Unmanned enforcement apparatus based on pedestrian traffic information and operating method of the same

Abstract

An unmanned enforcement device according to one embodiment disclosed in this document may include a protection reinforcement time zone derivation unit that dynamically calculates a threshold based on time-based pedestrian traffic volume data of an area of ​​interest and derives a time zone among a plurality of time zones where the pedestrian traffic volume exceeds the threshold as a protection reinforcement time zone, and a consistency evaluation information generation unit that generates consistency evaluation information based on the overlap range between the protection reinforcement time zone and the current enforcement time zone at each preset evaluation cycle.

Description

Technology Field

[0001] The embodiments disclosed in this document relate to an unmanned enforcement device based on pedestrian traffic information and a method of operation thereof. Background Technology

[0002] Conventional unmanned enforcement devices and time-based enforcement systems operate based on fixed pre-determined schedules, failing to reflect location- and time-specific variability. Consequently, this leads to inconsistent enforcement during off-peak hours and gaps in pedestrian protection during peak hours. Furthermore, conventional operating systems fail to provide separate procedures to evaluate or correct whether the currently applied enforcement time zones align with actual pedestrian traffic distribution, based on accumulated traffic data from each enforcement point. Moreover, the conventional approach of applying the same fixed threshold or absolute numerical standard to all enforcement points loses discriminative power when applied uniformly in environments with significant disparities in traffic volume across locations. Therefore, technical means for evaluation and updating based on relative thresholds derived from the distribution accumulated independently by each enforcement point are required.

[0003] For example, inconsistent enforcement that undermines drivers' acceptance of policies may occur because enforcement of the same intensity is uniformly applied even when only an average of 1 to 2 pedestrians pass through per hour during nighttime hours. Conversely, a gap in pedestrian protection may arise because enforcement of appropriate intensity is not carried out, even though more than 30 pedestrians pass through per hour during nighttime hours in school districts (e.g., 7 PM to 10 PM), simply because this time is not included in the current enforcement time zones. Furthermore, the absolute traffic volume at enforcement points within child protection zones varies significantly depending on the location; for instance, points near the main gate of large elementary schools see 80 to 100 pedestrians per hour during commuting hours, while quiet school routes see only 5 to 10 pedestrians per hour, resulting in a traffic volume gap of more than tenfold. The problem to be solved

[0004] The embodiments disclosed in this document are devised to solve the problems of the prior art as described above. The purpose is to diagnose and / or correct the inefficiency of a fixed pre-set schedule based on data by having an unmanned enforcement device in a child protection zone independently collect pedestrian traffic information and calculate statistics, derive a time period for enhanced protection from the actual pedestrian traffic patterns at crosswalks, and compare and analyze this with the current enforcement time period pre-set in the vehicle enforcement unit to automatically generate information on the consistency of enforcement standards and information on proposing changes to the enforcement time period.

[0005] Furthermore, the embodiments disclosed in this document aim to overcome the limitations of uniform absolute value standards and enable the integrated operation of multiple unmanned enforcement devices by reflecting the reality that the absolute volume and distribution of traffic vary depending on the location or time of each unmanned enforcement device, and by deriving enhanced protection time periods based on thresholds calculated from the pedestrian traffic volume distribution of each device itself.

[0006] Furthermore, the embodiments disclosed in this document aim to derive enhanced protection time zones from the perspective of risk exposure rather than simple traffic volume by calculating weights for vulnerable road users such as children and the elderly; to automatically generate temporary evaluation and / or proposal information without inputting external event information (referring to supplementary information provided by external systems or operators, such as event schedules, academic calendars, construction schedules, and weather information) when statistical outliers are detected compared to normal patterns; to resolve the cold start problem by providing a traffic volume distribution template of a similar location cluster as an initial value for newly installed unmanned enforcement devices; and to provide a self-updating structure for enforcement time zones approved by the operating entity. means of solving the problem

[0007] An unmanned enforcement device according to one embodiment disclosed in this document may include a protection reinforcement time zone derivation unit that dynamically calculates a threshold based on time-based pedestrian traffic volume data of an area of ​​interest and derives a time zone among a plurality of time zones where the pedestrian traffic volume exceeds the threshold as a protection reinforcement time zone, and a consistency evaluation information generation unit that generates consistency evaluation information based on the overlap range between the protection reinforcement time zone and the current enforcement time zone at each preset evaluation cycle.

[0008] A method of operation of an unmanned enforcement device according to one embodiment disclosed in this document may include the steps of: dynamically calculating a threshold based on time-based pedestrian traffic volume data of an area of ​​interest; deriving a time period among a plurality of time periods in which pedestrian traffic volume exceeds the threshold as a protection reinforcement time period; and generating consistency evaluation information based on the overlap range between the protection reinforcement time period and the current enforcement time period at each preset evaluation cycle. Effects of the invention

[0009] An unmanned enforcement device according to one embodiment can diagnose and / or correct the inefficiency of a pre-fixed schedule based on data by measuring pedestrian traffic patterns to derive a time period for enhanced protection and comparing and analyzing it with the current enforcement time period.

[0010] Furthermore, by using relative thresholds calculated from each unmanned enforcement device's own traffic volume distribution, consistent evaluation criteria are applied despite differences in traffic volume by location; additionally, integrated system-level operation is possible through the normalization comparison of multiple devices and the resolution of cold start issues for newly installed devices.

[0011] In addition, by calculating weights for vulnerable road users, time zones for enhanced protection are derived from the perspective of risk exposure rather than simple traffic volume; and when statistical outliers are detected relative to the usual distribution, temporary evaluation and / or suggestion information is automatically generated solely from the output of the pedestrian traffic information collection unit without the need for separate input of external event information, thereby ensuring operational adaptability to temporary traffic volume fluctuations such as events or inclement weather.

[0012] In addition, through edge-level anonymization and an operator approval structure, anonymization is maintained to comply with personal information protection principles, and implementation is possible by adding a pedestrian traffic information collection unit to the existing unmanned enforcement device infrastructure, thereby minimizing the burden of building new infrastructure. Brief explanation of the drawing

[0013] FIG. 1 is a block diagram illustrating the configuration of an unmanned enforcement device according to one embodiment disclosed in this document. FIG. 2 is a diagram conceptually showing the application situation of an unmanned enforcement device according to one embodiment disclosed in this document. FIG. 3(a) shows pedestrian traffic data by time of day for a first unmanned enforcement device with high pedestrian traffic volume, FIG. 3(b) shows pedestrian traffic data by time of day for a second unmanned enforcement device with low pedestrian traffic volume, and FIG. 3(c) shows the cumulative traffic volume, cumulative traffic volume ratio, and percentile calculated by sorting the pedestrian traffic data of the first unmanned enforcement device in order of size, respectively, to explain the operation of an unmanned enforcement device according to an embodiment disclosed in this document. FIGS. 4 and FIGS. 5 are flowcharts showing a method of operation of an unmanned enforcement device according to one embodiment disclosed in this document. Specific details for implementing the invention

[0014] Preferred embodiments of the present invention will be described in detail below with reference to the attached drawings. In this embodiment, an unmanned speed enforcement device is described as an example of an unmanned enforcement device; however, the present invention can be applied to unmanned enforcement devices that perform traffic enforcement related to pedestrian protection, such as unmanned traffic signal violation enforcement devices and unmanned lane violation enforcement devices.

[0015] FIG. 1 is a block diagram illustrating the configuration of an unmanned enforcement device according to an embodiment disclosed in this document, and FIG. 2 is a diagram conceptually showing the application situation of an unmanned enforcement device according to an embodiment disclosed in this document. FIG. 3 is a diagram for explaining the operation of an unmanned enforcement device according to an embodiment disclosed in this document.

[0016] First, referring to FIGS. 1 and 2, the unmanned enforcement device (100) includes a vehicle enforcement unit (110), a pedestrian traffic information collection unit (120), a protection reinforcement time zone derivation unit (130), and a consistency evaluation information generation unit (140), and may optionally further include an approval management unit (150) and a learning unit (160). The unmanned enforcement device (100) may be connected via a network to an external operating entity terminal (200) and a central management server (300), and may be connected to a variable information sign (400) linked to the unmanned enforcement device (100). As illustrated in FIG. 2, the unmanned enforcement device (100) is installed on a roadside post facing an area of ​​interest (e.g., a crosswalk (P) within a child protection zone) and is positioned so as to be able to observe both vehicles traveling in the uphill and downhill lanes and pedestrians passing through the area of ​​interest, and a variable information sign (400) may be installed near the unmanned enforcement device (100) to display current enforcement criteria information (e.g., enforcement conditions including enforcement time zones, etc.) to the driver.

[0017] Each component of the unmanned enforcement device (100) may be installed on-site and integrated within a single housing, or distributed within adjacent roadside devices and interconnected via wired or wireless connections, and some functions may be performed by a central management server (300) that controls or receives multiple unmanned enforcement devices (100). That is, the components (110 to 160) and functions of the unmanned enforcement device (100) described in this specification may be implemented in any one of the following forms: an edge form in which all functions are processed by an edge computing module mounted within the unmanned enforcement device (100) or an adjacent roadside device; a server form in which all functions are processed by a central management server (300); or a hybrid form in which some functions are processed at the edge and the remaining functions are processed at the server. For convenience of explanation, the following description focuses on an edge-based integrated configuration in which the main functions are processed within the unmanned enforcement device (100), but the scope of the present invention includes all of the above implementation forms. However, the vehicle enforcement unit (110) and pedestrian traffic information collection unit (120), which perform data acquisition through hardware such as cameras and sensors, are installed at a site where an area of ​​interest can be observed and perform the function of acquiring road images and pedestrian traffic information at the site, and the distinction between the edge type, server type, and hybrid type may mainly relate to the location where the data processing and analysis functions performed by the protection reinforcement time zone derivation unit (130), consistency evaluation information generation unit (140), approval management unit (150), and learning unit (160) are processed.

[0018] The vehicle enforcement unit (110) detects at least one of a traffic law violation, such as exceeding the speed limit, signal violation, tailgating, or lane violation, within a preset enforcement time period for a vehicle traveling on the road, and can obtain violation information including the vehicle number of the violating vehicle, the time of violation, the type of violation, and evidence video proving the violation, based on the acquired road image data. The evidence video may include at least one of a speeding violation video, a signal violation video, a tailgating video, a lane violation video, and a display video of a variable information sign linked to an unmanned enforcement device.

[0019] In addition, the vehicle enforcement unit (110) can store the above violation information based on the vehicle number.

[0020] The vehicle enforcement unit (110) may be implemented by including a speed measuring means such as a radar or lidar for measuring the speed of a vehicle, a vehicle number recognition camera, an infrared light, a power control unit, and a control unit that integrates and controls these components and performs a number recognition algorithm, and may optionally include an auxiliary camera. The vehicle enforcement unit (110) performs enforcement against vehicles exceeding a predetermined threshold based on the regulated speed of the road set in advance, and may be configured to enable time-based speed enforcement or variable speed enforcement by applying different regulated speeds depending on the time of day. The vehicle enforcement unit (110) may receive the setting value of the current enforcement time period updated from the approval management unit (150) described later, and perform enforcement against vehicles violating within the said current enforcement time period. In this specification, the variation of enforcement intensity includes at least one of the following: a form of applying different regulated speeds by time period, a form of determining whether to perform enforcement by time period, and a form of applying different enforcement thresholds by time period. Accordingly, the update of the current enforcement time period may be a concept that encompasses not only the decision to perform or not perform enforcement but also the decision to apply differential enforcement intensity. Meanwhile, the vehicle enforcement unit (110) may include at least one of a speed measuring sensor for measuring the speed of a vehicle, a signal detection means for detecting a signal violation, a camera for collecting situational information of a signal violation, a camera for collecting situational information of tailgating, and a camera for capturing the display content of a variable information sign (400) to prove variable speed enforcement, and may include an enforcement information module that stores the enforcement information obtained by these by matching it with the vehicle number of the corresponding vehicle.

[0021] A pedestrian traffic information collection unit (120) is provided on one side of an unmanned enforcement device (100) so as to be integrated or detachably oriented toward the direction of an area of ​​interest (e.g., a crosswalk (P)), and can detect pedestrians passing through the area of ​​interest to generate pedestrian traffic volume data in time units and accumulate and store the data. The pedestrian traffic information collection unit (120) may include a camera (121) for detecting pedestrians and a lidar and / or radar (122), and can detect pedestrians using at least one of a visible light camera, lidar, radar, or thermal imaging camera. The pedestrian traffic information collection unit (120) can accumulate and store pedestrian traffic volume data by counting detected pedestrian objects by time, and classify the detected pedestrian objects into age groups of children, the elderly, and the general public, and aggregate and store the number of pedestrians by age group by time. The classification of the age groups is performed based on at least one feature quantity among the body proportions, skeletal keypoints, head-to-body ratio, walking speed, and walking pattern of the pedestrian objects. In this specification, the term "time period" refers to a unit time for aggregating and analyzing pedestrian traffic volume, and the length of the unit time (time period) can be variably set according to the operating environment and the requirements for analysis precision. While a shorter unit time allows for the precise capture of short-term fluctuations such as school commuting peaks, a longer accumulation period (a period for accumulating traffic volume data to calculate the traffic volume distribution by time period) may be required to ensure statistical stability of the distribution. The pedestrian traffic information collection unit (120) can generate pedestrian traffic volume data including traffic volume by time period, date, day of the week, and age group.

[0022] In addition, the above time intervals can be variably set, for example, in 10-minute increments, 30-minute increments, or 1-hour increments.

[0023] For example, the pedestrian traffic volume data generated by the pedestrian traffic information collection unit (120) may be composed of records including fields for time period, date, day of the week, number of children, number of elderly people, and number of general people, and as an example, on Monday, March 1, 2026, 47 children, 8 elderly people, and 12 general people are recorded during the time period from 07:30 to 08:00, 88 children, 6 elderly people, and 5 general people during the time period from 08:00 to 09:00, and 3 children, 2 elderly people, and 5 general people during the time period from 21:00 to 21:30, respectively.

[0024] The pedestrian traffic information collection unit (120) may be configured as an edge computing module or a cloud server to protect personal information, in which case the original video that can identify a person is automatically deleted or masked within a preset time, and the data transmitted externally or delivered to subsequent components may be limited to de-identified metadata including age group labels, time of passage, and direction of passage.

[0025] The protection reinforcement time zone derivation unit (130) can derive a time zone that exceeds a threshold dynamically calculated from the time-based pedestrian traffic volume data of the area of ​​interest monitored by the unmanned enforcement device (100) among the pedestrian traffic volume data generated by the pedestrian traffic information collection unit (120) as a protection reinforcement time zone (T_n). In this specification, the threshold refers not to an absolute value pre-set externally, but to a value dynamically calculated from the statistical characteristics of the time-based pedestrian traffic volume distribution accumulated by the unmanned enforcement device (100) for a certain period in the area of ​​interest. This statistically based relative threshold can provide the effect of automatically deriving a protection reinforcement time zone (T_n) that corresponds to the location characteristics of each point in an environment of multiple enforcement points where the absolute amount and distribution of traffic volume differ depending on the location.

[0026] The protection reinforcement time zone derivation unit (130) can calculate a threshold value by applying a normalization operation based on the mean and standard deviation of the pedestrian traffic volume distribution by time zone, and specifically, can calculate a threshold value based on any one of the following or a combination thereof. The protection reinforcement time zone derivation unit (130) can calculate a threshold value (V_th) by applying a pre-set percentile reference value to the distribution of pedestrian traffic volume data by time zone as shown in Equation 1 below.

[0027] [Mathematical Formula 1]

[0028] V_th = Percentile_q(V)

[0029] Here, V represents the distribution of pedestrian traffic volume by time period, q represents a preset percentile reference value, and q can be changed by the operating entity.

[0030] In addition, the protection reinforcement time zone derivation unit (130) can calculate the threshold value (V_th) by adding the product of the standard deviation (σ) and a preset coefficient (K) to the average value (μ) of the distribution of pedestrian traffic volume data by time zone as shown in mathematical formula 2 below.

[0031] [Mathematical Formula 2]

[0032] V_th = μ + K·σ

[0033] In addition, the protection reinforcement time zone derivation unit (130) can derive a time zone (T_n) in which the normalized value (z(t)), obtained by normalizing the pedestrian traffic volume V(t) by time zone with the mean value (μ) and standard deviation (σ) of the distribution as shown in Equation 3 below, exceeds a preset normalized reference value (z_th).

[0034] [Mathematical Formula 3]

[0035] z(t) = (V(t)-μ) / σ ≥ z_th

[0036] The protection enhancement time zone derivation unit (130) may apply a weighted traffic volume (V_w(t)) calculated by applying a weighting factor to the traffic volume by age group and summing it, instead of the traffic volume by time zone (V(t)), in order to reflect the risk exposure of vulnerable road users such as children and the elderly. The weighted traffic volume is expressed by the following mathematical formula 4. Here, the weighted traffic volume (V_w(t)) can be collected by the pedestrian traffic information collection unit (120).

[0037] [Mathematical Formula 4]

[0038] V_w(t) = C(t)·W₁ + S(t)·W₂ + N(t)·W₃

[0039] Here, C(t) represents the traffic volume of the child group at time t, S(t) represents the traffic volume of the elderly group, and N(t) represents the traffic volume of the general public group, and the first weight (W1) of the child group, the second weight (W2) of the elderly group, and the third weight (W3) of the general public group can each be set to satisfy the relationship W1 > W2 > W3. When using weighted traffic volume, V or V(t) in Equations 1 to 3 above can be replaced with V_w(t). The percentile reference value (q), coefficient (K), normalization reference value (z_th), and weights (W1, W2, W3) used in calculating the threshold value can be varied by taking into account the operating policy of the operating entity and the specific characteristics of the surrounding environment where the unmanned enforcement device (100) is installed. In other words, when using weighted traffic volume, the basic data can be calculated by replacing it with a weighted score distribution (V_w) that reflects risk, such as '1 child = 5 points', and through this, the system can calculate weighted traffic volume based on actual risk exposure without being bound by the simple number of pedestrians.

[0040] The consistency evaluation information generation unit (140) can generate consistency evaluation information based on the overlap range between the protection reinforcement time zone (T_n) derived by the protection reinforcement time zone derivation unit (130) and the current enforcement time zone (T_c) set in the vehicle enforcement unit (110) at each preset evaluation cycle. The threshold is a criterion for defining the protection reinforcement time zone (T_n), and the consistency evaluation is a criterion for measuring whether the current enforcement time zone (T_c) corresponds to the protection reinforcement time zone (T_n). Based on the comparison results between the protection reinforcement time zone (T_n) and the current enforcement time zone (T_c), it is possible to determine whether to maintain, change, or supplement the current enforcement time zone (T_c). Accordingly, the inefficiency of a fixed enforcement schedule can be resolved, and an enforcement schedule optimized for the location characteristics of each enforcement point can be provided without continuous external intervention. The comparative analysis operation of the consistency evaluation information generation unit (140) is performed continuously regardless of the degree of inclusion or overlap between the protection reinforcement time period (T_n) and the current enforcement time period (T_c), and the result is produced in the form of consistency evaluation information. Therefore, even when consistency is good, the comparative analysis result is produced as evaluation information and can be provided to the operating entity terminal as data to prove or judge the appropriateness of the current enforcement time period setting or operation policy.

[0041] The consistency evaluation information generated by the consistency evaluation information generation unit (140) includes Coverage, which represents the effectiveness of pedestrian protection, as the ratio of the number of time periods that overlap with the current enforcement time period (T_c) among the protection reinforcement time periods (T_n), and Excess, which represents the ratio of unnecessary excessive enforcement that causes driver complaints, as the ratio of the number of time periods that do not overlap with the protection reinforcement time period (T_n) among the current enforcement time periods (T_c). The consistency evaluation information generation unit (140) calculates Coverage according to the following mathematical formula 5 and calculates Excess according to the following mathematical formula 6.

[0042] [Mathematical Formula 5]

[0043] Coverage = |T_n ∩ T_c| / |T_n|

[0044] [Mathematical Formula 6]

[0045] Excess = |T_c - T_n| / |T_c|

[0046] The alignment evaluation information may further include an alignment score calculated based on coverage and an excess rate, and the alignment score may be calculated by subtracting the value obtained by multiplying the excess rate by the second weighting factor (β) from the value obtained by multiplying the coverage by the first weighting factor (α) as shown in Equation 7 below. Here, the weighting factors (α, β) and the reference values ​​for the alignment score for generating change proposal information may be changed by the operating entity.

[0047] [Mathematical Formula 7]

[0048] Alignment = α·Coverage - β·Excess

[0049] The consistency evaluation information generation unit (140) determines that the consistency is low when the consistency score is less than a preset threshold.

[0050] The consistency evaluation information generation unit (140) generates information proposing a change in the enforcement time zone based on the protection reinforcement time zone (T_n) when the suitability score of the consistency evaluation information is lower than the preset threshold during a preset period. The change proposal information includes a recommended new enforcement time zone (i.e., the enforcement time zone to be applied after the change) and the degree of improvement (e.g., the amount or rate of increase of the suitability score) expected when applying the new enforcement time zone. Furthermore, the enforcement time zone change proposal information is not limited to the adjustment of the enforcement target time zone, but may further include information for adjusting the intensity of the enforcement criteria according to the distribution of pedestrian traffic volume by time zone. For example, it may include information for maintaining the existing enforcement criteria for time zones where pedestrian traffic is concentrated, and for low-density time zones where pedestrian traffic is low, information for suspending enforcement or increasing the speed limit to mitigate the enforcement intensity. Furthermore, to prevent unnecessary change proposals due to temporary fluctuations, the consistency evaluation information generation unit (140) may generate periodic change proposal information when the discrepancy between the protection reinforcement time zone (T_n) and the current enforcement time zone (T_c) accumulates for more than a preset number of days during a preset period. Here, the periodic change proposal information is distinguished from the enforcement time zone change proposal information generated when the suitability score is below a threshold value, and may refer to change proposal information generated periodically when the discrepancy between the protection reinforcement time zone and the current enforcement time zone satisfies the cumulative number of days condition. According to one embodiment, the default value of the cumulative number of days for generating periodic change proposal information may be 10 days per month, and the default threshold value for triggering the suitability score change proposal may be 0.7.

[0051] Referring to FIG. 3, the process of calculating thresholds and generating consistency evaluation information for unmanned enforcement devices installed in an area of ​​interest is explained in detail. FIG. 3(a) illustrates pedestrian traffic volume data by time of day for unmanned enforcement device A, which is installed near the main gate of a school and has a high volume of pedestrian traffic; FIG. 3(b) illustrates pedestrian traffic volume data by time of day for unmanned enforcement device B, which is installed on a quiet school route and has a low volume of pedestrian traffic; and FIG. 3(c) illustrates the cumulative traffic volume, cumulative traffic volume ratio, and percentile calculated by sorting the pedestrian traffic volume data of unmanned enforcement device A in ascending order of traffic volume. The protection reinforcement time period derivation unit (130) uses a threshold (V_th) that is dynamically calculated from the distribution of accumulated pedestrian traffic volume data, rather than a fixed absolute value assigned from the outside.

[0052] Referring to FIG. 3(c), the process of calculating the threshold (V_th) for the unmanned enforcement device A is explained. The protection reinforcement time zone derivation unit (130) sorts pedestrian traffic volume data by time zone in ascending order of traffic volume size. For example, since the information collection unit is 1 hour and a day is 24 hours, the total number of sorted data (N) is 24. Next, the protection reinforcement time zone derivation unit (130) calculates the position (index) of the data corresponding to the preset percentile value (q = 88 in this embodiment) according to the percentile position derivation formula Position = 1 + (N - 1) × (q / 100), resulting in Position = 1 + (24 - 1) × 0.88 = 21.24. That is, the threshold that serves as the boundary of the top 12% exists at the 21.24th position between the 21st and 22nd data among the sorted data.

[0053] In Fig. 3(c), the 21st data point is 45 people / hour (14:00–15:00) and the 22nd data point is 60 people / hour (16:00–17:00). Therefore, when the value at the 21.24th position is calculated by linear interpolation, it becomes 45 + 0.24 × (60 - 45) = 48.6 people / hour, and when this is rounded to the nearest decimal point, the threshold (V_th) is calculated to be approximately 50 people / hour. Accordingly, the top three time zones that exceed the threshold in sorting order, namely 16:00–17:00 (60 people), 07:00–08:00 (85 people), and 15:00–16:00 (90 people), are derived as the protection reinforcement time zones (T_n).

[0054] Referring to the cumulative traffic volume ratio shown in Fig. 3(c), out of a total of 537 pedestrians per day, the sum of pedestrians who traveled during the top three time periods is 235 (537 - 302), which accounts for approximately 43.7% of the total, confirming that pedestrian traffic is highly concentrated in specific time periods. When the threshold (V_th = approximately 50 people / hour) derived in this way is applied to the distribution of unmanned enforcement device A, the school arrival time (07:00–08:00) and school departure time (15:00–16:00, 16:00–17:00) that exceed this threshold are derived as protection reinforcement time periods (T_n), respectively. Assuming the current enforcement time zone (T_c) is from 06:00 to 19:00 (a total of 13 hours), the coverage is calculated as 3 / 3 = 1.0 because the entire enhanced protection time zone (T_n) is included within the current enforcement time zone (T_c), and the non-conformity enforcement rate (Excess) is 10 / 13 due to the 10 hours within the current enforcement time zone (T_c) that are not included in the enhanced protection time zone (T_n). It is calculated as 0.77, and if weighting factors α = 1 and β = 0.5 are applied, the goodness-of-fit score (Alignment) is 1 × 1.0 - 0.5 × 0.77 It is calculated as 0.615. Here, a weighting factor α = 1 means that securing pedestrian safety (coverage) is reflected 100% as the top priority goal, and β = 0.5 means that the reduction of excessive enforcement (non-conformity enforcement rate) that causes driver complaints is placed lower priority than safety, and that the penalty is reflected only half.

[0055] Since the above alignment score of 0.615 is less than the default threshold of 0.7, the alignment evaluation information generation unit (140) generates information proposing a change in the enforcement time zone to narrow the recommended enforcement time zone to the protection reinforcement time zone (T_n) itself, which is 07:00–08:00, 15:00–16:00, and 16:00–17:00. When this change proposal information is applied, the coverage is maintained at 1.0 while the excess of misalignment is reduced, thereby improving the alignment score. Thus, the misalignment between the pedestrian traffic volume distribution and the enforcement intensity can be resolved while maintaining the effectiveness of child protection.

[0056] Meanwhile, unmanned enforcement device B, exemplified in Fig. 3 (b), has a daily cumulative traffic volume of only about 52 people, which is about 1 / 10 the level of unmanned enforcement device A (537 people). However, based on the distribution characteristics, if the same sorting by size and percentile extraction process (q = 88) as above is applied, the relative threshold (V_th) in the distribution is derived as about 5 people / hour. As a result, the school arrival time (07:00–08:00) and school departure time (15:00–16:00, 16:00–17:00) that exceed 5 people / hour are derived as protection reinforcement time periods (T_n), identical to unmanned enforcement device A. Thus, even though the absolute traffic volume of unmanned enforcement device A and unmanned enforcement device B differs by about 10 times, protection reinforcement time periods (T_n) corresponding to their respective location characteristics are derived based on the same definition of time periods exceeding the 88th percentile of the distribution. In other words, by applying a distribution-based relative threshold, the problem where almost all time zones are classified as enhanced protection time zones (T_n) at some locations and none are identified as enhanced protection time zones at others when absolute value criteria are applied uniformly is resolved, and enhanced protection time zones (T_n) corresponding to the locational characteristics of each location can be automatically derived.

[0057] Additionally, the consistency evaluation information generation unit (140) determines that when traffic volume exceeding a preset multiple of the standard deviation relative to the time-based distribution is detected, it is a statistical outlier, and automatically generates temporary evaluation information and temporary proposal information based on pedestrian traffic volume data output by the pedestrian traffic information collection unit (120) without separate input of external calendars, weather information, event schedules, etc. Through this, the unmanned enforcement device (100) can secure field adaptability to propose immediate enforcement schedule correction (e.g., temporary enforcement reinforcement for 24 to 72 hours) even in the event of temporary and sudden fluctuations in traffic volume. Here, a statistical outlier can be defined as traffic volume exceeding K times (K ≥ 2, default value 2) the standard deviation of the usual time-based average pedestrian traffic volume distribution, and the validity period of the temporary evaluation information and temporary proposal information can be, for example, 24 to 72 hours. Temporary evaluation information and temporary proposal information are transmitted to the operating entity terminal on a separate track from the regular change proposal information and consistency evaluation information generated at each regular evaluation cycle, and may have a shorter validity period than the regular change proposal information. Additionally, temporary evaluation information and temporary proposal information are temporarily stored only during the validity period and are deleted upon the expiration of the validity period, allowing them to be managed on a temporary track separated from the regularly managed information.

[0058] The information generated by the consistency evaluation information generation unit (140) may be configured to include at least one of a consistency quantitative indicator including coverage, a non-conformity enforcement rate, and a suitability score, a recommended enforcement time zone after the change, visualization data of the pedestrian traffic volume distribution by time zone, and at least one of the degree of improvement of the consistency indicator expected when the change is applied. Additionally, the consistency evaluation information generation unit (140) may be configured to notify the operator terminal (200) when the statistical outlier is detected so that the operator can recognize it, and events that cause abrupt changes in pedestrian traffic patterns, such as events like festivals, the start of vacations, and the end of vacations, may be detected as the statistical outlier.

[0059] The approval management unit (150) is an optional component that transmits information generated by the consistency evaluation information generation unit (140) to the operating entity terminal (200) via a network, and updates the setting value of the current enforcement time zone of the vehicle enforcement unit (110) when it receives an approval signal from the operating entity terminal (200). The operating entity terminal (200) displays the content of the information, a visualization dashboard of pedestrian traffic volume by time zone, a comparison of the current enforcement time zone and the proposed enforcement time zone, and the expected effects when applying the change, and the operating entity can review this and select one of approval, rejection, or modification.

[0060] When the approval management unit (150) receives an approval signal from the operating entity terminal (200), it updates the setting value of the current enforcement time zone of the vehicle enforcement unit (110) and, at the same time, synchronizes the display content of the variable information sign (400) linked to the unmanned enforcement device (100) with the enforcement standard information corresponding to the updated current enforcement time zone. Through this structure, the change of the enforcement time zone is subject to the explicit approval of the operating entity, thereby protecting procedural legitimacy and reliability, and the history of the transmission, rejection, and approval of information is stored as an audit log and used as evidence in subsequent disputes or administrative reviews.

[0061] Additionally, the approval management unit (150) can receive and update a change input from the operating entity terminal (200) for at least one of the parameters used to calculate the threshold or reference value, the weighting coefficient used to calculate the consistency evaluation information, the cumulative reference day, and the coefficient used to determine statistical outliers.

[0062] The learning unit (160) can automatically update the parameters used to calculate the threshold of the protection reinforcement time zone derivation unit (130) and the threshold used to determine statistical outliers of the consistency evaluation information generation unit (140) by re-executing the distribution analysis on the traffic volume data collected by the pedestrian traffic information collection unit (120) at least once during the period of day, month, season, and semester. Through this, long-term fluctuations in pedestrian traffic patterns according to the academic calendar and season, etc., can be reflected in the threshold and the criteria for determining statistical outliers. Meanwhile, the detection of such statistical outliers can be used as an indicator to enable the operating entity to recognize whether an event that causes traffic patterns different from the usual, such as the start or end of a vacation or the occurrence of an event like a festival, has occurred.

[0063] The operating entity terminal (200) is a terminal used by an operating entity, such as the National Police Agency or a local government, and receives and displays consistency evaluation information or pedestrian traffic volume data from the unmanned enforcement device (100), and receives an approval signal from the operating entity and transmits it to the unmanned enforcement device (100).

[0064] A central management server (300) is network-connected to a plurality of unmanned enforcement devices (100) and an operating entity terminal (200), receives pedestrian traffic volume data by time period from the plurality of unmanned enforcement devices (100), and performs normalization processing for each unmanned enforcement device (100) to calculate a normalized distribution of traffic volume by location. Since the normalized distribution can be directly compared under the same evaluation criteria, the central management server (300) can apply a clustering operation to the normalized traffic volume distribution to classify the plurality of unmanned enforcement devices (100) by location type, and calculate a protection reinforcement time period (T_n) pattern derived within the same cluster as a recommended schedule template by location type and provide it to the operating entity terminal (200). Algorithms such as K-means, DBSCAN, or hierarchical clustering may be used for the clustering operation, and the location type may include, for example, at least one of a school front gate group, a school back gate group, an academy district group, a school commute route group, and a crosswalk adjacent to a residential area group.

[0065] In addition, the central management server (300) can solve the cold start problem, in which it is difficult to independently derive the protection reinforcement time period (T_n) due to insufficient accumulation of pedestrian traffic data for the newly installed unmanned enforcement device (100), by identifying the most similar cluster based on the location characteristics of the newly installed unmanned enforcement device (100), for example, the type of surrounding facilities, road width, or distance to a school, and provide the traffic volume distribution template or information regarding the protection reinforcement time period of the cluster as the initial value of the newly installed unmanned enforcement device (100) through the operator terminal (200).

[0066] When a newly installed unmanned enforcement device (100) accumulates its own data for a preset period, for example, 30 days or more, the template provided as an initial value is gradually converted into its own distribution and can be customized to the location characteristics of the unmanned enforcement device (100).

[0067] The variable information sign (400) is a sign that is linked to the unmanned enforcement device (100) and displays enforcement standard information corresponding to the current enforcement time period to the driver. When the setting value of the current enforcement time period is updated by the approval management unit (150), the displayed content is synchronized with the updated current enforcement time period. Additionally, an auxiliary camera for capturing the displayed content of the variable information sign (400) may be additionally installed in the unmanned enforcement device (100), and the vehicle enforcement unit (110) may store the displayed image of the variable information sign (400) captured by the auxiliary camera along with enforcement information as evidence of variable speed enforcement.

[0068] FIGS. 4 and FIGS. 5 are flowcharts showing a method of operation of an unmanned enforcement device according to an embodiment disclosed in this document. The operation of each step described below with reference to FIGS. 4 and FIGS. 5 can be performed by each component of the unmanned enforcement device (100) described with reference to FIG. 1.

[0069] First, referring to FIG. 4, in step S410, the vehicle enforcement unit (110) detects whether a vehicle violates regulations based on the acquired road image data and acquires the image of the vehicle and violation information corresponding to the violation. In step S420, the pedestrian traffic information collection unit (120) detects pedestrians passing through the area of ​​interest and generates pedestrian traffic volume data in time units. In step S430, the protection reinforcement time zone derivation unit (130) derives the time zone that exceeds a threshold dynamically calculated from the accumulated time-based pedestrian traffic volume distribution as the protection reinforcement time zone (T_n). In step S440, the consistency evaluation information generation unit (140) generates consistency evaluation information based on the overlap range between the protection reinforcement time zone (T_n) and the current enforcement time zone (T_c) at each preset evaluation cycle. In step S450, the consistency evaluation information generation unit (140) generates a proposal for changing the enforcement time zone based on the protection reinforcement time zone (T_n) when the suitability score of the consistency evaluation information is lower than the preset standard during a preset period. In step S460, the approval management unit (150) transmits the change proposal information to the operating entity terminal (200), updates the setting value of the current enforcement time zone of the vehicle enforcement unit (110) in response to receiving an approval signal from the operating entity terminal (200), and synchronizes the display content of the variable information sign (400) linked with the unmanned enforcement device (100) with the enforcement standard information corresponding to the updated current enforcement time zone. The current enforcement time zone updated in step S460 can be reused as a comparison input for the consistency evaluation information generation unit (140) in the next evaluation cycle.

[0070] Referring to FIG. 5, steps S510 to S540 may be substantially identical to steps S410 to S440 described with reference to FIG. 4. In step S550, the consistency evaluation information generation unit (140) determines whether a statistical outlier exceeding a preset multiple of the standard deviation relative to the distribution by time period has been detected. If a statistical outlier is detected as a result of the determination in step S550, in step S560, the consistency evaluation information generation unit (140) generates temporary proposal information based on pedestrian traffic volume data output by the pedestrian traffic information collection unit (120) without inputting external event information. Here, the external event information may be additional information including event schedules, academic schedules, construction schedules, weather information, etc., provided through an external system or an operating entity terminal (200). In step S570, the consistency evaluation information generation unit (140) determines whether the discrepancy between the protection reinforcement time zone (T_n) and the current enforcement time zone (T_c) has accumulated for more than a preset number of days. If, as a result of the determination in step S570, the discrepancy has accumulated for more than N days, then in step S580, the consistency evaluation information generation unit (140) generates the aforementioned periodic change proposal information. If, as a result of the determination in steps S550 and S570, it does not apply, then in step S590, the consistency evaluation information generation unit (140) generates enforcement time zone change proposal information based on the protection reinforcement time zone (T_n) when the suitability score of the consistency evaluation information during a preset period is lower than a preset standard. In step S600, the approval management unit (150) transmits information on a proposed change of enforcement time zone, a temporary proposal, or a regular change proposal to the operating entity terminal (200), updates the setting value of the current enforcement time zone of the vehicle enforcement unit (110) in response to receiving an approval signal from the operating entity terminal (200), and synchronizes the display content of the variable information sign (400) linked with the unmanned enforcement device (100) with the enforcement standard information corresponding to the updated current enforcement time zone.Each step of FIGS. 4 and FIGS. 5 described above may be partially omitted or the order may be changed, and a relearning operation by the learning unit (160) or protection reinforcement time zone derivation unit (130) described with reference to FIG. 1 may be performed in parallel, so that the parameters used to calculate the threshold and the statistical outlier determination threshold value for the accumulated data may be automatically updated in at least one cycle among days, months, seasons, and semesters.

Claims

Claim 1 An unmanned enforcement device comprising: a protection reinforcement time zone derivation unit that dynamically calculates a threshold based on time-based pedestrian traffic volume data of an area of ​​interest and derives a time zone among multiple time zones where the pedestrian traffic volume exceeds the threshold as a protection reinforcement time zone; and a consistency evaluation information generation unit that generates consistency evaluation information based on the overlap range between the protection reinforcement time zone and the current enforcement time zone set for enforcing traffic law violations by vehicles at each preset evaluation cycle, wherein the consistency evaluation information includes a first ratio which is the ratio of the time length overlapping the protection reinforcement time zone and the current enforcement time zone to the total time length of the protection reinforcement time zone, a second ratio which is the ratio of the time length of the current enforcement time zone that does not overlap with the protection reinforcement time zone to the total time length of the current enforcement time zone, and a suitability score calculated based on the first ratio and the second ratio. Claim 2 An unmanned enforcement device according to claim 1, wherein the protection enhancement time zone derivation unit calculates the threshold by applying a preset percentile reference value to the distribution of pedestrian traffic volume data by time zone. Claim 3 An unmanned enforcement device according to claim 1, wherein the protection enhancement time zone derivation unit calculates the threshold value by adding the product of the standard deviation of the distribution of pedestrian traffic volume data by time zone and a preset coefficient to the average value of the distribution of pedestrian traffic volume data by time zone. Claim 4 An unmanned enforcement device according to claim 1, wherein the consistency evaluation information generating unit generates information suggesting a change in the enforcement time zone based on the protection reinforcement time zone when the suitability score included in the consistency evaluation information is lower than a preset standard during a preset period. Claim 5 An unmanned enforcement device according to claim 4, wherein the proposed information for changing the enforcement time zone includes a new enforcement time zone and the degree of improvement of the consistency evaluation information expected upon the application of the new enforcement time zone. Claim 6 delete Claim 7 An unmanned enforcement device according to claim 1, wherein the consistency evaluation information generation unit generates at least one of temporary evaluation information and temporary proposal information based on the pedestrian traffic volume data without inputting external event information, based on the detection of a statistical outlier exceeding a preset multiple of the standard deviation relative to the mean in the distribution of the pedestrian traffic volume data by time period. Claim 8 An unmanned enforcement device according to claim 1, further comprising a learning unit that performs relearning on the pedestrian traffic volume data in at least one cycle among day, month, season, and semester to update the parameters used for calculating the threshold and the statistical outlier detection threshold. Claim 9 An unmanned enforcement device according to claim 1, further comprising a pedestrian traffic information collection unit that detects pedestrians passing through the area of ​​interest and generates pedestrian traffic volume data in time units. Claim 10 An unmanned enforcement device according to claim 9, wherein the pedestrian traffic information collection unit classifies the detected pedestrians into a child group, an elderly group, and a general public group, stores pedestrian traffic volume by age group by time period, and performs the classification based on at least one feature quantity among body ratio, skeletal key point, head-to-body ratio, walking speed, and walking pattern. Claim 11 An unmanned enforcement device according to claim 10, wherein the pedestrian traffic information collection unit uses the weighted traffic volume calculated by applying different weights to each of the pedestrian traffic volumes by age group as the pedestrian traffic volume by time period, wherein the weights are set such that the weight for the child group is greater than the weight for the elderly group, and the weight for the elderly group is greater than the weight for the general public group. Claim 12 An unmanned enforcement device according to claim 9, wherein the pedestrian traffic information collection unit detects the pedestrian using at least one of a lidar sensor, a radar sensor, and an image analysis sensor. Claim 13 An unmanned enforcement device according to claim 1, further comprising a vehicle enforcement unit that acquires images of vehicles and violation information based on road image data. Claim 14 An unmanned enforcement device according to claim 13, wherein the vehicle enforcement unit further includes a camera for capturing the display content of a variable information sign. Claim 15 An unmanned enforcement device according to claim 1, further comprising an approval management unit that transmits the consistency evaluation information to an operating entity terminal, updates the setting value of the current enforcement time zone based on receiving an approval signal from the operating entity terminal, and synchronizes the display content of the linked variable information sign with the enforcement standard information corresponding to the updated current enforcement time zone. Claim 16 An unmanned enforcement device according to claim 15, wherein the approval management unit receives and sets information regarding the protection enhancement time period corresponding to the location type of the newly installed unmanned enforcement device from the operating entity terminal when information regarding the protection enhancement time period is not set in the newly installed unmanned enforcement device. Claim 17 A method of operation of an unmanned enforcement device comprising: a step of dynamically calculating a threshold based on time-based pedestrian traffic volume data of an area of ​​interest and deriving a time period among a plurality of time periods in which the pedestrian traffic volume exceeds the threshold as a protection reinforcement time period; and a step of generating consistency evaluation information based on the overlap range between the protection reinforcement time period and the current enforcement time period set for enforcement of traffic law violations by vehicles at each preset evaluation cycle, wherein the consistency evaluation information comprises a first ratio which is the ratio of the time length overlapping the protection reinforcement time period and the current enforcement time period to the total time length of the protection reinforcement time period, a second ratio which is the ratio of the time length of the current enforcement time period that does not overlap with the protection reinforcement time period to the total time length of the current enforcement time period, and a suitability score calculated based on the first ratio and the second ratio.

Images