Pedestrian zebra crossing traffic safety signal and control system

Abstract

The invention relates to the technical field of safe traffic, and provides a pedestrian crossing zebra crossing traffic safety signal and control system, which comprises a data acquisition module used for acquiring pedestrian states, identifying special crowds and acquiring vehicle types, vehicle speeds and weight data; the behavior analysis module is connected with the data acquisition module through a data transmission technology, and is used for predicting the behavior trend of the pedestrian and predicting the vehicle passing condition; and the data analysis module is connected with the behavior analysis module through a data transmission technology, and is used for analyzing whether the pedestrians are allowed to pass through the zebra crossing or not through the prediction data and estimating the duration of the pedestrians passing through the zebra crossing. By capturing the change of the walking state of the pedestrian in real time, the pedestrian is effectively prevented from staying on the road due to unreasonable time setting of the signal lamp according to factors such as a possible speed recovery condition (PSRS) and an impossible speed recovery condition (ISRS) when an accident occurs when the pedestrian passes through the zebra crossing, and the safety and smoothness of the pedestrian crossing the street are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of safe traffic, and particularly to a traffic safety signal and control system for pedestrian crosswalks. Background Art

[0002] In urban traffic, the safety of pedestrians crossing the road and traffic fluency are important issues. The traditional control method of zebra crossing traffic lights usually switches at fixed time intervals. Regardless of how the number of pedestrians and vehicle flow at the intersection change, the switching time of the traffic lights is fixed. In the case of low pedestrian flow and high vehicle flow, this method will cause vehicles to wait for a long time. And in sections with low pedestrian and vehicle flow, the alertness of both pedestrians and vehicles is relatively low, resulting in a higher possibility of safety accidents when pedestrians cross the zebra crossing. For special pedestrians with inconvenient mobility such as the blind, wheelchair users, and those with crutches, the fixed signal switching mode is difficult to ensure their safe crossing of the road. If the green light time is too short, special pedestrians may not be able to reach the other side of the road in time, while if the green light time is too long, it will unnecessarily affect the vehicle passing efficiency. Therefore, there is an urgent need for a system that can detect the status of special pedestrians and intelligently control traffic lights accordingly to improve the safety of special pedestrians crossing the road and the overall traffic fluency. Summary of the Invention

[0003] Aiming at the problems existing in the prior art, the purpose of the present invention is to provide a traffic safety signal and control system for pedestrian crosswalks to solve the problems raised in the above background art.

[0004] To achieve the above purpose, the present invention provides a traffic safety signal and control system for pedestrian crosswalks, including: A data acquisition module, used to acquire pedestrian status, identify special groups, and acquire vehicle type, vehicle speed, and weight data; A behavior analysis module, connected to the data acquisition module through data transmission technology, used to predict the behavior tendency of pedestrians and predict vehicle passing situations; A data analysis module, connected to the behavior analysis module through data transmission technology, used to analyze whether to allow pedestrians to cross the zebra crossing based on the predicted data and estimate the duration for pedestrians to cross the zebra crossing; An early warning trigger module, connected to the data analysis module through data transmission technology, used to trigger an early warning signal when it is determined that pedestrians are not allowed to pass; A dynamic update module, connected to the data analysis module through data transmission technology, used to real-time update the predicted passing time and passing situation of pedestrians when pedestrians cross the zebra crossing; A signal control module, connected to the data analysis module and the early warning trigger module through data transmission technology, used to control the display of traffic lights, emit a sound and light alarm to pedestrians after triggering an early warning message, and real-time update the display situation of traffic lights.

[0005] Preferably, the steps for the data acquisition module to obtain the multi-modal pedestrian status are as follows: S111: Generate millimeter wave signals at a fixed frequency, calculate the reflected wave intensity according to the device-set transmission power, and the formula is: , where in the formula, is the reflected wave intensity, is the device-set transmission power, and are the transmitting and receiving antenna gains respectively, is the wavelength, and , is the speed of light, is the fixed frequency, is the target radar cross section, is the target distance; S112: Utilize the Doppler effect to calculate the pedestrian speed according to the frequency difference between the transmitted wave and the reflected wave, and the formula is: , where in the formula, is the current speed of the pedestrian, is the wavelength, is the fixed frequency difference; S113: Calculate the distance between the pedestrian and the radar by measuring the time delay between the transmitted signal and the received reflected signal, and the formula is: , where in the formula, is the target distance, that is, the distance between the pedestrian and the radar, is the speed of light, is the time delay; S114: Based on the target distance and pedestrian speed at different time points, use the Kalman filtering algorithm to predict the next position of the pedestrian, update the position by combining the newly measured data, and depict the pedestrian movement trajectory, and the formula is: , where in the formula, is the next position of the pedestrian, is the previous position of the pedestrian, is the state transition matrix, is the control matrix, is the control variable.

[0006] Preferably, the steps for the data acquisition module to identify special groups are as follows: S121: Continuously collect images at a fixed frame rate, compress the image data in the JPGE format, and transmit it to the special pedestrian recognition unit and the status analysis unit in real time through Ethernet or WI-FI; S122. Generate a depth image by combining the time-of-flight method with the calculated target distance data, and transmit the depth information in binary format to the special pedestrian recognition unit and the status analysis unit; S123. Use the Gaussian filtering algorithm to reduce image noise and update each new pixel value in the image. The formula is: , where is the new pixel value of the image after Gaussian filtering and noise reduction at the coordinate position, is the pixel value at the coordinate position in the original image, is the standard deviation of the Gaussian kernel; S124. Use histogram equalization to enhance the image contrast. The formula is: , where is the grayscale histogram of the image, and are both grayscale values, with the value range , is the cumulative distribution function, is the new pixel grayscale value of the image at the coordinate position after histogram equalization processing, is the pixel grayscale value of the original image at the coordinate position; S125. Through convolutional pooling operations, perform dot product accumulation calculations by sliding the convolutional kernel on the input feature map, and output the element at the coordinate position in the feature map. The formula is: , where is the element value at the coordinate and channel in the output feature map , and are the coordinate indices of the output feature map in the height and width directions respectively, is the size of the convolutional kernel in the high dimension, and are the indices in the high dimension and width dimension of the convolutional kernel respectively, is the channel index of the input feature map, is the coordinate in the convolutional kernel, the input channel is and the output channel is of the weight value, is the bias value of the weight channel , is the stride of the convolutional operation; S126. After multiple rounds of convolution and pooling, the feature map output by the last pooling layer is flattened to convert the multi-dimensional feature map into a one-dimensional vector. The formula is: , where is the one-dimensional image feature vector obtained after the flattening operation, , and are the height, width, and number of channels of the feature map output by the last convolutional pooling layer, respectively; S127. The extracted feature vector is input into a value classifier to calculate the scores for different pedestrian types, including but not limited to ordinary pedestrians, blind people, wheelchair users, cane users, and pedestrians with abnormal walking states. The formula is: , where is the score of the th class output by the classifier, is the total number of classes, is the class index.

[0007] Preferably, the steps for the data acquisition module to obtain vehicle driving state data are as follows: S131. Multiple geomagnetic sensors buried at a depth of 5 - 10 cm below the lane in a preset induction area of the zebra crossing are used to detect the voltage signals generated when the vehicle passes by, record the rising edge and falling edge times, and calculate the vehicle speed according to the time difference between the vehicle passing adjacent geomagnetic sensors and the preset length between adjacent geomagnetic sensors. The formula is: , where is the vehicle speed, is the preset length between adjacent geomagnetic sensors, is the time difference between the vehicle passing adjacent geomagnetic sensors; S132. Estimate the vehicle length according to the duration of the induction signal and the speed. The formula is: , where is the estimated vehicle length, is the time difference between the front and rear wheels of the vehicle passing this geomagnetic sensor; S133. Collect vehicle image information and use the same feature extraction method as the pedestrian feature extraction method to extract and compare the features of the detected vehicle image to identify the specific vehicle model; S134. Obtain the voltage signal when the vehicle passes through the weighing sensor set between adjacent geomagnetic sensors, and calculate the vehicle weight through a calibration method. The formula is: , where is the vehicle weight, is the calibration coefficient, is the voltage signal, is the offset.

[0008] Preferably, the behavior analysis module uses a weighted fusion algorithm to obtain the fusion speed from the speed data measured by the millimeter-wave radar and the speed estimation value obtained by combining the depth sensor with visual analysis. The formula is , where is the fusion speed, and are the speed measured by the millimeter-wave radar and the speed estimation value obtained by combining the depth sensor with visual analysis, respectively, and are the corresponding weights, and , and a speed threshold is set. Where indicates that the pedestrian is in a stationary state, indicates a slow walking state, indicates a normal walking state, indicates a fast walking state. By comparing the ratio of with to the set threshold speed, the walking state of the pedestrian is judged.

[0009] Preferably, the behavior analysis module calculates the braking distance of the vehicle according to the vehicle weight, driving speed and vehicle type. The formula is: , where is the braking distance, is the braking coefficient, is the acceleration due to gravity.

[0010] Preferably, the data analysis module judges and calculates the duration for pedestrians to pass through the traffic lights, including the following steps: S31. Compare the actual distance of the vehicle from the zebra crossing with the calculated braking distance to judge whether it is safe for pedestrians to cross the zebra crossing; S32. When it is judged that the leading vehicle in the traffic flow can still stop in front of the zebra crossing after adding the margin time and the yellow light time as the reaction time, according to the kinematic formula , where is the final speed of the leading vehicle in the current traffic flow, that is , is the current speed, is the acceleration, is the time, is the displacement distance, that is when the red light of the lane lights up. Where is the time required for the leading vehicle in the traffic flow to brake, is a margin time, and has a minimum value of , is the yellow light duration of the lane, and the value range is ; S33. According to the number of pedestrians in the queue, except for the first pedestrian, a safe interval is maintained between each subsequent pedestrian and the previous one, and the required time for all types of people to cross the zebra crossing is calculated. The formula is: , where in the formula, is the time required for the th person to cross the zebra crossing, is the total length of the pedestrian queue, is the length of the zebra crossing, is the average speed of the slowest person in the pedestrian queue crossing the zebra crossing, is the safe interval distance between pedestrians; S34. Add another margin time to the time required for the th person to cross the zebra crossing, and calculate the green light time of the zebra crossing. The formula is: , where in the formula, is the green light time of the zebra crossing, is the time required for the first vehicle in the traffic flow to brake, is another margin time, and has a minimum value of . At the same time, add a yellow light time for the zebra crossing, and calculate the red light time of the lane. The formula is , where in the formula, is the remaining red light time of the lane, is the yellow light time of the zebra crossing, and the value range is .

[0011] Preferably, in step S31, when it is determined that it is not safe for pedestrians to cross the zebra crossing, a signal is sent to the warning trigger module. First, an audible and visual alarm signal is triggered, and then the green light countdown for vehicles and the red light countdown for pedestrians are calculated. The generation of the countdown includes the following steps: S41. Calculate the vehicles in the current traffic flow that can safely stop at the front end of the zebra crossing, and calculate the vehicles in the traffic flow that can safely stop with a speed of 0 according to the kinematic formula. The formula is , where in the formula, is the speed of the vehicle at the moment when braking starts, is the time experienced by the vehicle from the start of braking to safe parking, that is, the braking duration, is the displacement traveled by the vehicle from the start of braking to safe parking, that is, the braking distance; S42. Add a safety margin time to the braking time to ensure that the vehicle has enough time to react and stop safely. The formula is: , where in the formula, is the remaining green light duration of the lane, is another margin duration, and has a minimum value of ; S43. Calculate the red light duration of the zebra crossing based on the remaining green light duration of the lane. The formula is: , where in the formula, is the red light duration of the zebra crossing, is the yellow light duration of the lane, and the value range is , is another margin duration, and has a minimum value of .

[0012] Preferably, the steps for the dynamic update module to update the passing time of pedestrians on the zebra crossing are as follows: S51. Real-time sense the pressure change through the pressure sensors deployed on the zebra crossing, record the landing points and movement information of pedestrians' footsteps, and calculate the speed change of pedestrians on the zebra crossing in combination with image acquisition. The formula is: , where in the formula, and are the walking speeds of pedestrians at adjacent times respectively, is the speed change value of pedestrians' walking, is the time interval; S52. When the pedestrian speed appears , , or the pedestrian posture shows obvious imbalance, pause and other situations that do not conform to the normal model through image analysis, and the pressure sensor data shows chaotic foot movement, etc., it is determined that the walking state is abnormal. In the formula, and are the average speed and acceleration of the pedestrian on the zebra crossing respectively, , , and are the set minimum, maximum average speed and acceleration of pedestrians on the zebra crossing respectively; S53. Calculate the remaining distance of the pedestrian to the end point. The formula is: , where in the formula, is the remaining distance of the pedestrian to the end point, is the total length of the zebra crossing, is the distance that the pedestrian has walked; S54. Calculate the time for the pedestrian to pass the remaining distance of the zebra crossing. The formula is , where in the formula, is the predicted time for the pedestrian to pass the remaining distance of the zebra crossing, is the recovery coefficient, is the possible acceleration recovery situation of the pedestrian, is the impossible acceleration recovery situation of the pedestrian; S55. Update the green light duration of the zebra crossing and the red light duration of the lane. The calculation formula is , where in the formula, is the updated green light duration of the zebra crossing, is the duration for which the green light of the zebra crossing has been on, is the updated red light duration of the lane, is the duration for which the red light of the lane has been on.

[0013] Preferably, the signal control module controls the signal lamp through the following steps: S61. Receive instructions through wired or wireless communication, and verify the format and data integrity; S62. Convert the digital instructions into analog control signals suitable for driving the circuit; S63. Drive the signal lamp driving circuit to control the on / off switching of the traffic lights and the start / stop of the audible and visual alarm; S64. Feed back the working state of the signal lamp to the data analysis module for data storage and update.

[0014] A traffic safety signal and control system for a pedestrian crosswalk provided by the present invention has the following beneficial effects: 1. Through the multi-modal sensor fusion technology, the walking state of pedestrians is monitored comprehensively and in real time. By accurately controlling the pedestrian crossing time and intelligently regulating the signal lamp, the efficient coordination between pedestrians and vehicles is achieved. While ensuring the safety of pedestrians crossing the street, the unnecessary waiting time of vehicles is reduced, the traffic flow at the intersection is optimized. Compared with the traditional fixed-duration signal lamp system, it can greatly improve the alertness of pedestrians and drivers passing through the intersection at intersections with few pedestrians and vehicles, which is beneficial to improving the safety of road traffic.

[0015] 2. By accurately identifying special pedestrians and their states, sufficient green light time and voice guidance are provided for them. And by obtaining information such as vehicle type, length, speed and weight, the braking distance that the vehicle can achieve is analyzed, which can greatly optimize the road traffic effect. At the same time, it can effectively prevent special pedestrians from being in danger due to unreasonable signal lamp time when crossing the road. While ensuring the safety of special pedestrians, combined with vehicle detection information, the signal lamp time is reasonably controlled to reduce the unnecessary waiting time of vehicles.

[0016] 3. The dynamic update module endows it with powerful adaptability. It can not only capture the changes in the walking state of pedestrians in real time, but also, based on factors such as the possible speed recovery situation (PSRS) and impossible speed recovery situation (ISRS) of pedestrians in case of accidents during crossing the zebra crossing, use advanced algorithms to accurately predict the remaining crossing time of pedestrians. This dynamic and intelligent adjustment mechanism enables the signal light duration to always match the actual crossing needs of pedestrians. When detecting abnormal walking states of pedestrians, it can quickly and accurately judge, based on a scientific calculation model, accurately update the pedestrian crossing time, and timely feedback it to the signal light control system and voice prompt system, effectively avoiding pedestrians staying in the middle of the road due to unreasonable signal light time settings, and significantly improving the safety and smoothness of pedestrian crossing. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0018] Figure 1 It is a schematic system flow diagram of a traffic safety signal and control system for a pedestrian crosswalk provided by this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The following will further describe in detail the specific embodiments of the present invention in conjunction with the drawings in the specification and the embodiments. The following embodiments are only used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0020] As Figure 1 shown, this embodiment proposes a traffic safety signal and control system for a pedestrian crosswalk, including: A data acquisition module, used to acquire pedestrian states, identify special groups, and acquire vehicle types, vehicle speeds, and weight data; A behavior analysis module, connected to the data acquisition module through data transmission technology, used to predict the behavior trends of pedestrians and predict vehicle passing situations; A data analysis module, connected to the behavior analysis module through data transmission technology, used to analyze whether to allow pedestrians to cross the zebra crossing and estimate the duration for pedestrians to cross the zebra crossing through predicted data analysis; An early warning trigger module, connected to the data analysis module through data transmission technology, used to trigger an early warning signal when it is judged that pedestrians are not allowed to pass; A dynamic update module, connected to the data analysis module through data transmission technology, used to update the predicted crossing time and passing situation of pedestrians in real time when pedestrians cross the zebra crossing; The signal control module is connected to the data analysis module and the warning trigger module through data transmission technology, and is used to control the display of the traffic signal lights, trigger warning information to send out sound and light alarms to pedestrians, and update the display of the traffic signal lights in real time.

[0021] Specifically, through the multi-modal sensor fusion technology, the walking state of pedestrians is monitored comprehensively and in real time. By accurately controlling the pedestrian crossing time and intelligently regulating the traffic signal lights, the efficient coordination of pedestrian and vehicle traffic is realized. While ensuring the safety of pedestrians crossing the street, it reduces the unnecessary waiting time of vehicles, optimizes the traffic flow at intersections. Compared with the traditional fixed-duration traffic signal system, it can greatly improve the alertness of pedestrians and drivers passing through intersections at intersections with few people and few vehicles, which is conducive to enhancing the safe passage of roads.

[0022] In this embodiment, the steps for the data acquisition module to obtain the multi-modal pedestrian state are as follows: S111: Generate millimeter-wave signals at a fixed frequency, calculate the reflected wave intensity according to the device-set transmission power, and the formula is: , where is the reflected wave intensity, is the transmission power set by the device, and are the transmitting and receiving antenna gains respectively, is the wavelength, and , is the speed of light, is the fixed frequency, is the target radar cross section, is the target distance; S112: Utilize the Doppler effect to calculate the pedestrian speed according to the frequency difference between the transmitted wave and the reflected wave, and the formula is: , where is the current speed of the pedestrian, is the wavelength, is the fixed frequency difference; S113: Calculate the distance between the pedestrian and the radar by measuring the time delay between the transmitted signal and the received reflected signal, and the formula is: , where is the target distance, that is, the distance between the pedestrian and the radar, is the speed of light, is the time delay; S114: Based on the target distance and pedestrian speed at different time points, use the Kalman filtering algorithm to predict the next position of the pedestrian, update the position by combining the newly measured data, and depict the pedestrian movement trajectory, and the formula is: , where is the next position of the pedestrian, is the position of the pedestrian at the previous moment, is the state transition matrix, is the control matrix, is the control variable.

[0023] In this embodiment, the data acquisition module identifies special populations through the following steps: S121: Continuously acquire images at a fixed frame rate, compress the image data in the JPGE format, and transmit it in real time to the special pedestrian recognition unit and the state analysis unit via Ethernet or WI-FI; S122: Combine with the calculated target distance data through the time-of-flight method to generate a depth image, and transmit the depth information in binary format to the special pedestrian recognition unit and the state analysis unit; S123: Use the Gaussian filtering algorithm to denoise the image and update each new pixel value in the image. The formula is: , where is the new pixel value of the image at the coordinate position after Gaussian filtering and denoising, is the pixel value at the coordinate position in the original image, is the standard deviation of the Gaussian kernel; S124: Use histogram equalization to enhance the image contrast. The formula is: , where is the grayscale histogram of the image, and are both grayscale values, with the value range , is the cumulative distribution function, is the new pixel grayscale value of the image at the coordinate position after histogram equalization processing, is the pixel grayscale value of the original image at the coordinate position; S125: Perform convolution pooling operations by sliding the convolution kernel on the input feature map for dot product and accumulation calculations, and output the element at the coordinate position in the feature map. The formula is: , where is the output feature map at the coordinate and the channel is element value, and are the coordinate indices of the output feature map in the height and width directions respectively, is the size of the convolution kernel in the high dimension, and are the indices in the height dimension and width dimension of the convolution kernel respectively, is the channel index of the input feature map, is the coordinate in the convolution kernel as , the input channel is and the output channel is the weight value, is the weight channel the bias value, is the stride of the convolution operation; S126. After multiple rounds of convolution and pooling, the feature map output by the last pooling layer is flattened, converting the multi-dimensional feature map into a one-dimensional vector. The formula is: , where in the formula, is the one-dimensional image feature vector obtained after the flattening operation, , and are the height, width and number of channels of the feature map output by the last convolution pooling layer respectively; S127. The extracted feature vector is input into the classifier to calculate the scores for different pedestrian types, including but not limited to ordinary pedestrians, blind people, wheelchair users, cane users, and pedestrians with abnormal walking states. The formula is: , where in the formula, is the score of the th class output by the classifier, is the total number of classes, is the class index.

[0024] In this embodiment, the data acquisition module acquires vehicle driving state data including the following steps: S131. Multiple geomagnetic sensors buried at a depth of 5 - 10 cm below the lane in the preset induction area of the zebra crossing are used to detect the voltage signals generated when the vehicle passes by, record the rising edge and falling edge times, and calculate the vehicle speed according to the time difference between the vehicle passing adjacent geomagnetic sensors and the preset length between adjacent geomagnetic sensors. The formula is: , where in the formula, is the vehicle speed, is the preset length between adjacent geomagnetic sensors, is the time difference between the vehicle passing adjacent geomagnetic sensors; S132. Estimate the vehicle length according to the induction signal duration and speed. The formula is: , where in the formula, is the estimated vehicle length, is the time difference between the vehicle's front and rear wheels passing the geomagnetic sensor; S133. Collect vehicle image information, and use the same feature extraction method as that for pedestrian feature extraction to extract and compare features of the detected vehicle images to identify the specific vehicle model; S134. Obtain the voltage signal when the vehicle passes through according to the weighing sensors set between adjacent geomagnetic sensors, and calculate the vehicle weight through a calibration method. The formula is: , where in the formula, is the vehicle weight, is the calibration coefficient, is the voltage signal, is the offset.

[0025] In this embodiment, the behavior analysis module uses the weighted fusion algorithm to obtain the fusion speed from the speed data measured by the millimeter-wave radar and the speed estimation value obtained by combining the depth sensor with visual analysis. The formula is , where in the formula, is the fusion speed, and are respectively the speed measured by the millimeter-wave radar and the speed estimation value obtained by combining the depth sensor with visual analysis, and are respectively the corresponding weights, and , and a speed threshold is set. Where in the formula, means the pedestrian is in a stationary state, means the slow walking state, means the normal walking state, means the fast walking state. By and the ratio of the set threshold speed, judge the walking state of the pedestrian.

[0026] In this embodiment, the data analysis module judges and calculates the duration for pedestrians to pass through the traffic lights, including the following steps: , where in the formula, is the braking distance, is the braking coefficient, is the acceleration due to gravity.

[0027] In this embodiment, the data analysis module judges and calculates the duration for pedestrians to pass through the traffic lights, including the following steps: S31. Compare the actual distance of the vehicle from the zebra crossing with the calculated braking distance to judge whether it is safe for pedestrians to pass through the zebra crossing; S32. When it is judged that the leading vehicle in the traffic flow can still stop in front of the zebra crossing when adding the margin time and the yellow light time as the reaction time, according to the kinematic formula , where in the formula, is the final speed of the leading vehicle in the current traffic flow, that is, , is the current speed, is the acceleration, is the time, is the displacement distance, that is, when the red light of the lane lights up, where, is the time required for the first vehicle in the traffic flow to brake, is a margin time, and the minimum value is , is the yellow light duration of the lane, and the value range is ; S33. According to the number of pedestrians in the queue, except for the first pedestrian, a safe interval is maintained between each subsequent pedestrian and the previous pedestrian, and the time required for all types of people to pass the zebra crossing is calculated. The formula is: , where, is the time required for the th person to pass the zebra crossing, is the total length of the pedestrian queue, is the length of the zebra crossing, is the average speed of the slowest person passing the zebra crossing in the pedestrian queue, is the safe interval distance between pedestrians; S34. Add another margin time to the time required for the th person to pass the zebra crossing, and calculate the green light time of the zebra crossing. The formula is: , where, is the green light time of the zebra crossing, is the time required for the first vehicle in the traffic flow to brake, is another margin time, and the minimum value is , and at the same time add a yellow light time of the zebra crossing to find the red light time of the lane. The formula is , where, is the remaining red light time of the lane, is the yellow light time of the zebra crossing, and the value range is .

[0028] In this embodiment, in step S31, when it is determined that it is unsafe for pedestrians to pass the zebra crossing, a signal is sent to the warning trigger module, first triggering an audible and visual alarm signal, and then calculating the green light countdown of the vehicle and the red light countdown of the pedestrian. The countdown generation includes the following steps: S41. Calculate the vehicles in the current traffic flow that can safely stop at the front end of the zebra crossing, and calculate the vehicles in the traffic flow that can safely stop with a speed of 0 according to the kinematic formula. The formula is , where, is the speed of the vehicle at the moment when braking starts, is the time it takes for the vehicle to come to a safe stop from the start of braking, i.e., the braking duration. The displacement traveled by the vehicle during the process from the start of braking to a safe stop, i.e., the braking distance. S42. Add a safety margin time to the braking time to ensure that the vehicle has enough time to react and stop safely. The formula is: , where in the formula, is the remaining green light duration of the lane. is another margin duration, and the minimum value is . S43. Calculate the red light duration of the zebra crossing according to the remaining green light duration of the lane. The formula is: , where in the formula, is the red light duration of the zebra crossing. is the yellow light duration of the lane, and the value range is . is another margin duration, and the minimum value is .

[0029] Specifically, by accurately identifying special pedestrians and their states, providing them with sufficient green light time and voice guidance, and by obtaining information on vehicle type, length, speed, and weight, analyzing the braking distance that the vehicle can achieve, it can greatly optimize the road traffic effect. At the same time, it can effectively prevent special pedestrians from being in danger due to unreasonable signal timing when crossing the road, realizing the reasonable control of signal timing while ensuring the safety of special pedestrians, and reducing the unnecessary waiting time of vehicles by combining vehicle detection information.

[0030] In this embodiment, the steps for the dynamic update module to update the pedestrian passing time on the zebra crossing are as follows: S51. Real-time sense the pressure change through the pressure sensors deployed on the zebra crossing, record the foot landing points and movement information of pedestrians, and calculate the speed change of pedestrians on the zebra crossing in combination with image acquisition. The formula is: , where in the formula, and are the walking speeds of pedestrians at adjacent moments respectively. is the speed change value of pedestrians walking. is the time interval. S52. When the pedestrian speed shows , , or when the pedestrian posture shows obvious imbalance, pause and other situations that do not conform to the normal model through image analysis, and the pressure sensor data shows chaotic foot movement, etc., it is determined that the walking state is abnormal. Where in the formula, and are the average speed and acceleration of the pedestrian on the zebra crossing at present respectively. , , and are respectively the minimum, maximum average speeds and accelerations of the set pedestrians on the zebra crossing; S53. Calculate the remaining distance of the pedestrian to the end point, and the formula is: , where is the remaining distance of the pedestrian to the end point, is the total length of the zebra crossing, is the distance the pedestrian has walked; S54. Calculate the time for the pedestrian to pass the remaining distance of the zebra crossing, and the formula is , where is the predicted time for the pedestrian to pass the remaining distance of the zebra crossing, is the recovery coefficient, is the possible acceleration recovery situation of the pedestrian, is the impossible acceleration recovery situation of the pedestrian; S55. Update the green light duration of the zebra crossing and the red light duration of the lane, and the calculation formula is , where is the updated green light duration of the zebra crossing, is the duration for which the green light of the zebra crossing has been on, is the updated red light duration of the lane, is the duration for which the red light of the lane has been on.

[0031] In this embodiment, the signal control module controls the signal lamp including the following steps: S61. Receive instructions through wired or wireless communication, and verify the format and data integrity; S62. Convert the digital instructions into analog control signals suitable for driving the circuit; S63. Drive the signal lamp driving circuit to control the on / off switching of the red and green lights and the start / stop of the audible and visual alarm; S64. Feed back the working state of the signal lamp to the data analysis module for data storage and update.

[0032] Specifically, the dynamic update module has a powerful adaptive ability. It can not only capture the changes in the walking state of pedestrians in real time, but also, based on factors such as the possible speed recovery situation (PSRS) and impossible speed recovery situation (ISRS) when pedestrians encounter unexpected situations during crossing the zebra crossing, use advanced algorithms to accurately predict the remaining passing time of pedestrians. This dynamic and intelligent adjustment mechanism enables the signal light duration to always match the actual crossing needs of pedestrians. When detecting abnormal walking states of pedestrians, it can quickly and accurately judge, based on a scientific calculation model, accurately update the passing time of pedestrians, and promptly feedback it to the signal light control system and voice prompt system, effectively avoiding pedestrians being stranded in the middle of the road due to unreasonable signal light time settings, and significantly improving the safety and smoothness of pedestrians crossing the street.

[0033] The above embodiments are only used to illustrate the present invention, rather than to limit the present invention. Although the present invention has been described in detail with reference to the embodiments, those of ordinary skill in the art should understand that various combinations, modifications, or equivalent replacements of the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and should all be covered within the scope of the claims of the present invention.

Claims

1. A traffic safety signal and control system for pedestrian crosswalks, characterized in that, Including: A data acquisition module, which is used to acquire pedestrian status, identify special groups, and acquire vehicle type, vehicle speed, and weight data; The steps for the data acquisition module to acquire multi-modal pedestrian status are as follows: S111: Generate millimeter-wave signals at a fixed frequency, calculate the reflected wave intensity according to the device-set transmission power; S112: Utilize the Doppler effect to calculate the pedestrian speed based on the frequency difference between the transmitted wave and the reflected wave; S113: Calculate the distance between the pedestrian and the radar by measuring the time delay between the transmitted signal and the received reflected signal; S114: Based on the target distance and pedestrian speed at different time points, use the Kalman filtering algorithm to predict the next position of the pedestrian, update the position by combining the newly measured data, and depict the pedestrian's movement trajectory, including the following steps: S121: Continuously acquire images at a fixed frame rate, compress the image data in the JPGE format, and transmit it to the special pedestrian recognition unit and the status analysis unit in real time through Ethernet or WI-FI; S122: Generate a depth image by combining the time-of-flight method with the calculated target distance data, and transmit the depth information to the special pedestrian recognition unit and the status analysis unit in binary format; S123: Use the Gaussian filtering algorithm to reduce noise in the image, update each new pixel value in the image, and enhance the image contrast by applying histogram equalization; S124. Through convolution pooling operation, perform dot product accumulation calculation by sliding a convolution kernel on the input feature map, and output the element at the coordinate position in the feature map; S125: After multiple rounds of convolution and pooling, flatten the feature map output by the last pooling layer to convert the multi-dimensional feature map into a one-dimensional vector; S126: Input the extracted feature vector into a classifier to calculate the pedestrian types, and the types include but are not limited to ordinary pedestrians, blind people, wheelchair users, cane users, and pedestrians with abnormal walking states, including the following steps: S131: Detect the voltage signals generated when the vehicle passes through by multiple geomagnetic sensors buried at a depth of 5-10 cm under the lane within the preset induction area of the zebra crossing, record the rising edge and falling edge times, and calculate the vehicle speed according to the time difference between the vehicle passing through adjacent geomagnetic sensors and the preset length between adjacent geomagnetic sensors; S132: Estimate the vehicle length according to the induction signal duration and speed, collect vehicle image information, and use the same feature extraction method as the pedestrian feature extraction method to extract and compare the features of the detected vehicle image to identify the specific vehicle type; S133: Obtain the voltage signal when the vehicle passes through according to the weighing sensors set between adjacent geomagnetic sensors, and calculate the vehicle weight through a calibration method; A behavior analysis module, which is connected to the data acquisition module through data transmission technology and is used to predict the behavior tendency of pedestrians and predict the passing situation of vehicles; A data analysis module, which is connected to the behavior analysis module through data transmission technology and is used to analyze whether to allow pedestrians to pass through the zebra crossing and estimate the time for pedestrians to pass through the zebra crossing through predictive data analysis, including the following steps: S31: Compare the actual distance of the vehicle from the zebra crossing with the calculated braking distance to determine whether it is safe for pedestrians to pass through the zebra crossing; S32. When it is judged that the first vehicle in the traffic flow can still stop in front of the zebra crossing when adding the margin time and the yellow light time as the reaction time, calculate the vehicle stop position; S33. According to the number of the pedestrian queue, except for the first pedestrian, keep a safe interval between each subsequent pedestrian and the previous one, and calculate the time required for all types of people to pass through the zebra crossing; S34. Add another margin time to the time required for the nth person to cross the zebra crossing, and calculate the green light time of the zebra crossing; The warning trigger module is connected to the data analysis module through data transmission technology, and is used to trigger a warning signal when it is judged that pedestrians are not allowed to pass, and calculate the green light countdown for vehicles and the red light countdown for pedestrians, including the following steps: S41. Calculate the vehicles in the current traffic flow that can safely stop at the front end of the zebra crossing, and calculate the vehicles in the traffic flow with a safe stop speed of 0 according to the kinematic formula; S42. Add a safety margin time to the braking time to ensure that the vehicle has enough time to react and stop safely; S43. Calculate the red light duration of the zebra crossing according to the remaining duration of the lane green light; The dynamic update module is connected to the data analysis module through data transmission technology, and is used to update the predicted passing time and passing situation of pedestrians in real time when pedestrians pass through the zebra crossing, including the following steps: S51. Real-time sense the pressure change through the pressure sensors deployed on the zebra crossing, record the foot landing points and movement information of pedestrians, and calculate the speed change of pedestrians on the zebra crossing in combination with image acquisition; S52. When the pedestrian speed changes or obvious imbalances, pauses, etc. that do not conform to the normal model appear through image analysis of the pedestrian posture, and the pressure sensor data shows chaotic foot movement, etc., it is determined that the walking state is abnormal; S53. Calculate the remaining distance of the pedestrian to the end point, and calculate the time for the pedestrian to pass through the remaining zebra crossing distance, and update the green light duration of the zebra crossing and the red light duration of the lane; The signal control module is connected to the data analysis module and the warning trigger module through data transmission technology, and is used to control the display of the signal lights, trigger warning information, send out sound and light alarms to pedestrians, and update the display situation of the signal lights in real time.

2. The traffic safety signal and control system for pedestrian crosswalk according to claim 1, characterized in that, In the step S111, the formula for calculating the reflected wave intensity is: , where is the reflected wave intensity, is the transmission power set by the device, and are the transmitting and receiving antenna gains respectively, is the wavelength, and , is the speed of light, is the fixed frequency, is the target radar cross section area, is the target distance. In the step S112, the formula for calculating the pedestrian speed is: , where is the current speed of the pedestrian, is the wavelength, is the fixed frequency difference. In the step S113, the formula for calculating the distance between the pedestrian and the radar is: , where is the target distance, that is, the distance between the pedestrian and the radar, is the speed of light, is the time delay. In the step S114, the formula for depicting the pedestrian motion trajectory is: , where is the position of the pedestrian at the next moment, is the position of the pedestrian at the previous moment, is the state transition matrix, is the control matrix, is the control variable.

3. The traffic safety signal and control system for a pedestrian crosswalk according to claim 2, characterized in that In the step S123, the formula for image denoising is: , where is the new pixel value of the image at the coordinate after Gaussian filtering denoising processing, is the pixel value at the coordinate in the original image, is the standard deviation of the Gaussian kernel. The formula for enhancing image contrast is: , where is the grayscale histogram of the graph, and are both grayscale values, and the value range is , is the cumulative distribution function, is the new pixel grayscale value of the image at the coordinate after histogram equalization processing, is the pixel grayscale value of the original image at the coordinate . In the step S124, the formula for the element at the coordinate in the output feature map is: , where is the element value at the coordinate and channel in the output feature map , and are the coordinate indexes of the output feature map in the height and width directions respectively, is the size of the convolutional kernel in the high dimension, and are the indexes in the high dimension and width dimension of the convolutional kernel respectively, is the channel index of the input feature map, is the coordinate in the convolutional kernel , the input channel is and the output channel is of the weight value, is the bias value of the weight channel , is the stride of the convolution operation. In the step S125, the formula for flattening the feature map is: , where is the one-dimensional image feature vector obtained after the flattening operation, , and are the height, width and number of channels of the output feature map of the last convolutional pooling layer respectively. In the step S126, the formula for calculating different pedestrian types is: , where is the Class score, is the total number of categories, is the category index.

4. The traffic safety signal and control system for pedestrian crosswalk according to claim 3, characterized in that In the step S131, the formula for calculating the vehicle speed is: , where is the vehicle speed, is the preset length between adjacent geomagnetic sensors, is the time difference for the vehicle to pass between adjacent geomagnetic sensors. In the step S132, the formula for estimating the vehicle length is: , where is the estimated vehicle length, is the time difference for the front and rear wheels of the vehicle to pass the geomagnetic sensor. In the step S133, the formula for calculating the vehicle weight is: , where is the vehicle weight, is the calibration coefficient, is the voltage signal, is the offset.

5. A traffic safety signal and control system for a pedestrian crosswalk according to claim 4, characterized in that, The behavior analysis module uses a weighted fusion algorithm to obtain the fused speed from the speed data measured by the millimeter-wave radar and the speed estimate obtained by combining the depth sensor with visual analysis. The formula is , where is the fused speed, and are respectively the speed measured by the millimeter-wave radar and the speed estimate obtained by combining the depth sensor with visual analysis, and are the corresponding weights respectively, and . And a speed threshold is set. Where indicates that the pedestrian is in a stationary state, indicates a slow walking state, indicates a normal walking state, indicates a fast walking state. By and the ratio of the set threshold speed, the walking state of the pedestrian is judged.

6. The traffic safety signal and control system for a pedestrian crosswalk according to claim 4, characterized in that The said behavior analysis module calculates the braking distance of the vehicle according to the vehicle weight, driving speed and vehicle type. The formula is as follows: , where is the braking distance, is the braking coefficient, is the acceleration due to gravity.

7. The traffic safety signal and control system for pedestrian crosswalk according to claim 6, characterized in that, In the said step S32, the formula for calculating the vehicle stop position is: , where is the final velocity of the first vehicle in the current traffic flow, i.e., , is the current velocity, is the acceleration, is the time, is the displacement distance, i.e., when the red light of the lane lights up. In the formula, is the time required for the first vehicle in the traffic flow to brake, is a margin time, and the minimum value is , is the yellow light duration of the lane, and the value range is . In the said step S33, the formula for calculating the time required for all types of people to cross the zebra crossing is: , where is the time required for the th person to cross the zebra crossing, is the total length of the pedestrian queue, is the length of the zebra crossing, is the average velocity of the slowest person in the pedestrian queue crossing the zebra crossing, is the safe interval distance between pedestrians. In the said step S34, the formula for calculating the green light time of the zebra crossing is: , where is the green light time of the zebra crossing, is the time required for the first vehicle in the traffic flow to brake, is another margin time, and the minimum value is . At the same time, add a yellow light time of the zebra crossing to find the red light time of the lane. The formula is , where is the remaining red light time of the lane, is the yellow light time of the zebra crossing, and the value range is .

8. A pedestrian crosswalk traffic safety signal and control system according to claim 7, characterized in that, In the step S41, the formula of dynamics is , in the formula, is the speed of the vehicle at the instant when braking starts, is the time experienced by the vehicle from the start of braking to safe stop, i.e., the braking duration, is the displacement traveled by the vehicle during the process from the start of braking to safe stop, i.e., the braking distance. In the step S42, the formula for calculating the braking time is: , in the formula, is the remaining duration of the green light of the lane, is another margin duration, and the minimum value is . In the step S43, the formula for calculating the red light duration of the zebra crossing is: , in the formula, is the red light duration of the zebra crossing, is the yellow light duration of the lane, and the value range is , is another margin duration, and the minimum value is .

9. A traffic safety signal and control system for pedestrian crosswalks according to claim 7, characterized in that, In the step S51, the formula for calculating the speed change of a pedestrian is: , where and are the walking speeds of the pedestrian at adjacent times respectively, is the speed change value of the pedestrian's walking, is the time interval. In the step S52, the formula for comparing the pedestrian speed changes is: , where and are the average speed and acceleration of the pedestrian on the zebra crossing at present respectively, , , and are the set minimum, maximum average speeds and accelerations of the pedestrian on the zebra crossing respectively. In the step S53, the formula for calculating the remaining distance of the pedestrian to the end point is: , where is the remaining distance of the pedestrian to the end point, is the total length of the zebra crossing, is the distance the pedestrian has walked. The formula for calculating the time for the pedestrian to pass the remaining distance of the zebra crossing is , where is the predicted time for the pedestrian to pass the remaining distance of the zebra crossing, is the recovery coefficient, is the possible acceleration recovery situation of the pedestrian, is the impossible acceleration recovery situation of the pedestrian. The formula for updating the green light duration of the zebra crossing and the red light duration of the lane is: , where is the updated green light duration of the zebra crossing, is the elapsed time of the green light of the zebra crossing, is the updated red light duration of the lane, is the elapsed time of the red light of the lane.

10. A traffic safety signal and control system for a pedestrian crosswalk as claimed in claim 1, wherein, The signal control module controls the signal lights including the following steps: S61. Receive instructions through wired or wireless communication, and check the format and data integrity; S62. Convert the digital instructions into analog control signals suitable for driving the circuit; S63. Drive the signal light driving circuit to control the on / off switching of the red and green lights and the start / stop of the sound and light alarm; S64. Feed back the working state of the signal lights to the data analysis module for data storage and update.

Images