Intelligent interval cooperative control method and system for bicycle fleet

Abstract

The invention discloses an intelligent interval cooperative control method and system for a bicycle fleet. The method comprises the steps that real-time speed information, real-time position coordinates and bicycle body posture data of all bicycles are collected; the wireless communication module uploads own operation parameters and shares operation parameters of other vehicles in the motorcade; based on the real-time speed information, the real-time position coordinates and the bicycle body posture data, a dynamic safety distance threshold value of the adjacent bicycles is calculated through a dynamic safety distance algorithm; the actual distance between the adjacent bicycles is calculated in real time and compared with a dynamic safety distance threshold value, and then collision risk levels including low risks, medium risks and high risks are determined; light flickering is used for early warning in the middle risk, and sound warning and strong early warning through saddle vibration are used for high risk; based on the real-time speed, the collision risk level and the motorcade driving path, a motorcade-level speed regulation instruction is generated through a cooperative speed regulation algorithm, and the bicycle speed is automatically adjusted. The method is used for solving the safety and efficiency defects of manual cooperation of a traditional bicycle fleet.

Description

Technical Field

[0001] This invention relates to the field of collaborative control technology, specifically to a method and system for intelligent spacing collaborative control of bicycle fleets. Background Technology

[0002] In cycling (such as road cycling and long-distance convoy cycling), coordinated convoy riding is an important way to improve overall efficiency and reduce wind resistance. However, traditional convoy riding relies on riders manually judging distance and adjusting speed, which has the following significant technical shortcomings and makes it difficult to meet the requirements of dynamic safety and coordinated control: (1) The safety distance is static and has poor adaptability. Traditional vehicle convoy safety distances often use fixed values ​​(such as the empirical formula "1 meter / 10 km / h"), without taking into account the dynamic changes in real-time speed, vehicle attitude (such as tilt and pitch), and road conditions (curves and slopes).

[0003] (2) Information exchange is delayed and collaboration is lacking. The status of each vehicle in the convoy (speed, position, attitude) relies on the rider's visual observation, and information transmission has a delay of 0.5-2 seconds, and is easily affected by obstructed vision (such as curves, traffic flow). When a sudden risk occurs (such as the vehicle in front avoiding an obstacle), the following vehicles cannot obtain the dynamics of the vehicle in front in real time, leading to a "chain reaction" collision accident.

[0004] (3) Risk warning is singular and response efficiency is low. Traditional early warning systems rely on riders honking their horns or using hand gestures, which have the following limitations: The warning methods are limited: relying solely on sound or visual signals makes them susceptible to environmental interference (such as noisy road sections or backlit scenes). No graded warning: It cannot distinguish between the risk of "minor approach" and "emergency collision", which may lead riders to misjudge or overreact; Disconnect between warning and control: After the warning is issued, riders rely on manual deceleration, and within the reaction time (about 0.8-1.5 seconds), they may have already entered a dangerous area.

[0005] (4) Independent speed control results in poor vehicle stability. The speed adjustment of each vehicle is entirely operated independently by the rider, which has the following problems: the following vehicle only follows the single target adjustment of the preceding vehicle without considering the overall path of the team (such as the speed distribution inside and outside the curve), which can easily lead to lateral compression or "snake-like" movement.

[0006] (5) Insufficient data acquisition accuracy and weak control foundation Traditional speed and position data acquisition relies on a single sensor (such as a Hall sensor or GPS), which has the following drawbacks: Limitations of Hall effect sensors: They are susceptible to wheel slippage and changes in tire pressure, and cannot reflect the actual driving trajectory; GPS signal dependence: Signal drift in urban high-rise buildings, tunnels and other scenarios, with speed errors reaching ±2m / s, and the position update frequency (1Hz) cannot meet the dynamic control requirements.

[0007] The aforementioned problems result in a high collision risk for traditional bicycle teams in high-speed, complex road conditions, and a difficulty in balancing riding efficiency and safety. Therefore, there is an urgent need for an intelligent control method that can collect multi-dimensional data in real time, dynamically calculate safe distances, provide tiered warnings, and coordinate speed adjustments to achieve overall safety and efficient teamwork for the bicycle team. Summary of the Invention

[0008] In order to overcome the shortcomings of the existing technology, the purpose of this invention is to provide a method and system for intelligent spacing coordination control of bicycle teams, which solves the safety and efficiency defects of traditional manual coordination of bicycle teams, thereby realizing intelligent, precise and coordinated control of team operation.

[0009] To solve the above problems, the technical solution adopted by the present invention is as follows: A method for intelligent spacing coordination control of a bicycle fleet includes the following steps: S1. Collect real-time speed information, real-time position coordinates, and vehicle posture data of each bicycle through the on-board sensor module; S2. Each bicycle uploads its own operating parameters and shares the operating parameters of other bicycles in the team through the wireless communication module; S3. Based on real-time speed information, real-time position coordinates, and vehicle posture data, calculate the dynamic safe distance threshold between adjacent bicycles using a dynamic safe distance algorithm; S4. Calculate the actual distance between adjacent bicycles in real time, compare it with the dynamic safe distance threshold, and determine the collision risk level, including: low risk, medium risk, and high risk; S5. In cases of medium risk, a flashing light will be used as a warning; in cases of high risk, a strong warning will be issued by an audible alarm and seat vibration. S6. Based on real-time speed, collision risk level, and convoy driving path, a convoy-level speed adjustment command is generated through a collaborative speed adjustment algorithm to automatically adjust the bicycle speed.

[0010] Preferably, real-time speed information acquisition includes: The linear velocity of the wheel is collected by a three-axis accelerometer, and combined with GPS positioning speed information. The real-time speed information is obtained after data fusion through dynamic weight allocation and Kalman filtering.

[0011] Preferably, data fusion includes: The GPS positioning speed information and linear velocity are synchronized in time and data interpolated and adapted. The fusion weights are dynamically allocated according to the GPS signal strength, and noise is eliminated by Kalman filtering to output the fused real-time speed information. Specifically, when the GPS signal strength is less than or equal to a preset threshold, the fusion weight of the positioning speed information is reduced to 0.2-0.4; when the GPS signal strength is greater than the preset threshold, the fusion weight of the positioning speed information is increased to 0.5-0.7.

[0012] Preferably, real-time location coordinate acquisition includes: Raw location coordinate data is collected using a GPS locator, combined with motion trend data and heading angle change rate from a three-axis accelerometer, and after outlier removal and timestamp alignment, real-time location coordinates are generated using a multi-source fusion algorithm. When the GPS signal strength is greater than or equal to the threshold, the motion trend data is fused using the preprocessed original position coordinate data as a reference to correct the positioning drift by using the Kalman filter algorithm to fuse the motion trend data. When the GPS signal strength is less than the threshold, the position coordinates are predicted in real time by extended Kalman filtering based on the effective position coordinates, linear velocity and heading angle change rate of the previous moment. The confidence weight of the estimated position decreases linearly with the duration of GPS signal interruption.

[0013] Preferably, vehicle body posture data acquisition includes: Raw three-dimensional acceleration data is collected by a triaxial accelerometer. After outlier removal and zero drift calibration, the roll angle and pitch angle of the vehicle body are calculated based on the acceleration components of the X, Y, and Z axes. The roll angle and pitch angle data are smoothed by a second-order Butterworth low-pass filter with a cutoff frequency of 5Hz before being output and timestamped with real-time speed information and real-time position coordinates.

[0014] Preferably, the dynamic safety distance threshold calculation includes: The longitudinal basic safety distance is calculated based on the real-time speed information of the vehicle in front, the preset driver reaction time, and the maximum braking deceleration of the bicycle. The maximum braking deceleration of the bicycle is dynamically corrected with the pitch angle. The actual lateral offset is calculated based on the real-time lateral difference between the rear vehicle and the front vehicle, the preset bicycle wheelbase parameters, and the roll angle. The lateral safety distance correction term is calculated based on the actual lateral offset and real-time speed information, combined with the lateral safety factor. The dynamic safety distance threshold is obtained by vector synthesis of the longitudinal basic safety distance and the lateral safety distance correction term; When the convoy enters a curve, the lateral safety factor automatically increases to 1.6-1.8.

[0015] Preferably, the actual spacing calculation includes: Based on the real-time location coordinates and driving direction of each bicycle in the convoy, the path planning algorithm is used to dynamically identify adjacent bicycle pairs, and the latitude and longitude coordinates are converted into plane rectangular coordinates by UTM projection based on the curvature radius of the trochanter circle, the curvature radius of the meridian circle, and the longitude of the central meridian. The projection parameters are dynamically matched based on the current geographical location of the convoy. Based on Cartesian coordinates, the actual distance is calculated using the Euclidean distance formula. When the elevation difference exceeds a threshold, an elevation correction term is introduced for three-dimensional correction, and the result is output after sliding window averaging filtering.

[0016] Preferably, the generation of fleet-level speed control commands includes: When the vehicle ahead poses a high risk, a deceleration command is generated, including: Based on the real-time speed of the preceding vehicle and the dynamic safe distance threshold, calculate the number of following vehicles that need to slow down to ensure that at least two following bicycles are covered. For each vehicle that needs to decelerate, a distance weight is calculated, where the distance weight increases as the actual distance between the following vehicle and the preceding vehicle decreases. The target deceleration magnitude for each following vehicle that needs to decelerate is generated based on its current speed, risk level coefficient, and distance weight. When a following vehicle receives deceleration commands from multiple preceding vehicles simultaneously, the command from the closest preceding vehicle is prioritized and time-slice rotation is performed.

[0017] Preferably, the generation of fleet-level speed control commands also includes: When the entire convoy enters the curve, the speed adjustment commands generated include: Calculate the radius of curvature of the curve based on the bicycle wheelbase and average roll angle. ; The radial distance relative to the center of the curve in the Cartesian coordinate system is less than The vehicles marked as inside the curve bicycles, larger than The markings indicate the outer bicycle; Calculate the speed increase of the inner bicycle based on the distance deviation between the inner vehicle and the center of the curve and the current speed of the inner bicycle. Calculate the deceleration magnitude of the outer bicycle based on the distance deviation between the outer vehicle and the center of the curve and the current speed of the outer bicycle; By dynamically increasing the speed increase of the inner bicycle and the speed decrease of the outer bicycle, the lateral distance between adjacent bicycles is made greater than or equal to the safe lateral distance threshold for curves.

[0018] A smart spacing collaborative control system for bicycle fleets includes: Vehicle-mounted sensor module: used to collect real-time speed information, real-time position coordinates, and vehicle attitude data of each bicycle; Wireless communication module: used to upload its own operating parameters and share the operating parameters of other vehicles in the fleet; Central processing unit: Based on real-time speed information, real-time position coordinates and vehicle posture data, it calculates the dynamic safety distance threshold between adjacent bicycles through a dynamic safety distance algorithm; it calculates the actual distance between adjacent bicycles in real time, compares it with the dynamic safety distance threshold to determine the collision risk level, including: low risk, medium risk and high risk; based on real-time speed, collision risk level and fleet driving path, it generates fleet-level speed adjustment commands through a collaborative speed adjustment algorithm. Multimodal alert module: used to issue a warning with flashing lights when there is a medium risk, and to issue a strong warning with sound alarm and seat vibration when there is a high risk; Collaborative speed control execution module: Used to receive team-level speed control commands and automatically adjust bicycle speed.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) Dynamic safety distance is accurately adapted to complex scenarios Dynamic safety distance is precisely adapted to complex scenarios: By integrating real-time speed, vehicle attitude (roll angle, pitch angle) and road conditions (curves, slope), the safety distance threshold is dynamically calculated, solving the problem of poor adaptability of traditional fixed distance formulas. UTM projection coordinate transformation and three-dimensional Euclidean distance calculation are adopted, combined with an elevation correction term, to eliminate the error of traditional planar distances ignoring terrain undulations; the actual distance calculation error is <0.3m.

[0020] Collision risk classification and prevention: Based on the comparison between actual distance and dynamic threshold, the risk is divided into three levels: low, medium and high, corresponding to multimodal warnings, which solves the shortcomings of traditional warnings that are single and without classification, and reduces the rider's misjudgment rate.

[0021] (2) The vehicle team's collaborative control capability has been significantly enhanced. Fleet-level dynamic speed regulation and coordination optimization: Real-time fleet data sharing is achieved through wireless communication modules, and global commands are generated based on a coordinated speed regulation algorithm: When the vehicle in front is at high risk, differentiated deceleration commands are sent to at least two following vehicles (the closer the distance, the greater the deceleration); when cornering, the inner vehicle accelerates while the outer vehicle decelerates to maintain a safe lateral distance and avoid "snake-like" maneuvers or side collisions. A command conflict handling mechanism is provided to ensure the orderly coordination of multi-vehicle responses.

[0022] Improved adaptability to curves and complex paths: The curve status is determined based on the roll angle, the radius of curvature is calculated by the wheelbase and the average roll angle, the inner and outer vehicles are dynamically divided, and the speed adjustment range is adjusted according to the distance deviation (the inner side speeds up to compensate for the radius difference, and the outer side decelerates to reduce centrifugal force), so as to achieve closed-loop control of the safe distance in curves.

[0023] (3) The efficiency of early warning and response has been greatly improved. Multimodal warning combined with active control: Breaking through the limitations of traditional horn / gesture warnings, medium-risk warnings are triggered by flashing lights (visual), while high-risk warnings are triggered by sound and vibration (multi-sensory stimulation), improving the effectiveness of warnings in complex environments (noisy, backlit). In high-risk situations, a deceleration command is automatically generated, eliminating the need for manual rider operation and shortening response latency (traditional manual reaction time is 0.8-1.5 seconds, while system response is ≤100ms).

[0024] Precise coverage of risk propagation: When the vehicle ahead is at high risk, the number of vehicles that need to slow down is dynamically calculated based on the speed (at least 2 vehicles, or 3 vehicles if the speed is 20m / s). The deceleration range is allocated by distance weighting to achieve "advanced transmission" of risk and avoid chain collisions.

[0025] (4) Data acquisition and fusion accuracy has been significantly optimized. Multi-source data fusion enhances reliability: Velocity acquisition integrates a triaxial accelerometer (high frequency ≥100Hz) and GPS (absolute velocity). Through dynamic weight allocation and Kalman filtering, the error is controlled within ±0.3m / s, resolving the issues of mechanical error or signal drift from a single sensor. Position acquisition combines GPS and a triaxial accelerometer. When the signal is weak, dead reckoning (extended Kalman filter to predict position) is initiated. The confidence weight decreases linearly with the duration of the interruption, ensuring data continuity in scenarios such as tunnels and high-rise buildings (output frequency ≥100Hz).

[0026] High-precision vehicle attitude data acquisition: Roll and pitch angles are calculated using a three-axis accelerometer (sampling ≥100Hz, resolution ≥16-bit), smoothed by a 5Hz second-order Butterworth filter, and aligned with the timestamps of speed and position data to provide accurate input for dynamic safety distance and cornering control.

[0027] (5) Enhanced system robustness and environmental adaptability Ensuring data reliability in complex scenarios: Preprocessing such as outlier removal (3σ criterion), time synchronization, and data interpolation adaptation, combined with sliding window filtering (window = 5 cycles), eliminates fluctuations caused by road bumps and sensor noise, improving spacing data stability by over 40%. The wireless communication module supports low-power Bluetooth / LoRa networking with a transmission latency of ≤100ms, ensuring real-time data interaction within a 50-meter range for the fleet.

[0028] Multi-model and cycling scenario compatibility: Supports dynamic wheelbase parameter adaptation (1.1-1.3m) to adapt to mixed formations of different models such as mountain bikes and road bikes; Response time (0.8-1.2s adjustable) caters to the needs of both professional and amateur teams, balancing safety and riding efficiency.

[0029] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0030] Figure 1 This is a flowchart illustrating the steps of the intelligent spacing collaborative control method according to an embodiment of the present invention; Figure 2 This is a flowchart of the real-time location coordinate acquisition process according to an embodiment of the present invention; Figure 3 This is a flowchart of the vehicle posture data acquisition process according to an embodiment of the present invention. Detailed Implementation

[0031] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0032] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0033] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.

[0034] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0035] Example 1, see Figure 1 The present invention provides a step-by-step diagram of an intelligent spacing collaborative control method, as shown in the figure. Figure 1 The method for intelligent spacing coordination control of a bicycle fleet, as shown, includes the following steps: S1. Real-time acquisition of multi-dimensional data: Through the on-board sensor module installed on each bicycle in the fleet, multi-dimensional operating parameters of each bicycle are collected in real time. The multi-dimensional operating parameters include: real-time speed information, real-time position coordinates and bicycle posture data. S2. Team Dynamic Information Interaction: Each bicycle uploads multi-dimensional operating parameters to the central processing unit via a wireless communication module, and receives operating parameters from other bicycles in the team, enabling real-time sharing of the operating status of multiple bicycles in the team; S3. Dynamic safety distance calculation: The central processing unit calculates the dynamic safety distance threshold between adjacent bicycles based on the real-time speed information, real-time position coordinates and vehicle attitude data of each bicycle in the fleet, using a preset dynamic safety distance algorithm. S4. Graded Risk Warning and Judgment: The central processing unit calculates the actual distance between adjacent bicycles in real time and compares the actual distance with the dynamic safe distance threshold. Based on the comparison result, the collision risk level is determined. The collision risk levels include: low risk, medium risk and high risk. S5. Multimodal warning prompt: When the risk level is determined to be medium, a warning will be issued by flashing lights through the multimodal prompt module of the corresponding bicycle; when the risk level is determined to be high, the multimodal prompt module will issue a strong warning by simultaneously sounding an alarm (such as a buzzer) and vibrating the seat. S6. Team Cooperative Speed ​​Control: Based on the real-time speed of each bicycle, the collision risk level, and the overall driving path of the team, the central processing unit generates team-level speed control commands through a preset cooperative speed control algorithm, and automatically adjusts the bicycle speed.

[0036] In one possible embodiment, when the actual distance is greater than 1.2 times the dynamic safety distance threshold, the collision risk level is determined to be low risk; when the actual distance is less than 0.8 times the dynamic safety distance threshold and less than or equal to 1.2 times the dynamic safety distance threshold, the collision risk level is determined to be medium risk; and when the actual distance is less than or equal to 0.8 times the dynamic safety distance threshold, the collision risk level is determined to be high risk.

[0037] In one possible embodiment, the collaborative speed control algorithm includes: when the preceding vehicle is determined to be at high risk, sending a deceleration command to at least two subsequent bicycles, and the deceleration magnitude of the following bicycles increases as the distance to the preceding bicycle decreases; when the entire convoy enters a curve, sending an acceleration command to the bicycles inside the curve and a deceleration command to the bicycles outside the curve, in order to maintain a safe lateral distance within the curve and achieve dynamic collaborative optimization of the overall convoy distance.

[0038] Background Description: In collaborative cycling scenarios, real-time, high-precision speed information is the core foundation for ensuring dynamic safe spacing control and coordinated speed adjustment within the cyclist convoy. Traditional bicycle speed acquisition methods suffer from the limitations of relying on a single sensor, making it difficult to meet the high reliability and real-time requirements of collaborative cyclist control for speed data. Specifically: (1) Shortcomings of a single Hall effect sensor Traditional bicycle speed measurement often relies on Hall effect sensors mounted on the wheel (which calculate linear velocity by detecting the rotation period of the magnetic trigger element on the spokes). Their advantages include a high sampling frequency (typically ≥50Hz) and the ability to reflect instantaneous changes in wheel rotation, but they also have significant drawbacks: Susceptible to mechanical errors: such as wheel slippage (when braking or on a wet road surface) and changes in tire pressure causing changes in wheel radius, all of which can cause deviations in linear velocity calculation; Lack of absolute speed reference: Based only on the relative speed of wheel rotation, it cannot reflect the absolute speed of the bicycle's actual travel trajectory (e.g., the wheels rotate when pushing the bicycle but the actual displacement is zero).

[0039] (2) The inadequacy of GPS positioning speed alone GPS locators can provide absolute speed information based on satellite signals, but their limitations are also obvious: Low update frequency: Conventional GPS modules have a speed update frequency of only 1Hz, which cannot capture the frequent dynamic speed changes (such as sudden deceleration and acceleration) during convoy driving. Strong signal dependence: In urban high-rise buildings, tunnels, and shady areas, GPS signals are easily blocked, causing positioning drift with a speed error of ±2m / s or more, which cannot meet the requirements of safety control.

[0040] (3) Special requirements of vehicle fleet cooperative control for speed data When a bicycle convoy is in motion, a dynamic safe distance needs to be calculated based on the speed difference between adjacent bicycles, and coordinated speed adjustment commands (such as speed differentiation control between bicycles on the inside and outside of curves) need to be generated based on real-time speed. If there is a delay (>100ms) or error (>0.5m / s) in the speed data, it may result in: Distorted safety distance calculations and delayed collision risk warnings; Delayed response to speed control commands can cause drastic fluctuations in convoy spacing, and even side collisions. Based on this: In step S1 above, the real-time speed information is collected in real time, including: The linear velocity of the wheel is collected in real time by a triaxial accelerometer installed in the middle of the bicycle frame seat tube; Real-time speed information is obtained by combining the positioning speed information collected by the GPS locator with the linear speed through dynamic weight allocation and Kalman filtering. Among them, the sampling frequency of the triaxial accelerometer is ≥100Hz, the positioning speed information update frequency is 1Hz, and the output frequency of the real-time speed information obtained after fusion is consistent with the sampling frequency of the triaxial accelerometer.

[0041] In this embodiment of the invention, it is necessary to further explain that the proposed multi-source velocity acquisition technology of "triaxial accelerometer + GPS + data fusion" combines the high-frequency sampling (≥100Hz) of the triaxial accelerometer with the absolute velocity reference of GPS. Data fusion compensates for the shortcomings of a single sensor, ensuring that the velocity data can reflect instantaneous dynamics (such as sudden deceleration) and correct for mechanical errors (such as wheel slippage). Simultaneously, through a data fusion strategy of dynamic weight allocation (adjusting weight ratios based on GPS signal strength) and Kalman filtering (eliminating noise interference), the following is ultimately achieved: High-frequency real-time performance: The fused speed output frequency is consistent with that of the three-axis accelerometer (≥100Hz), meeting the millisecond-level response requirements of the vehicle fleet dynamic control; High precision and reliability: When the GPS signal is good, the speed error is controlled within ±0.3m / s through data fusion; when the signal is weak, the relative accuracy of the triaxial accelerometer can still be maintained to avoid data failure. Environmental robustness: Adaptable to complex scenarios such as cities, mountains, and tunnels, laying a key data foundation for intelligent spacing and collaborative control of bicycle fleets.

[0042] In one possible embodiment, data fusion combining dynamic weight allocation and Kalman filtering includes: Time synchronization processing: Align the timestamp of the positioning speed information with the timestamp of the linear velocity to ensure that they are based on the same time reference; Data interpolation adaptation: Linear interpolation or spline interpolation is performed on the positioning velocity information to generate a continuous velocity data sequence that matches the sampling frequency (≥100Hz) of the triaxial accelerometer. Dynamic weight allocation: The fusion weights for positioning speed information and linear velocity are dynamically allocated based on the GPS signal strength. Specifically, when the GPS signal strength is less than or equal to a preset threshold (e.g., ≤30dBm), the fusion weight of the positioning speed information is reduced to 0.2-0.4; when the GPS signal strength is greater than the preset threshold, the fusion weight of the positioning speed information is increased to 0.5-0.7. Kalman filter fusion: The weighted positioning velocity information and linear velocity are used as the observation input of the Kalman filter algorithm. The optimal estimated velocity is calculated iteratively by constructing state equations and measurement equations to eliminate noise interference and output fused velocity data.

[0043] In this embodiment of the invention, it is necessary to further explain that the fusion weight of linear velocity and GPS positioning speed is dynamically allocated based on GPS signal strength. The core objective is to prioritize and trust more reliable sensor data in different environments. (1) Weight allocation logic Strong GPS signal (e.g., in open areas, signal strength > 30dBm): GPS provides absolute position and velocity references, which can correct the cumulative deviation of the three-axis accelerometer. In this case, a high weight of 0.5-0.7 is assigned.

[0044] When GPS signal is weak (e.g., in urban high-rise buildings or tunnels, signal strength ≤30dBm): GPS positioning drift increases and reliability decreases. In this case, its weight is reduced to 0.2-0.4, and dynamic response is mainly maintained by high-frequency sampling (≥100Hz) of the triaxial accelerometer.

[0045] Kalman filtering is a recursive estimation algorithm that iteratively optimizes state prediction and measurement updates to eliminate noise interference and output the optimal velocity estimate. The specific steps are as follows: State equation construction: Predict the current velocity state based on physical motion models (such as uniformly accelerated motion), taking into account system noise (such as sensor drift).

[0046] Measurement equation update: The GPS velocity and linear velocity after dynamic weight allocation are used as observation inputs. The residual (observation error) is calculated by comparing the measurement matrix with the predicted state.

[0047] Optimal estimation iteration: The prediction state is corrected by Kalman gain (balancing the reliability of prediction and observation) to obtain the optimal velocity after fusion; the covariance matrix is ​​dynamically updated to adapt to changes in noise characteristics (such as changes in observation noise caused by GPS signal fluctuations).

[0048] Through the above mechanism, this embodiment achieves the complementary advantages of "high-frequency dynamic response + absolute reference calibration", effectively improving the reliability and applicability of speed data.

[0049] Background Description: In collaborative cycling scenarios, high-precision, real-time location coordinates are the core foundation for achieving dynamic safe distance calculation and platoon path coordination. Traditional bicycle positioning methods rely on a single GPS module, which has significant technical limitations and cannot meet the stringent location data requirements of collaborative platoon control. Therefore: See Figure 2 The flowchart for real-time location coordinate acquisition is as follows: In step S1 above, the real-time location coordinate acquisition includes: Raw location coordinate data (including longitude, latitude and elevation information) is collected at a frequency of 1Hz using a GPS locator, and the vehicle's motion trend data and heading angle change rate are collected using a three-axis accelerometer. Preprocess the raw location coordinate data: Outlier removal: Identify and remove GPS positioning jump data (such as single positioning deviation > 5m) based on the 3σ criterion; Timestamp alignment: aligning the timestamps of the original location coordinate data with the timestamps of the real-time velocity information; Real-time location coordinates are generated using a multi-source fusion algorithm: When the GPS signal strength is ≥30dBm, the preprocessed original position coordinate data is used as the reference, and the motion trend data of the three-axis accelerometer is fused by the Kalman filter algorithm to correct the positioning drift. When the GPS signal strength is <30dBm, the dead reckoning mode is activated: based on the effective position coordinates of the previous moment, the linear velocity and heading angle change rate collected by the three-axis accelerometer, the position coordinates are predicted in real time through extended Kalman filtering, and the confidence weight of the estimated position decreases linearly with the duration of GPS signal interruption. The output frequency of the real-time position coordinates is adapted to ≥100Hz through linear interpolation, which is consistent with the sampling frequency of the triaxial accelerometer.

[0050] In this embodiment of the invention, it should be further explained that the real-time position coordinate acquisition is achieved through the collaboration of a GPS locator and a triaxial accelerometer, with the two functions complementing each other to meet the requirements of high precision and high reliability. GPS locator: Installed on the bicycle frame, it collects raw location coordinate data at a frequency of 1Hz, including longitude, latitude and elevation information, and provides absolute location reference.

[0051] Three-axis accelerometer: Fixed in the middle section of the frame riser tube, it collects vehicle motion trend data (such as acceleration, tilt angle) and heading angle change rate to help correct position deviations during dynamic motion.

[0052] Preprocessing the raw GPS data eliminates noise and time synchronization errors, laying the foundation for subsequent data fusion.

[0053] Based on a dynamic switching and fusion strategy using GPS signal strength, location accuracy is guaranteed across all scenarios. Strong GPS signal (≥30dBm): Based on the preprocessed GPS coordinates, the motion trend data of the three-axis accelerometer is fused using the Kalman filter algorithm to correct GPS positioning drift (such as multipath effect error in urban environments).

[0054] Weak GPS signal (<30dBm): Dead reckoning mode activated: Based on the previous valid position coordinates, linear velocity (reflecting distance traveled), and rate of change of heading angle (reflecting direction of travel), the position is predicted in real time using an extended Kalman filter. The confidence weight of the estimated position decreases linearly with the duration of GPS signal interruption, avoiding error accumulation due to long-term estimation.

[0055] Finally, to match the high real-time requirements of vehicle fleet cooperative control for position data (such as dynamic safety distance calculation and cornering cooperative speed adjustment), the position coordinate output frequency is increased to ≥100Hz through linear interpolation to keep in line with the sampling frequency of the three-axis accelerometer, ensuring that the data update rate meets the control response requirements.

[0056] Background Description: Vehicle posture (such as tilt angle and lean angle) is a core parameter reflecting vehicle stability, directly affecting key functions such as cornering, braking safety, and dynamic balance control. For example, excessive tilting during cornering can lead to rollover risk; abnormal changes in pitch angle during acceleration or braking can cause front or rear wheel slippage. Therefore, real-time and accurate acquisition of vehicle posture data, and its integration with multi-dimensional operating parameters such as speed and position, is fundamental to achieving intelligent functions such as dynamic safety distance calculation, cornering cooperative control, and rider posture warning. Based on this: See Figure 3 The flowchart for vehicle body posture data acquisition is as follows. In step S1 above, when acquiring vehicle body posture data in real time, the process includes: The three-dimensional acceleration data of the vehicle body is collected by a triaxial accelerometer. The measurement axes of the triaxial accelerometer are aligned with the vehicle body coordinate system (X-axis points forward along the longitudinal direction of the vehicle body, Y-axis points to the left along the transverse direction of the vehicle body, and Z-axis points upward perpendicular to the ground). The sampling frequency is ≥100Hz and the resolution is ≥16 bits. Preprocessing of the acquired raw three-dimensional acceleration data: Outlier removal: Based on the 3σ criterion, abnormal data caused by road bumps or sensor vibrations are identified and removed. When the absolute value of uniaxial acceleration is greater than 3g (g is the acceleration due to gravity), it is determined to be an outlier and replaced by linear interpolation. Zero-drift calibration: With the bicycle stationary (the change in acceleration of all three axes is <0.1 m / s² for 3 consecutive seconds). 2 The average acceleration of each axis is calculated as the zero drift compensation reference value, and the zero drift compensation reference value is subtracted from the original three-dimensional acceleration data to eliminate the sensor static error. Calculate vehicle attitude data based on preprocessed 3D acceleration data: Roll angle θ (body tilt angle): Calculated using the formula θ=arctan2(acc_Y, acc_Z), where acc_Y is the Y-axis acceleration component (unit: m / s²). 2 ), acc_Z is the Z-axis acceleration component; Pitch angle φ (vehicle roll angle): Calculated using the formula φ=arctan2(-acc_X, sqrt(acc_Y)). 2 + acc_Z 2 The calculation is performed, where acc_X is the X-axis acceleration component; The calculated vehicle body attitude data is then filtered and optimized. A second-order Butterworth low-pass filter with a cutoff frequency of 5Hz is used to smooth the roll angle and pitch angle data to eliminate high-frequency vibration noise. The output frequency of the filtered vehicle attitude data is consistent with the sampling frequency of the three-axis accelerometer (≥100Hz). The filtered vehicle attitude data is timestamped with real-time speed information and real-time position coordinates to ensure that the vehicle attitude data and other multi-dimensional operating parameters are based on the same time reference, which is used for dynamic safety distance calculation and curve cooperative control.

[0057] In this embodiment of the invention, it is necessary to further explain that the core hardware for acquiring the vehicle posture data is a three-axis accelerometer, which is fixed in the middle section of the bicycle frame stem tube to ensure that the sensor is rigidly connected to the vehicle body, reduce relative vibration interference during riding, and the sensor's measurement axis is strictly aligned with the vehicle body coordinate system.

[0058] Sampling frequency ≥ 100Hz: ensures capture of rapid dynamic changes in vehicle body posture (such as sharp bends and sudden bumps); resolution ≥ 16 bits: ensures the quantization accuracy of acceleration data, with the smallest resolvable acceleration ≤ 0.001g (g is the acceleration due to gravity).

[0059] This embodiment uses preprocessed three-dimensional acceleration data to calculate the vehicle body attitude angles through trigonometric function relationships: Roll angle θ (body tilt angle): The angle at which the body rotates around the X-axis, reflecting the degree of tilt (such as tilting when leaning into a corner). Principle: When stationary, the Y-axis acceleration component acc_Y≈0, the Z-axis acceleration component acc_Z≈g (g is the acceleration due to gravity), and the roll angle θ≈0°; when the body tilts to the left, the Y-axis acceleration component increases, the Z-axis acceleration component decreases, and θ increases (positive value); when tilting to the right, θ is negative.

[0060] Pitch angle φ (vehicle roll angle): The angle at which the vehicle body rotates around the Y-axis, reflecting the degree of forward and backward tilting (e.g., pitching uphill and tilting downhill). Principle: When stationary, the X-axis acceleration component acc_X≈0, and the pitch angle φ≈0°; when going uphill, the vehicle body tilts forward, the X-axis acceleration component is negative (-acc_X is positive), and the pitch angle φ increases (positive value); when going downhill, φ is negative.

[0061] Filter selection: A second-order Butterworth low-pass filter with a cutoff frequency of 5Hz is selected. Its characteristics include a flat amplitude-frequency response within the passband, effectively suppressing high-frequency noise (>5Hz) while preserving low-frequency dynamic changes in attitude angles (such as the tilting trend during normal riding). The filtered data output frequency is consistent with the sensor sampling frequency (≥100Hz) to ensure a balance between real-time performance and smoothness.

[0062] Vehicle attitude data needs to be timestamped with parameters such as real-time speed and position coordinates to ensure that all data are based on the same time base, providing spatiotemporally consistent input for subsequent dynamic safety distance calculations (such as curve cooperative control).

[0063] Background Description: In traditional bicycle team safety distance calculations, longitudinal distance often uses a fixed formula (such as "speed × reaction time + braking distance"), which leads to the following problems: Fixed braking deceleration parameters: The effect of gradient on braking performance was not considered; The reaction time setting is uniform: it does not differentiate the reaction time differences in different riding scenarios (such as novice / experienced rider, straight road / curve), and uses a fixed value (such as 1 second) for all.

[0064] In addition, traditional lateral distance calculations rely solely on the lateral coordinate difference (ΔY) from GPS positioning, without considering the actual offset caused by vehicle tilt, resulting in significant errors.

[0065] Traditional lateral safety distances often use fixed coefficients (such as "lateral offset × 1.2"), failing to consider the amplifying effect of speed on lateral risk, leading to the following problems: Speed ​​insensitivity: At high speeds, the danger of lateral drift increases non-linearly with speed (e.g., if the speed doubles, the lateral collision energy increases to 4 times), and the fixed coefficient cannot accommodate this characteristic.

[0066] Staticizing the lateral safety factor: The system doesn't differentiate between straight and curved scenarios. Lateral centrifugal force increases on curves, requiring higher safety redundancy. A fixed factor (e.g., 1.2) can easily lead to insufficient lateral distance on curves. Therefore: In step S3 above, when calculating the dynamic safety distance threshold between adjacent bicycles using a preset dynamic safety distance algorithm, the following steps are included: S31. Longitudinal base distance calculation: based on the real-time speed information of the vehicle in front. Through formula Calculate the longitudinal foundation safety distance ,in: The preset driver reaction time (range 0.8-1.2s). The maximum braking deceleration of the bicycle (range: 2.5-3.5 m / s²). 2 Furthermore, it dynamically corrects for the pitch angle (φ) in the vehicle's attitude data: when the pitch angle φ > 5° (uphill), Reduce by 10%-15%; when φ < -5° (downhill), Increase by 5%-10%; S32. Lateral offset calculation: based on the real-time lateral difference in position coordinates between the rear vehicle and the front vehicle. ( (Obtained by subtracting the horizontal axis components of the GPS position coordinates of the two vehicles) and the roll angle (θ) from the rear vehicle's body attitude data, using the formula... Calculate the actual lateral offset ,in The preset bicycle wheelbase parameters (range 1.1-1.3m). S33. Lateral safety distance correction: based on the actual lateral offset. And real-time speed information to calculate lateral safety distance correction item The formula is ,in: The horizontal safety factor (range 1.2-1.5). Reference speed (value 10 m / s); S34. Dynamic safety distance threshold synthesis: through formula Calculate the final dynamic safety distance threshold ,when At that time, take To simplify calculations; Among them, when the convoy enters a curve (determined by the roll angle θ being greater than 8° for 3 consecutive seconds in the vehicle attitude data), the lateral safety factor is... Automatically upgraded to 1.6-1.8.

[0067] In this embodiment of the invention, it needs to be further explained that the maximum braking deceleration is adjusted in real time using the pitch angle (φ) in the vehicle attitude data. ): Reduce braking distance by 10%-15% on uphill slopes (φ>5°) and increase it by 5%-10% on downhill slopes (φ<-5°) to ensure that the calculated braking distance matches the actual physical characteristics under different slopes. Simultaneously, the reaction time ( The adjustable range of 0.8-1.2s can be set to flexibly configure according to the riding experience of the team (e.g., 0.8s for professional teams and 1.2s for amateur teams), taking into account both safety and riding efficiency.

[0068] This embodiment introduces a coupled calculation of roll angle and wheelbase: the vehicle body tilt (θ) and wheelbase (θ) are calculated together. Including lateral offset calculations more accurately reflects the actual space occupied by the vehicle body (e.g., when tilting during curves). Compare Increase (To avoid the risk of side collisions).

[0069] Dynamic wheelbase parameter adaptation: Allows adjustment of wheelbase (1.1-1.3m) according to bicycle model (such as mountain bike, road bike), improving the universality of lateral distance calculation when different models are mixed in formation.

[0070] This embodiment introduces a velocity factor. At high speed ( > =10m / s) Automatically increases the correction term to ensure that the risk of high-speed lateral drift is fully covered. When the convoy enters a curve (θ > 8° for 3 consecutive seconds), the lateral safety factor is... The speed has been increased from 1.2-1.5 to 1.6-1.8, enhancing lateral redundancy in curves and preventing rollovers or scrapes.

[0071] This embodiment uses the Pythagorean theorem (… This achieves vector superposition of longitudinal and lateral distances, which better conforms to spatial geometry; when the lateral offset... At that time, take directly This simplifies calculations and improves real-time performance.

[0072] After determining the curve, the lateral safety factor is increased. Indirectly increase , so that the synthesized It automatically adapts to the centrifugal force requirements of curves, achieving a dynamic balance between "efficient calculation for straight lines and enhanced safety for curves".

[0073] Background Description: In cooperative cycling scenarios, real-time, high-precision calculation of the actual distance between adjacent bicycles is crucial for collision risk warning and cooperative speed control. Traditional bicycle platoon distance calculation methods rely on single sensor data or simplified models, which have significant technical limitations and cannot meet the safety control requirements of dynamic platoon environments. These limitations include: static pairing of adjacent bicycles leading to poor dynamic adaptability; data packet loss causing distance calculation interruptions; direct calculation of distance using latitude and longitude resulting in large plane coordinate transformation errors; ignoring the impact of elevation differences on actual distance; and noise in the raw data causing drastic fluctuations in distance. Therefore: In step S4 above, when the central processing unit calculates the actual distance between adjacent bicycles in real time, it includes: Adjacent vehicle pairing and data association: Based on the real-time position coordinates and driving direction of each bicycle in the convoy, the central processing unit uses a path planning algorithm to dynamically identify adjacent bicycle pairs: along the convoy's driving path (a trajectory curve fitted by continuous position coordinates), the vehicle closest to the current bicycle in front is marked as the front bicycle, and the vehicle closest to the rear bicycle is marked as the rear bicycle. The driving direction is determined by the trend of position coordinate changes over 5 consecutive sampling periods (≥50Hz). When the rate of change of displacement direction angle is <3° / s, it is determined to be straight driving; otherwise, it is turning driving. Location data interpolation adaptation: For location coordinates where data packet loss occurs (e.g., wireless transmission interruption > 200ms), linear prediction is performed using the location change rate of the first 3 cycles to supplement the data and ensure data continuity; Conversion from latitude and longitude coordinates to Cartesian coordinates: Converting the real-time position coordinates (longitude) of adjacent bicycles... ,latitude Converted to Cartesian coordinates via UTM projection. , The projection parameters are dynamically matched based on the vehicle's current geographical location (e.g., using latitude zone parameters in the Northern Hemisphere), and the conversion formula is: ; ; in, The radius of curvature of the circle is denoted as . Let be the radius of curvature of the meridian. Longitude of the central meridian; Two-dimensional actual spacing calculation: based on the transformed Cartesian coordinates ( , The actual distance between adjacent bicycles is calculated using the Euclidean distance formula: ; in, , Let the coordinates of the preceding vehicle be the plane coordinates. , Let the plane coordinates of the rear vehicle be; when the elevation difference is... At 1m, an elevation correction term is introduced. : ; Dynamic smoothing filtering: A sliding window averaging filter (window size = 5 sampling periods) is used to smooth the actual spacing data in real time. The formula is as follows: ; in, The original values ​​of the actual distance for the current period and the previous four periods are used. The output frequency of the filtered actual distance data is consistent with the sampling frequency of the real-time position coordinates (≥100Hz).

[0074] In this embodiment of the invention, it is necessary to further explain that, in order to solve the problem of poor adaptability of static pairing, this embodiment dynamically identifies adjacent vehicles based on driving path fitting and direction determination: by fitting the trajectory curve with continuous position coordinates, and combining the displacement direction angle change rate (<3° / s is determined to be straight driving) to distinguish the straight / turning state, it ensures that the "front vehicle-rear vehicle" relationship is accurately matched along the driving path of the convoy, and avoids interference from lateral vehicles.

[0075] To address the packet loss issue in wireless transmission, a linear prediction of the position change rate in the first three cycles is used to supplement missing data, ensuring the continuity of the position coordinate sequence (if the transmission interruption is greater than 200ms, the current position is predicted by historical movement trends), and maintaining the real-time performance of the spacing calculation (output frequency ≥ 100Hz).

[0076] To eliminate spherical coordinate errors, UTM projection transformation is introduced to convert latitude and longitude coordinates into plane rectangular coordinates. , By dynamically matching the central meridian parameters (such as the latitude division of the Northern Hemisphere), the coordinate transformation error in short-distance scenarios is controlled within 0.1%, providing a geometric benchmark for plane spacing calculation.

[0077] Considering the impact of terrain undulations, when the elevation difference between adjacent vehicles is greater than 1m, the actual distance is corrected by the three-dimensional Euclidean distance formula (incorporating the Z-axis elevation component) to avoid underestimating the distance in uphill and downhill scenarios (such as when riding a mountain bike, the elevation difference can reach 3-5m, and the deviation between the planar distance and the actual spatial distance is greater than 10%).

[0078] To reduce the impact of fluctuations in the raw data, a sliding window averaging filter (window size = 5 sampling periods) is used to smooth the spacing data in real time, eliminating high-frequency noise (such as instantaneous position jumps caused by road bumps), improving the stability of the spacing output by more than 40%, and providing reliable input for risk level determination.

[0079] Through the above improvements, this embodiment achieves full-process optimization of "dynamic pairing - data repair - accurate conversion - multiple maintenance correction - smooth output", ensuring the accuracy (error < 0.3m), real-time performance (≥100Hz) and environmental adaptability (adapting to complex scenarios such as cities, mountains, and tunnels) of the actual distance calculation between adjacent vehicles, laying a core data foundation for subsequent graded risk warning and fleet coordinated speed adjustment.

[0080] Background Description: In collaborative cycling scenarios, when the leading bicycle is deemed high-risk due to a sudden situation (such as emergency braking or obstacle avoidance), precise deceleration commands are needed to rapidly transmit risk and coordinate a collaborative response across the cyclist group. Traditional methods of independent braking or simple following deceleration have significant technical limitations, including: insufficient risk propagation range, high rear-end collision risk, unreasonable deceleration magnitude distribution, poor cyclist stability, lack of handling for multiple conflicting commands, and chaotic control logic. Therefore: In step S6 above, when generating fleet-level speed control commands using a preset collaborative speed control algorithm, the following steps are included: When the vehicle in front is deemed high-risk, the deceleration command generated includes: Risk propagation range determined based on the real-time speed of the vehicle ahead. and dynamic safety distance threshold Calculate the number of following vehicles that need to slow down. ,in The function is designed to round up, ensuring that at least the next two bicycles are covered. Distance weighting calculation: For each following vehicle that needs to slow down ,pass Calculate the distance weights, where For each following vehicle The actual distance from the vehicle in front, Follow Decrease and increase (value range 0.3~0.8); Deceleration amplitude generation: via formula Generate the following vehicles that need to slow down. Target deceleration ,in, For each vehicle that needs to slow down The current speed, It is the risk level coefficient (0.4~0.6 for high risk) and satisfies the following conditions: ; Command priority sorting: When a following vehicle receives deceleration commands from multiple preceding vehicles at the same time, the command from the closest preceding vehicle is executed first. Conflicting commands are scheduled in a round-robin fashion through the central processing unit (scheduling period ≤ 50ms).

[0081] In this embodiment of the invention, it is necessary to further explain that by dynamically calculating the number of following vehicles that need to decelerate (covering at least 2 following vehicles), this embodiment ensures that the risk is "transmitted in advance" in the convoy, and reserves sufficient braking distance for the following vehicles (e.g., when the speed of the leading vehicle is 20m / s, covering 3 following vehicles can provide a safety buffer zone of about 60m).

[0082] Dynamically allocate deceleration amplitude based on the actual distance between the following vehicle and the preceding vehicle: The closer the distance, the greater the deceleration: By using a distance weight (0.3~0.8), the vehicle immediately following behind is ensured to decelerate significantly first, avoiding rear-end collisions; Risk level adaptation: When the risk is high, the deceleration effect is enhanced by the risk coefficient (0.4~0.6) to ensure that the actual distance quickly recovers to above the safety threshold.

[0083] Using the "nearest distance first" principle and time-slice round-robin scheduling (≤50ms): Conflict resolution: The following vehicle only executes the deceleration command of the nearest preceding vehicle to avoid multiple commands overlapping; Real-time response: The scheduling cycle is ≤50ms, ensuring that the instruction execution delay is much lower than the human reaction time, thus improving the real-time performance of control.

[0084] Background Description: In collaborative cycling scenarios, curves are high-risk areas for accidents. Traditional curve control methods rely on riders' manual judgment and operation, which has significant technical limitations and fails to meet the requirements for dynamic safety distances and riding stability. Specific problems are as follows: In traditional convoy driving, cornering speeds often employ a "uniform reduction" or "experience-based deceleration" approach, without considering the effects of corner curvature, vehicle position (inner / outer) and centrifugal force. Homogeneity of speed between the inner and outer sides: The inner vehicle has a smaller turning radius. If it maintains the same speed as the outer vehicle, it is prone to excessive body tilt due to insufficient centrifugal force (roll angle θ exceeds the limit); the outer vehicle has a larger turning radius. If the speed is too high, the centrifugal force will surge, which may cause sideslip or side collision accidents.

[0085] Poor curvature adaptability: Different curvatures of curves (such as sharp curves R=10m vs gentle curves R=50m) have significantly different speed requirements. Static deceleration strategies cannot dynamically match curvature changes, leading to the contradiction of "too slow affects efficiency" or "too fast endangers safety".

[0086] Traditional cycling teams rely on riders to visually determine relative positions, which has the following drawbacks: Position determination lag: The dynamic changes in the relative positions of vehicles in a curve, and the delay in manual recognition (usually >0.5s), can easily lead to the inner vehicle not accelerating in time and the outer vehicle not decelerating in time, resulting in a rapid reduction in lateral distance (in actual tests, it can be as low as 50% below the safety threshold).

[0087] Boundary ambiguity: Without a clear standard for dividing the inside and outside (such as the radial distance relative to the center of the curve), it is easy to "misjudge the inside vehicle as the outside" or vice versa, resulting in incorrect speed control command direction (such as the inside vehicle decelerating incorrectly).

[0088] Traditional speed control relies on rider experience and cannot achieve precise speed difference control. Insufficient or excessive acceleration / deceleration: The acceleration / deceleration of the inner side is not correlated with the curvature of the curve or the vehicle's position deviation, which may result in "insufficient acceleration on the inner side to maintain the distance" or "excessive deceleration on the outer side causing a rear-end collision".

[0089] Lack of closed-loop feedback: The lateral spacing is not monitored in real time. When road bumps, crosswind interference, etc. cause the spacing to deviate from the safety threshold, the speed adjustment cannot be dynamically adjusted, and the collision risk continues to approach.

[0090] When driving on curves, lateral safety distance is a key indicator for preventing side collisions, but traditional control methods have limitations: Centrifugal force interference: The centrifugal force of the outer vehicle increases with the square of the speed. If the vehicle is not decelerated in a targeted manner, the lateral distance may drop from the safe threshold (e.g., 2m) to the dangerous value (<1m) within 0.5s.

[0091] Lack of coordination: The fleet operates with vehicles adjusting speed independently, lacking a coordinated mechanism of "inner lane acceleration - outer lane deceleration," easily leading to overlapping driving patterns where "inner lanes are slower and outer lanes are faster," significantly increasing the risk of lateral collisions. Based on this: In step S6 above, when generating fleet-level speed control commands using a preset collaborative speed control algorithm, the following is also included: When the entire convoy enters a curve (determined by the roll angle θ being greater than 8° for 3 consecutive seconds in the vehicle attitude data), speed adjustment commands are generated, including: Curve parameter identification: Based on vehicle attitude data from 5 consecutive sampling periods (≥100Hz), using a formula... Calculate the radius of curvature of a curve ,in This refers to the wheelbase of a bicycle (1.1–1.3m). This represents the average roll angle. Distinguishing between inner and outer vehicles: using Cartesian coordinates The radial distance relative to the center of the curve is less than The vehicles marked as inside the curve bicycles, larger than The markings indicate the outer bicycle; Speed ​​adjustment range calculation: The speed increase of the inner bicycle ,in This represents the distance deviation between the inner vehicle and the center of the curve (actual radial distance - ideal radial distance). 0 indicates a bias towards the outer edge, requiring an increase in the rate of increase to correct the position. The current speed of the inner bicycle; The deceleration of the bicycle on the outside ,in This represents the distance deviation between the outer vehicle and the center of the curve (actual radial distance - ideal radial distance). (This indicates a deviation towards the outside, requiring increased deceleration to correct the position.) The current speed of the outer bicycle; Lateral spacing closed-loop control: Real-time monitoring of the lateral spacing between adjacent bicycles ,when hour( (For the safe lateral spacing threshold of curves), dynamically increased. and The absolute value, until .

[0092] In this embodiment of the invention, it is necessary to further explain that this embodiment determines whether the convoy has entered a curve by using the roll angle θ (which reflects the degree of left and right tilt of the vehicle body) in the vehicle body posture data: When driving normally in a straight line, the roll angle θ of the vehicle body is usually <3°, while in a curve, the rider will actively tilt the vehicle body to balance the centrifugal force. The value of θ increases with the curvature of the curve (such as θ can reach 15°~20° in a sharp curve). Therefore, continuous exceeding of θ can be regarded as the core physical characteristic of driving in a curve.

[0093] This embodiment calculates the curvature radius R of the curve in real time using vehicle posture data, providing a geometric reference for speed adjustment: when the vehicle is tilted, the ratio of the wheelbase to the sine of the roll angle is approximately equal to the curvature radius of the curve trajectory (based on the balance mechanics model of the "bicycle-rider" system, the larger the tilt angle, the smaller the corresponding turning radius).

[0094] The inner and outer vehicles are distinguished by their relative positions to the center of the curve using the vehicle's Cartesian coordinates (X, Y coordinates after UTM projection transformation). Determining the center of the curve: Based on the curve fitting of the driving trajectory of the lead vehicle in the convoy (generated from continuous position coordinates), calculate the coordinates of the center of the curve segment.

[0095] Radial distance determination: Inner vehicle: The radial distance between the vehicle and the center of the curve < (Closer to the inside of the curve, smaller turning radius); Outer vehicle: Radial distance between the vehicle and the center of the curve > (Located near the outside of the curve, with a larger turning radius); Vehicles in the middle area: radial distance in ~ For now, do not adjust the speed and maintain the current state.

[0096] Based on the distance deviation between the inner and outer vehicles and the center of the curve, the target speed adjustment range is calculated to achieve coordinated control of "inner vehicles accelerating and outer vehicles decelerating": The principle of speed increase for inner vehicles: Inner vehicles need to compensate for their smaller turning radius by increasing speed to avoid being "squeezed" into lateral space by outer vehicles; the greater the distance deviation, the higher the speed increase, so as to quickly return to the ideal trajectory.

[0097] The principle of deceleration for vehicles on the outside: Vehicles on the outside need to reduce centrifugal force by decelerating to avoid sideslip due to large turning radius and high speed; the greater the distance deviation, the greater the deceleration amplitude, so as to reduce the risk of lateral drift.

[0098] This embodiment monitors the lateral distance between adjacent vehicles in real time and dynamically adjusts the speed to ensure that the distance remains stable above a safe threshold. Spacing monitoring: Calculates the lateral distance between adjacent bicycles using the vehicle's planar coordinates. (The absolute value of the difference in Y-axis coordinates).

[0099] Closed-loop trigger condition: when At that time, the system determined that the distance was too close.

[0100] Dynamic correction strategy: Dynamically increase the absolute value of the sum (e.g., increase by 20%~50%) until... This forms a closed-loop control system of "monitoring-adjustment-feedback" to avoid the risk of continuous approaching collisions.

[0101] Example 2: The present invention provides an intelligent spacing collaborative control system for bicycle fleets, comprising: an on-board sensing module, a wireless communication module, a central processing unit, a multimodal prompting module, and a collaborative speed adjustment execution module.

[0102] The vehicle-mounted sensing module is installed on the frame and wheels of each bicycle and includes a GPS locator and a three-axis accelerometer. It is used to collect multi-dimensional operating parameters of each bicycle in real time, including real-time speed information, real-time position coordinates and vehicle posture data.

[0103] The wireless communication module is integrated into the handlebar controller of each bicycle, supporting Bluetooth Low Energy or LoRa networking communication, and is used to realize bidirectional data transmission between bicycles in the team and the central processing unit within a range of 50 meters, with a transmission latency of ≤100ms.

[0104] The central processing unit is deployed in the lead bicycle of the team and has built-in dynamic safety distance algorithm module and collaborative speed adjustment algorithm module. It is used to receive multi-dimensional operating parameters of the team, calculate the dynamic safety distance threshold and actual distance between adjacent bicycles using the preset dynamic safety distance algorithm, determine the collision risk level, and generate team-level speed adjustment command based on the real-time speed of each bicycle, the collision risk level and the overall driving path of the team through the preset collaborative speed adjustment algorithm, and send it to the collaborative speed adjustment execution module.

[0105] The multimodal warning module includes an LED warning light, a buzzer, and a seat vibration motor mounted on the handlebars. It triggers the corresponding warning mode according to the warning command from the central processing unit. When the risk is determined to be medium, the LED warning light will flash to issue a warning. When the risk is determined to be high, the buzzer will sound an alarm, and the seat vibration motor will provide a vibration warning.

[0106] The coordinated speed control module works in conjunction with the bicycle's electronic shifting system and / or braking system. After receiving the team-level speed control command from the central processing unit, it automatically adjusts the bicycle speed by controlling the gear ratio and / or braking stroke, with an adjustment response time of ≤500ms.

[0107] In one possible embodiment, a triaxial accelerometer is positioned in the middle of the bicycle frame seat tube to collect the linear velocity of the wheels in real time.

[0108] In one possible embodiment, the GPS locator is used to collect raw location coordinate data, including longitude, latitude, and elevation information; The three-axis accelerometer is used to collect data on the vehicle's motion trend, the rate of change of heading angle, and raw three-dimensional acceleration data. Its measurement axis is aligned with the vehicle's coordinate system. The X-axis points forward along the longitudinal direction of the vehicle body, the Y-axis points to the left along the transverse direction of the vehicle body, and the Z-axis points upward perpendicular to the ground. The sampling frequency is ≥100Hz and the resolution is ≥16 bits.

[0109] In one possible embodiment, the vehicle-mounted sensing module further includes a second-order Butterworth low-pass filter with a cutoff frequency of 5Hz, used to smooth the roll and pitch angle data and eliminate high-frequency vibration noise.

[0110] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent spacing coordination control of a bicycle fleet, characterized in that, Includes the following steps: S1. Collect real-time speed information, real-time position coordinates, and vehicle posture data of each bicycle through the on-board sensor module; S2. Each bicycle uploads its own operating parameters and shares the operating parameters of other bicycles in the team through the wireless communication module; S3. Based on real-time speed information, real-time position coordinates, and vehicle posture data, calculate the dynamic safe distance threshold between adjacent bicycles using a dynamic safe distance algorithm; S4. Calculate the actual distance between adjacent bicycles in real time, compare it with the dynamic safe distance threshold, and determine the collision risk level, including: low risk, medium risk, and high risk; S5. In cases of medium risk, a flashing light will be used as a warning; in cases of high risk, a strong warning will be issued by an audible alarm and seat vibration. S6. Based on real-time speed, collision risk level, and convoy driving path, a convoy-level speed adjustment command is generated through a collaborative speed adjustment algorithm to automatically adjust the bicycle speed.

2. The method according to claim 1, characterized in that, Real-time speed information collection includes: The linear velocity of the wheel is collected by a three-axis accelerometer, and combined with GPS positioning speed information. The real-time speed information is obtained after data fusion through dynamic weight allocation and Kalman filtering.

3. The method according to claim 2, characterized in that, Data fusion includes: The GPS positioning speed information and linear velocity are synchronized in time and data interpolated and adapted. The fusion weights are dynamically allocated according to the GPS signal strength, and noise is eliminated by Kalman filtering. The fused real-time speed information is then output. Specifically, when the GPS signal strength is less than or equal to a preset threshold, the fusion weight of the positioning speed information is reduced to 0.2-0.4; when the GPS signal strength is greater than the preset threshold, the fusion weight of the positioning speed information is increased to 0.5-0.

7.

4. The method according to claim 1, characterized in that, Real-time location coordinate acquisition includes: Raw location coordinate data is collected using a GPS locator, combined with motion trend data and heading angle change rate from a three-axis accelerometer, and after outlier removal and timestamp alignment, real-time location coordinates are generated using a multi-source fusion algorithm. When the GPS signal strength is greater than or equal to the threshold, the motion trend data is fused using the preprocessed original position coordinate data as a reference to correct the positioning drift by using the Kalman filter algorithm to fuse the motion trend data. When the GPS signal strength is less than the threshold, the position coordinates are predicted in real time by extended Kalman filtering based on the effective position coordinates, linear velocity and heading angle change rate of the previous moment. The confidence weight of the estimated position decreases linearly with the duration of GPS signal interruption.

5. The method according to claim 1, characterized in that, Vehicle posture data collection includes: Raw three-dimensional acceleration data is collected by a triaxial accelerometer. After outlier removal and zero drift calibration, the roll angle and pitch angle of the vehicle body are calculated based on the acceleration components of the X, Y, and Z axes. The roll angle and pitch angle data are smoothed by a second-order Butterworth low-pass filter with a cutoff frequency of 5Hz before being output and timestamped with real-time speed information and real-time position coordinates.

6. The method according to any one of claims 1, 2, 4, and 5, characterized in that, Dynamic safety distance threshold calculation includes: The longitudinal basic safety distance is calculated based on the real-time speed information of the vehicle in front, the preset driver reaction time, and the maximum braking deceleration of the bicycle. The maximum braking deceleration of the bicycle is dynamically corrected with the pitch angle. The actual lateral offset is calculated based on the real-time lateral difference between the rear vehicle and the front vehicle, the preset bicycle wheelbase parameters, and the roll angle. The lateral safety distance correction term is calculated based on the actual lateral offset and real-time speed information, combined with the lateral safety factor. The dynamic safety distance threshold is obtained by vector synthesis of the longitudinal basic safety distance and the lateral safety distance correction term; When the convoy enters a curve, the lateral safety factor automatically increases to 1.6-1.

8.

7. The method according to claim 6, characterized in that, Actual spacing calculation includes: Based on the real-time location coordinates and driving direction of each bicycle in the convoy, the path planning algorithm is used to dynamically identify adjacent bicycle pairs, and the latitude and longitude coordinates are converted into plane rectangular coordinates by UTM projection based on the curvature radius of the trochanter circle, the curvature radius of the meridian circle, and the longitude of the central meridian. The projection parameters are dynamically matched based on the current geographical location of the convoy. Based on Cartesian coordinates, the actual distance is calculated using the Euclidean distance formula. When the elevation difference exceeds a threshold, an elevation correction term is introduced for three-dimensional correction, and the result is output after sliding window averaging filtering.

8. The method according to claim 7, characterized in that, Team-level speed control command generation includes: When the vehicle ahead poses a high risk, a deceleration command is generated, including: Based on the real-time speed of the preceding vehicle and the dynamic safe distance threshold, calculate the number of following vehicles that need to slow down to ensure that at least two following bicycles are covered. For each vehicle that needs to decelerate, a distance weight is calculated, where the distance weight increases as the actual distance between the following vehicle and the preceding vehicle decreases. The target deceleration magnitude for each following vehicle that needs to decelerate is generated based on its current speed, risk level coefficient, and distance weight. When a following vehicle receives deceleration commands from multiple preceding vehicles simultaneously, the command from the closest preceding vehicle is prioritized and time-slice rotation is performed.

9. The method according to claim 1 or 5, characterized in that, The generation of fleet-level speed control commands also includes: When the entire convoy enters the curve, the speed adjustment commands generated include: Calculate the radius of curvature of the curve based on the bicycle wheelbase and average roll angle. ; The radial distance relative to the center of the curve in the Cartesian coordinate system is less than The vehicles marked as inside the curve bicycles, larger than The markings indicate the outer bicycle; Calculate the speed increase of the inner bicycle based on the distance deviation between the inner vehicle and the center of the curve and the current speed of the inner bicycle. Calculate the deceleration magnitude of the outer bicycle based on the distance deviation between the outer vehicle and the center of the curve and the current speed of the outer bicycle; By dynamically increasing the speed increase of the inner bicycle and the speed decrease of the outer bicycle, the lateral distance between adjacent bicycles is made greater than or equal to the safe lateral distance threshold for curves.

10. A smart spacing collaborative control system for a bicycle fleet, characterized in that, include: Vehicle-mounted sensor module: used to collect real-time speed information, real-time position coordinates, and vehicle attitude data of each bicycle; Wireless communication module: used to upload its own operating parameters and share the operating parameters of other vehicles in the fleet; Central processing unit: Based on real-time speed information, real-time position coordinates and vehicle posture data, it calculates the dynamic safe distance threshold between adjacent bicycles through a dynamic safe distance algorithm; The system calculates the actual distance between adjacent bicycles in real time and compares it with a dynamic safe distance threshold to determine the collision risk level, including low risk, medium risk, and high risk. Based on real-time speed, collision risk level, and convoy driving path, a convoy-level speed adjustment command is generated through a collaborative speed adjustment algorithm. Multimodal alert module: used to issue a warning with flashing lights when there is a medium risk, and to issue a strong warning with sound alarm and seat vibration when there is a high risk; Collaborative speed control execution module: Used to receive team-level speed control commands and automatically adjust bicycle speed.

Images