systems include: NWU coordinate system (NorthWestUp), ENU coordinate system (EastNorthUp), NED coordinate system (NorthEastDown); Suppose the intelligent cycling glasses adopt the following device coordinate system: x-axis: pointing upwards (perpendicular to the ground); y-axis: pointing to the left (parallel to the face of the cyclist, pointing to the left side of the face); z-axis: pointing to the rear (opposite to the cycling direction); Convert to NWU coordinate system; Gyro_NWU = [-Gyro_z, Gyro_y, Gyro_x]; Accel_NWU = [-Accel_z, Accel_y, Accel_x]; In actual use, the relationship between the device coordinate system and the global coordinate system is determined according to the actual installation direction of the device, the original sensor data is converted, and subsequent processing and algorithm implementation are carried out in a unified coordinate system.

[0030] Sensor fusion algorithm: The improved complementary filter is adopted to realize the attitude estimation. The core of the fusion algorithm is to balance the short-term accuracy of the gyroscope and the long-term stability of the accelerometer.

[0031] Attitude representation: The attitude is represented by a quaternion to avoid gimbal lock; Quaternion definition: where is the scalar part.

[0032] Gyroscope integration: Update the quaternion according to the angular velocity: ; where Indicates angular velocity. Represents quaternion multiplication; Accelerometer correction: Correct attitude based on the direction of gravity: ; Where a is the normalized acceleration, g is the direction of gravity [0,0,1], and R is the rotation matrix corresponding to the current attitude.

[0033] Complementary filter fusion: The contributions from the gyroscope and accelerometer are balanced by adjusting the gain parameters:

[0034] Where K is the gain parameter and dt is the sampling period.

[0035] Gyroscope bias estimation: Real-time update of gyroscope bias while stationary: ; Where α is the learning rate and threshold is the resting threshold.

[0036] Emergency braking detection algorithm: Emergency braking detection is based on forward and backward acceleration (e.g., z-axis) analysis, achieved through multi-level filtering and feature extraction.

[0037] (1) Signal preprocessing: Two-stage low-pass filter: First-stage filter (cutoff frequency: 2.0 Hz): ; Second-stage filtering (cutoff frequency: 1.0 Hz): ; Where x[n] represents the original IMU acceleration data. ; dt is the sampling period. This corresponds to the time constant; (2) Multi-feature deceleration analysis and judgment logic: Threshold analysis: Calculate the mean of the deceleration window (meanDecel) and compare it with the threshold (brakeThreshold); Basic condition: meanDecel >= brakeThreshold; Trend analysis: Linear regression estimation of deceleration trend; Trend determination condition: trendSlope < 0.5f (increasing deceleration trend); trendSlope= ; Velocity Estimation: Estimate velocityChange by acceleration integration; Velocity Decision Condition: velocityChange < 0.5f (velocity decreasing trend) Comprehensive Decision: thresholdBased || (trendBased && velocityBased); (3) Head Motion Compensation: Tracking Yaw Change: yawChange = newestYaw - oldestYaw; Adjusting Brake Detection Threshold: adjustmentFactor = 1.0f + 0.5f * (|yawChange| / yawChangeThreshold); Correcting Detection Results: if (brakingIntensity < brakeThreshold * adjustmentFactor) isBraking = false; (4) Hysteresis Processing: Increasing Detection Stability: if (isBraking); brakeHysteresis = min(brakeHysteresis + 1, 10) else; brakeHysteresis = max(brakeHysteresis 1, 0); isBraking = brakeHysteresis > 3; Rocking State Detection Algorithm: Rocking state detection mainly focuses on the vertical movement of the rider's body, recognizing the typical cycling rocking mode by analyzing the acceleration signal characteristics; 1. Signal Preprocessing: Vertical acceleration low-pass filter: cutoff frequency 3.0Hz; Pitch low-pass filter: cutoff frequency 2.0Hz; 2. Double Window Analysis Method: Long window (1.0s): for stability analysis; Short window (0.5s): for fast response detection; 3. Dynamic Feature Extraction: Acceleration standard deviation calculation: ; Peak identification and counting: isPeak = (a[i] > a[i-1] && a[i] > a[i+1]); Swing frequency estimation: f = PeakCount / (windowSize / sampleRate); Asymmetric hysteresis mechanism: Swaying state needs to be accumulated continuously to trigger (through a slowly growing counter), but once the swing stops, the state will quickly recover (through a rapidly decaying counter); Slow growth: swayingDuration += 1 (when swing is detected); Fast decay: swayingDuration = max(0, swayingDuration 5) (when swing is not detected); 5. Swaying state determination: Basic condition: (accStd > STD_THRESHOLD) || (accStdShort > STD_THRESHOLD *1.2f); Duration requirement: swayingDuration > 2.0f * sampleRate; Intensity calculation: rockingIntensity = max(accStd / STD_THRESHOLD_MAX, accStdShort / (STD_THRESHOLD_MAX*1.2f)); Fall detection algorithm: The fall detection algorithm focuses on abnormal motion patterns during cycling, achieving high accuracy in fall recognition through multi-modal feature fusion; 1. Impact detection: Combined acceleration calculation: ; Impact threshold determination: when |a| > IMPACT_THRESHOLD (e.g. 3.5g), trigger impact event; 2. Attitude change monitoring: Attitude change rate calculation: ; Threshold determination: when Δθ / Δt > ATTITUDE_RATE_THRESHOLD, record attitude sudden change event; 3. Free fall detection: Near-zero gravity state recognition: when ||a|g| < FREE_FALL_THRESHOLD and duration exceeds a certain value; Usually followed by a short free-fall state before falling down; 4. Static state determination: Acceleration variance calculation: ; Static determination: when Var(a) < STATIC_THRESHOLD and duration exceeds STATIC_DURATION; 5. Post-fall state analysis: Posture stability analysis: usually small posture change but large difference from normal riding posture after falling down; Motion state analysis: usually short intense motion followed by near-static state after falling down; 6. Fall event determination process: Warning stage: detect impact event or sudden posture change; Confirmation stage: analyze fall feature sequence (impact -> posture change -> possible static state); Event confirmation: feature sequence matches preset pattern and confidence exceeds threshold; Emergency response: trigger alarm, automatic recording, optional automatic help; 7. False alarm suppression mechanism: Accidental impact filtering: distinguish between fall impact and other impacts (such as bumpy road); Normal dismounting discrimination: analyze dismounting action features and distinguish from falling down; Environmental factor consideration: combine visual information to judge environmental conditions; 8. Fall type recognition: Forward falling down: characterized by strong forward impact, posture forward leaning; Lateral falling down: characterized by lateral impact, posture lateral leaning; Backward falling down: characterized by backward impact, posture backward leaning; Riding posture and fatigue monitoring; Long-term posture analysis: Head inclination angle trend monitoring: ; where is the time weight factor, is the sampled posture angle; Posture stability index: ; where is the normal posture angle range; Head micro-motion analysis: Short cycle head movement frequency statistics; Irregular head movement amplitude measurement; Head shaking frequency spectrum analysis; Fatigue assessment model: Synthetic index calculation: ; Where is the posture fatigue index: ; is the optimal riding posture angle, and are weight parameters ( + =1); is the motion fatigue index; ; : is the head nodding frequency; : is the nodding frequency threshold; : is the head movement amplitude; : is the normal movement amplitude reference value; : is the power spectrum of irregular head movement; : is the total power spectrum; , , is the weight parameter, ; is the time fatigue index; ; is the total riding time; is the predefined critical fatigue time (e.g. 60 minutes); is the continuous riding time (without rest); is the recommended maximum continuous riding time (e.g. 120 minutes); is the parameter affecting the continuous riding fatigue growth rate; Fatigue level classification: Mild fatigue: ; Moderate fatigue: ; Severe fatigue: ; Comprehensive status assessment and early warning; 1. Multi-state fusion analysis: State-related analysis: such as fall risk assessment after sudden braking; State transition monitoring: such as the transition from a stable riding state to a rocking state; Comprehensive safety risk score: a weighted score combining multiple status indicators; 2. Tiered early warning mechanism: Level 1 warning: A mild warning, such as a reminder of mild fatigue; Level 2 warning: Attention-based warning, such as excessive swaying of the vehicle; Level 3 Warning: Emergency warning, such as danger of falling or a fall has already occurred; 3. Intelligent feedback adaptation: Different feedback methods are selected based on the warning level; Adjust the feedback intensity according to environmental conditions; Personalized feedback patterns based on user habits; Event logging and analysis; Detected events will be recorded and can be selectively triggered with corresponding actions: 1. Event log format: Event type: Braking / Fall / Rocking / Fatigue warning; Start time; End time; Duration; Maximum strength; Confidence level; Relevant sensor data; 2. Filtering logic: Braking events: Events lasting >0.3 seconds are recorded; Fall events: All fall events with a confidence level > 0.7 were recorded; Car rocking event: Events lasting longer than 2.0 seconds are recorded; Fatigue warning: Severe fatigue state has been recorded; 3. Event Statistics: Total number of events during the ride; Event time distribution; Event intensity distribution; Event correlation analysis; 4. Cycling report generation: Safe riding rating; Cycling habit analysis; Security recommendations generated; Long-term trend analysis; System integration; The application can be integrated with other functional modules of intelligent riding glasses to form a complete riding assistance system: Integrated with visual system: Trigger camera recording when detecting emergency events; Combined with computer vision algorithm for forward obstacle recognition; Environmental perception auxiliary state judgment; Integrated with alarm system: Provide differentiated feedback according to event type and emergency level; Dangerous state through vibration, sound or display interface; Extreme cases (such as falling) automatically send help information; Integrated with health monitoring system: Combined with heart rate monitoring to evaluate the physical state of the rider; Analysis of the relationship between shaking frequency and riding efficiency; Fatigue state and physiological index correlation analysis; Integrated with navigation system: Record the location of dangerous events; Mark dangerous sections and give early warning in future riding; Adjust the route recommendation according to the rider's state; Integrated with cloud service: Upload event data for long-term analysis.

[0038] The embodiments of the application are described in detail above in combination with the drawings, but the application is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the purpose of the application.

Claims

1. A method for monitoring cycling status based on smart cycling glasses, characterized in that: The specific steps include: S1 data acquisition: The smart cycling glasses have multiple built-in sensors to collect key data; Inertial Measurement Unit (IMU): Includes a triaxial accelerometer (range ±16g, sampling rate >= 200Hz), a triaxial gyroscope (range ±2000° / s, sampling rate >= 200Hz), and a triaxial magnetometer (range ±4000μT, sampling rate >= 100Hz), used to acquire acceleration, angular velocity, and magnetic field data; Heart rate sensor: Measured using photoplethysmography, with a sampling rate ≥60Hz, to collect the cyclist's heart rate data; S2 coordinate transformation: Determine the relationship between the actual installation orientation of the smart cycling glasses and the navigation coordinate system (such as the NWU coordinate system), and convert the glasses coordinate system data to the navigation coordinate system data; commonly used conversion formulas are as follows: Gyroscope data conversion: GyroNWU = [-Gyroz, Gyroy, Gyrox] Accelerometer data conversion: AccelNWU = [-Accelz, Accelly, Accelx] S3 sensor fusion and attitude acquisition; S4 multi-state monitoring; The S4 multi-state monitoring includes S401 emergency braking detection, S402 deceleration characteristic analysis, S403 head movement monitoring, S404 fall detection, S405 swaying state detection, and S406 riding posture monitoring. S5 Comprehensive Assessment and Early Warning; S6 Event Recording and Analysis; S7 System Integration Assistance.

2. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: The S3 sensor fusion and attitude acquisition: An improved complementary filter is used to achieve attitude estimation, and the attitude is represented by quaternions; The specific steps are as follows: S301 Gyroscope Integration: Integrate the angular velocity data from the gyroscope to obtain the predicted attitude value; S302 Accelerometer Correction: Uses accelerometer data to correct attitude and eliminate drift error of gyroscope integration; S303 Complementary Filter Fusion: Weighted fusion of the gyroscope's predicted value and the accelerometer's corrected value to obtain a more accurate attitude estimate; S304 Gyroscope Bias Estimation: Real-time estimation of gyroscope bias to further improve the accuracy of attitude estimation. Based on the obtained attitude quaternion, the influence of gravity is eliminated and linear acceleration is extracted.

3. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: The S401 emergency braking detection; Data filtering: The forward and backward linear accelerations are subjected to a two-stage low-pass filter, with a first-stage cutoff frequency of 2.0Hz and a second-stage cutoff frequency of 1.0Hz.

4. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: Analysis of the deceleration characteristics of S402: Threshold analysis: A deceleration threshold is set. When the deceleration exceeds this threshold, emergency braking is considered to be possible. Trend analysis: Analyze the trend of deceleration changes to determine whether it is a sharp deceleration; Speed ​​estimation: By combining acceleration data and time information, the change in cycling speed is estimated.

5. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: The S403 head motion monitoring: extracts and tracks changes in head yaw angle, and dynamically adjusts the braking detection threshold according to the head motion state; Multi-level decision logic: The emergency braking detection result is output through multi-level decision logic and hysteresis processing.

6. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: The S404 fall detection: Impact event detection: Monitor the peak value of the combined acceleration. When |a| > IMPACT_THRESHOLD (e.g., 3.5g), an impact event is triggered. Attitude change rate analysis: Analyze the attitude change rate, and record the attitude sudden change event when Δθ / Δt>ATTITUDE_RATE_THRESHOLD; Free fall detection: Detects near-zero gravity conditions and identifies the free fall phase; Determining whether a person is stationary after falling: Multi-feature fusion judgment: The system fuses multiple feature sequences, including impact, posture change, free fall, and static state, to determine fall events and classify fall types (forward fall, side fall, backward fall).

7. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: The S405 jacking status detection: Data filtering: Extract the vertical linear acceleration and perform low-pass filtering on the linear acceleration and pitch angle. The cutoff frequency for vertical acceleration is 3.0Hz and the cutoff frequency for pitch angle is 2.0Hz. Standard deviation analysis: Analysis of acceleration standard deviation based on long and short time windows (long window 1.0 second, short window 0.5 second); Swaying frequency calculation: Swaying frequency is calculated by identifying wave peaks; State maintenance mechanism: The swaying state is maintained by applying an asymmetric hysteresis mechanism; State determination: The shaking state is determined based on the standard deviation threshold and duration.

8. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: The S406 riding posture monitoring: Head tilt monitoring: Long-term monitoring of head tilt trends; Head movement analysis: Analyzing the frequency and amplitude of head movements; Fatigue assessment: Cycling fatigue is comprehensively assessed based on multiple indicators, including long-term head posture trends and statistics on the frequency of head micro-movements. Fatigue warning: Generate graded fatigue warnings based on fatigue level.

9. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: The S5 comprehensive assessment and early warning: State correlation analysis: Analyze the correlation and transition patterns of multiple states such as emergency braking, falling, and swaying. Security risk scoring: Calculate the overall security risk score; Tiered early warning mechanism: Implement a tiered early warning mechanism (Level 1: Informative warning; Level 2: Cautionary warning; Level 3: Emergency warning); Intelligent feedback adaptation: Provides intelligent feedback adaptation based on warning level, environmental conditions and user habits.

10. The cycling status monitoring method based on smart cycling glasses according to claim 1, characterized in that: The S6 event recording and analysis: Event logging: When various events such as braking, falling, swaying, and fatigue warning are detected, they are recorded in a specific format, including event type, start time, end time, duration, maximum intensity, confidence level, and related sensor data; Event filtering: Record corresponding events based on filtering logic (e.g., braking event duration > 0.3 seconds, fall event confidence > 0.7, rocking event duration > 2.0 seconds, severe fatigue state); Event statistics: Perform event statistics (total number of events of all types during the ride, time distribution, intensity distribution, correlation analysis) and generate a riding report (safe riding score, riding habit analysis, safety recommendations, long-term trend analysis); S7 System Integration Assistance: Integration with vision systems: When an emergency event is detected, the camera is triggered to record, and computer vision algorithms are used to identify obstacles ahead and assist in environmental perception to determine the status. Integration with alarm systems: Provides differentiated feedback based on event type and urgency; Integration with health monitoring systems: Combines heart rate monitoring to assess the cyclist's physical condition; Integration with navigation systems: Records the location of dangerous events, marks dangerous sections of road and provides advance warnings for future rides, and dynamically adjusts route recommendations based on the rider's condition; Integration with cloud services: Upload event data for long-term analysis.

Images