A robot advertisement publishing method based on people flow feature analysis

Abstract

The application relates to the field of mobile robot advertisement publishing control, and discloses a robot advertisement publishing method based on human flow feature analysis; the method collects personnel distribution information and spatial distance information in a candidate publishing unit through an environment perception component, extracts a human flow density, a main flow direction stability, an average moving speed, a proportion of personnel facing an advertisement display component, and a short-time staying proportion, and carries out a publishable window judgment accordingly; when the judgment condition is met, a lateral buffer publishing position located on the side of a main human flow line is generated, a mobile advertisement robot is controlled to move to the position, and a pre-publishing state and a main publishing state are executed; during the publishing process, the effective number of viewers, the continuous facing time length, the local congestion degree change and the safety distance change are continuously monitored, and when the keeping condition is not met, the current lateral buffer publishing position is exited and a preset cruise path is returned to; and the application can be used for improving the cooperation between an advertisement publishing time and a robot pose.

Description

Technical Field

[0001] This invention relates to the field of mobile robot advertising control, and more specifically to a robot advertising method based on pedestrian flow feature analysis. Background Technology

[0002] Mobile advertising robots are widely used in shopping mall passages, exhibition halls, waiting areas and public service halls. They typically integrate a mobile chassis, display screen, sensing components and control modules, and output advertising content by cruising along a preset route or stopping in a designated area, thereby replacing some fixed advertising spaces and improving the flexibility of advertising display.

[0003] In existing technologies, mobile advertising robots mostly use a combination of preset cruising routes and fixed-point docking to deliver advertisements. Some solutions use cameras, distance sensors, or human detection modules to identify people in the vicinity, and then trigger advertisement playback, switch displayed content, or adjust docking position based on whether people are detected, changes in the number of people, or partial facial attribute information. Other solutions determine whether to continue playing advertisements based on preset time periods, preset area popularity, or simple viewing behavior.

[0004] However, most of the above solutions are still based on fixed paths and single-trigger logic, lacking coordinated control over the dynamic flow of people within the passageway. Specifically, in existing structures, after the robot enters the publishing area, it often decides to stop and play based solely on whether people or local viewing behavior are detected. It fails to simultaneously constrain the relationship between the direction of the flow of people, the speed of passage, the state of dwelling, and the robot's stopping position. This results in the robot potentially stopping near the main passageway to publish even when people are passing through quickly or when the main passageway is relatively concentrated, making it difficult to establish a stable viewing relationship. This further leads to problems such as inaccurate timing of advertisement publication, poor display continuity, and interference with normal passage on site. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a robot advertising method based on pedestrian flow characteristic analysis to solve the technical problems existing in the prior art.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: A method for robot advertising delivery based on pedestrian flow characteristic analysis includes the following steps: S1: Control the mobile advertising robot to travel along a preset cruising path and divide the work area into multiple candidate publishing units; S2: Collect personnel distribution information and spatial distance information in each candidate publishing unit through the environmental perception component, and extract the personnel density, mainstream stability, average moving speed, personnel proportion facing the advertising display component, and short-term dwell proportion corresponding to each candidate publishing unit; S3: Determine the publishable window based on pedestrian density, mainstream stability, average movement speed, proportion of people facing the advertising display component, and short-term dwelling proportion. When pedestrian density is not higher than the preset traffic interference threshold, average movement speed is not higher than the preset speed threshold, proportion of people facing the advertising display component is not lower than the preset attention threshold, short-term dwelling proportion is not lower than the preset dwelling threshold, and mainstream stability meets the preset stability condition, the corresponding candidate publishable unit is determined as the current publishable unit. S4: For the currently publishable unit, generate a lateral buffer publishing position located next to the main flow line based on the main flow line, channel boundary, body size of the mobile advertising robot and preset safety gap, and control the mobile advertising robot to move to the lateral buffer publishing position so that the minimum distance between the outer contour of the mobile advertising robot body and the main flow line is not less than the preset safety gap, and the advertising display component faces the direction of the main flow. S5: After the mobile advertising robot reaches the lateral buffer publishing position, it executes the advertising publishing according to the pre-publishing state and the main publishing state. The pre-publishing state is used to output guiding information, and the main publishing state is used to output the target advertising content after the valid viewing judgment conditions are met. S6: During the ad publishing process, continuously monitor the number of effective viewers, continuous facing time, local congestion changes, and changes in the safe distance between the mobile ad robot and surrounding people. When the number of effective viewers is lower than the preset number threshold, the continuous facing time is lower than the preset time threshold, the local congestion is higher than the preset congestion threshold, or the safe distance is lower than the preset safe distance threshold, end the current publishing state, and control the mobile ad robot to exit the current lateral buffer publishing position and return to the preset cruise path to continue cruise.

[0007] Preferably, the environmental perception component includes an image acquisition unit and a distance measurement unit; The image acquisition unit is used to acquire images of people within the candidate publishing unit, and the distance measurement unit is used to acquire distance data within the candidate publishing unit. The personnel distribution information is obtained from personnel images, and the spatial distance information is obtained from distance data.

[0008] Preferably, in step S1, a front boundary and a rear boundary are set along the extension direction of the preset cruise path, and left and right boundaries are set in combination with the channel boundary of the work area, so that multiple candidate release units are enclosed by the front boundary, the rear boundary and the left and right boundaries; adjacent candidate release units are arranged continuously in the extension direction of the preset cruise path.

[0009] Preferably, in step S2, the mainstream stability is determined based on the consistency of the direction of movement trajectories of each person within a preset sampling period; The average moving speed is determined based on the displacement and time interval of each person's movement trajectory within a preset sampling period; The proportion of people facing the advertising display component is determined based on the proportion of the number of people whose heads or faces fall within the observation angle range of the advertising display component during the preset sampling period to the total number of people in the current candidate publishing unit. The short-stay percentage is determined based on the proportion of the number of people who enter the current candidate release unit within a preset sampling period and whose stay time is not less than a preset stay time threshold to the total number of people entering the current candidate release unit.

[0010] Preferably, in step S3, when multiple candidate publishing units meet the criteria for publishing windows, the current publishing unit is determined by comparing them in the following order: the proportion of people facing the advertising display component from high to low, the proportion of short-term stay from high to low, the flow density from low to high, and the distance from the current location of the mobile advertising robot from near to far.

[0011] Preferably, in step S4, the lateral buffer release position is offset from the main flow line to the side of the channel according to the normal direction of the main flow line. The offset endpoint satisfies that the outer contour of the mobile advertising robot is located inside the channel boundary, and the minimum distance between the outer contour of the mobile advertising robot and the main flow line is not less than the preset safety gap. The orientation angle of the advertising display component is determined according to the projection center point of the currently publishable unit on the main traffic line, with its display normal pointing towards the center point of the projection.

[0012] Preferably, in step S5, when the number of people located in the current publishable unit and facing the advertising display component within a preset continuous time period is not less than a preset number of people, and the average moving speed of these people is not greater than a preset viewing speed value, it is determined that the valid viewing determination condition is met, and the state is switched from the pre-publishing state to the main publishing state. When the pre-release state lasts for the preset pre-release duration and the valid viewing criteria are still not met, the current release state ends, and the mobile advertising robot exits the current lateral buffer release position and returns to the preset cruise path to continue cruise.

[0013] Preferably, in step S6, the local congestion level is determined based on the number of people and the area of ​​the monitoring area formed by extending a preset safety distance outward from the outer contour of the mobile advertising robot's body. After ending the current publishing state, control the mobile advertising robot to first retreat away from the current lateral buffer publishing position in a direction away from the main flow line, and then merge into the preset cruise path.

[0014] Preferably, the method further includes: recording the number of valid viewers, continuous viewing duration, posting duration, local congestion change value, and number of interruptions according to the candidate posting unit and time period, and generating posting records corresponding to the candidate posting unit and time period; During subsequent patrols, the preset passage interference threshold, preset attention threshold, preset dwell threshold, and candidate release order of each candidate release unit are adjusted based on the release records.

[0015] In summary, the present invention has the following main beneficial effects: This application divides the work area into multiple candidate advertising units and continuously extracts data such as pedestrian density, mainstream stability, average movement speed, percentage of people facing the advertising display component, and percentage of short-term dwell time for each candidate unit during the mobile advertising robot's patrol. This aims to establish advertising placement based on a comprehensive judgment of pedestrian flow status at the unit level. In existing solutions, advertising placement is often based on detecting a single person, identifying their attributes, or observing temporary viewing behavior. This can lead to issues such as triggering placement even when people are passing quickly or stopping to display ads despite high traffic pressure. This application, through joint judgment of publishable windows, achieves the effect of triggering advertising placement only under conditions of minimal traffic interference, relatively stable flow, and actual viewing potential. This makes the timing of advertising placement more closely match the on-site pedestrian flow status, preventing the advertising robot from blindly stopping or displaying ads in unsuitable scenarios.

[0016] By generating a lateral buffer placement position next to the main traffic flow based on the main traffic flow, passage boundaries, the size of the mobile advertising robot, and a preset safety gap, and aligning the advertising display components towards the direction of the main traffic flow, the goal of coordinating the advertising placement position with the on-site passage space is achieved. In existing solutions, even if the advertising robot can adjust its parking position according to the number of viewers or their viewing location, it often lacks specific control over the occupancy of the main passageway and the lateral safety gap, easily leading to the robot entering the main passage area and affecting the continuous passage of people. This application, by limiting the lateral buffer placement position to the side of the main traffic flow and maintaining a distance of no less than a preset safety gap between the robot's outer contour and the main traffic flow, achieves the effect of facing the advertising information while ensuring non-obstruction of the main passageway, thereby increasing the likelihood of advertising content being received while ensuring normal passage of people.

[0017] By setting pre-release and main release states, and continuously monitoring the number of effective viewers, duration of continuous viewing, changes in local congestion, and changes in the safe distance between the mobile advertising robot and surrounding people during the release process, the aim is to dynamically maintain and promptly exit the advertising release process. Simultaneously, this application records the number of effective viewers, duration of continuous viewing, release duration, changes in local congestion, and number of interruptions corresponding to candidate release units and time periods. Based on this, the thresholds and candidate release order in subsequent patrol processes are adjusted, allowing the mobile advertising robot to gradually adapt to changes in the real-world scenario during repeated runs. This not only prevents the advertising robot from continuing to linger after a weakening of viewing relationships or an increase in traffic pressure, but also enables subsequent release control to have historical data-based correction capabilities, thereby improving the stability, targeting, and continuous operational adaptability of advertising releases. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention.

[0019] Figure 2 This is a schematic diagram of the candidate publishing unit division of the mobile advertising robot in the working area according to the present invention.

[0020] Figure 3 This is a schematic diagram illustrating the principle of generating lateral buffer release bits in this invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1 refer to Figure 1-3 A method for robot advertising based on pedestrian flow characteristic analysis includes the following steps: S1: Control the mobile advertising robot to travel along a preset cruising path and divide the work area into multiple candidate publishing units; S2: Collect personnel distribution information and spatial distance information in each candidate publishing unit through the environmental perception component, and extract the personnel density, mainstream stability, average moving speed, personnel proportion facing the advertising display component, and short-term dwell proportion corresponding to each candidate publishing unit; S3: Determine the publishable window based on pedestrian density, mainstream stability, average movement speed, proportion of people facing the advertising display component, and short-term dwelling proportion. When pedestrian density is not higher than the preset traffic interference threshold, average movement speed is not higher than the preset speed threshold, proportion of people facing the advertising display component is not lower than the preset attention threshold, short-term dwelling proportion is not lower than the preset dwelling threshold, and mainstream stability meets the preset stability condition, the corresponding candidate publishable unit is determined as the current publishable unit. S4: For the currently publishable unit, generate a lateral buffer publishing position located next to the main flow line based on the main flow line, channel boundary, body size of the mobile advertising robot and preset safety gap, and control the mobile advertising robot to move to the lateral buffer publishing position so that the minimum distance between the outer contour of the mobile advertising robot body and the main flow line is not less than the preset safety gap, and the advertising display component faces the direction of the main flow. S5: After the mobile advertising robot reaches the lateral buffer publishing position, it executes the advertising publishing according to the pre-publishing state and the main publishing state. The pre-publishing state is used to output guiding information, and the main publishing state is used to output the target advertising content after the valid viewing judgment conditions are met. S6: During the ad publishing process, continuously monitor the number of effective viewers, continuous facing time, local congestion changes, and changes in the safe distance between the mobile ad robot and surrounding people. When the number of effective viewers is lower than the preset number threshold, the continuous facing time is lower than the preset time threshold, the local congestion is higher than the preset congestion threshold, or the safe distance is lower than the preset safe distance threshold, end the current publishing state, and control the mobile ad robot to exit the current lateral buffer publishing position and return to the preset cruise path to continue cruise.

[0023] This method is applied to a mobile advertising robot equipped with an environmental perception component and an advertising display component. The mobile advertising robot includes a mobile chassis, a controller, an advertising display component, an image acquisition unit, and a distance measurement unit. The mobile chassis drives the mobile advertising robot to move along a preset cruising path. The controller performs tasks such as candidate publishing unit segmentation, pedestrian flow feature extraction, publishable window determination, lateral buffer publishing position generation, advertising publishing state switching, exit control, and publishing record updating. The advertising display component outputs guidance information and target advertising content. The image acquisition unit acquires continuous images of the area in front of and to the sides of the mobile advertising robot. The distance measurement unit acquires distance data around the mobile advertising robot.

[0024] In this embodiment, the image acquisition unit is fixedly installed on the upper part of the advertising display component, with its lens facing the front of the mobile advertising robot and covering the front areas on both sides; the distance measurement unit is fixedly installed on the front of the mobile chassis to acquire distance information between the mobile advertising robot and personnel, channel boundaries, and obstacles. The controller performs time synchronization processing on the continuous image frames output by the image acquisition unit and the distance data output by the distance measurement unit, and transforms the synchronized image coordinate system with the robot's body coordinate system to form personnel distribution information and spatial distance information for subsequent candidate publishing unit status analysis.

[0025] In this embodiment, the mobile advertising robot does not rely on a single person's age, gender, clothing type, facial attributes, or a single interaction to determine whether to publish an advertisement, nor does it follow a person's movement as the method of advertising execution. Instead, it uses whether the flow of people within the candidate publishing unit meets the conditions for non-blocking publishing as the basis for initiating advertising publishing, and whether the mobile advertising robot's posture can form a lateral buffer publishing position next to the main flow line as the spatial execution prerequisite for advertising publishing. Therefore, the advertising publishing behavior is not a targeted push to a single person, but a controlled publishing based on the specific unit-level scene state.

[0026] During the deployment phase, the controller reads electronic map data of the work area and preset cruise path data. The preset cruise path consists of several sequentially connected path segments, each of which includes a path centerline, a starting point, an ending point, and a permitted passage boundary. The controller sets a front boundary and a rear boundary along the extension direction of the preset cruise path, and sets left and right boundaries in conjunction with the left and right passage boundaries of the work area, so that multiple candidate deployment units are enclosed by the front boundary, rear boundary, left boundary, and right boundary.

[0027] Specifically, the controller uses the extension direction of the cruise path as the longitudinal dividing direction, dividing adjacent path segments into multiple sequentially arranged unit segments according to the path length; then, using the left and right channel boundaries as lateral limiting boundaries, each unit segment and the channel boundaries together enclose a rectangular or approximately rectangular area, which serves as a candidate release unit. Each candidate release unit has a unique number and is mapped to its corresponding path segment, unit center position, left and right boundary positions, front and rear boundary positions, and corresponding time period.

[0028] The candidate publishing unit is not simply a pedestrian flow statistics area, but rather the smallest control object for the mobile advertising robot to determine whether to publish, where to publish, and when to exit control. In other words, the controller uses the candidate publishing unit as the unified carrier for perception, judgment, execution, and recording in all subsequent processing, thereby avoiding the disconnect between the perceived result belonging to one area and the control action acting on another area.

[0029] To ensure that each candidate release unit has publishable window determination parameters that can be directly invoked at different time periods, the controller establishes an initial threshold table during the initial deployment.

[0030] In this embodiment, the controller directs the mobile advertising robot to perform several rounds of sampling operations along a preset cruising path during pre-opening hours and low-traffic periods. During the sampling operation, the controller does not perform advertising placement, but only collects data on pedestrian density, mainstream stability, average moving speed, percentage of people facing the advertising display component, percentage of short-term dwell time, changes in local congestion, and changes in the minimum safe distance between people and the robot within each candidate placement unit. The controller categorizes and stores the above samples according to the candidate placement unit number and time period.

[0031] For each candidate release unit, the controller generates an initial threshold table based on its historical sample statistics within the corresponding time period. The initial threshold table includes at least: a preset traffic interference threshold, a preset speed threshold, a preset attention threshold, a preset dwell threshold, a preset stability condition, a preset number of people threshold, a preset duration threshold, a preset congestion threshold, a preset safe distance threshold, and a preset viewing speed value.

[0032] After the initial threshold table is generated, it is written into the task configuration file by the controller and stored according to the candidate publishing unit number and time period index. During the actual operation of the mobile advertising robot, the controller calls the threshold table of the candidate publishing unit corresponding to the current location based on the time period to perform subsequent publishing window determination, valid viewing determination, and exit control. Therefore, all the preset thresholds involved in the claims have clear data sources, generation processes, and calling paths.

[0033] As the mobile advertising robot moves along a preset cruising path, the controller performs human target detection on continuous image frames output by the image acquisition unit and correlates the positions of the same person in adjacent image frames to form a human trajectory within a preset sampling period. Simultaneously, the controller uses distance data from the distance measurement unit to spatially correct the positions of people in the images, eliminating distance errors caused by relying solely on image pixel positions.

[0034] For the For each individual, let their trajectory starting point within the preset sampling period be... The endpoint of the trajectory is Then its direction of motion is the unit vector. for:

[0035] In the formula, Indicates the first The displacement length of each person within a preset sampling period. Assume there are a total of [number] people in the current candidate release unit. If there are valid trajectories, then the mainstream stability is... for:

[0036] in, The value ranges from 0 to 1. A value closer to 1 indicates a more consistent movement direction among personnel within the current candidate release unit, and a clearer main flow line. A value closer to 0 indicates dispersed personnel directions, and the current unit is not suitable for establishing a stable unilateral release pose during this period. The controller further calculates the average position of the midpoints of each valid trajectory. :

[0037] and the direction vector Normalization as the main flow direction vector Passing by point , direction is The straight line is defined as the master flow line of the current candidate release unit.

[0038] Therefore, the main flow line is not a static centerline fixed in the electronic map, but is dynamically generated by the real-time pedestrian trajectories within the current candidate publishing unit. The purpose of this setting is to ensure that the subsequently generated lateral buffer publishing positions match the actual pedestrian flow direction, rather than matching the abstract path geometric centerline.

[0039] The controller extracts the following pedestrian flow features within the current candidate release unit: Crowd density: Let the total number of people in the current candidate release unit during the preset sampling period be... The area of ​​the current candidate release unit is Then the population density for: Crowd density reflects the occupancy level of the current candidate ad placement unit during the current time period. When the crowd density is too high, even if there are people facing the ad display component, it may not be suitable to perform ad placement due to excessive channel pressure.

[0040] Average moving speed: Let the first... The trajectory length of each person within the preset sampling period is: The trajectory duration is The average movement speed within the current candidate release unit. for:

[0041] Average movement speed is used to characterize the speed at which people pass through the current candidate display unit. When the average movement speed is high, it indicates that people mainly pass through quickly. In this case, even if there are brief actions towards the advertising display component, it is difficult to form an effective viewing relationship.

[0042] Percentage of personnel facing the advertising display component: The viewing angle range of the advertising display component is jointly limited by the preset angle range on both sides of the display normal and the effective viewing distance range. For each person in the current candidate publishing unit, the controller calculates the angle between their head orientation vector and the display normal of the advertising display component. When this angle is not greater than a preset orientation angle threshold, and the person is within the effective viewing distance range of the advertising display component, the person is determined to be facing the advertising display component. Let the number of personnel facing the advertising display component during the preset sampling period be... The percentage of personnel targeting advertising display components. for: The percentage of people facing the advertising display component here is not an abstract attention parameter, but a specific statistic obtained based on head orientation and distance constraints.

[0043] Short-stay percentage: The controller calculates the dwell time of a single person within the current candidate release unit based on the timestamps of the person's trajectory crossing the front and back boundaries of the current candidate release unit. Let the total number of people entering the current candidate release unit within the preset sampling period be... The number of people whose stay duration is not less than the preset stay duration threshold is The proportion of short stays for: The short-term dwell time percentage is used to reflect whether there is a certain proportion of non-fast-moving users in the current candidate release unit, thus providing conditions for the advertising display component to form a viewing window.

[0044] The controller performs a publishable window determination for each candidate publishable unit. The controller only determines a candidate publishable unit as meeting the publishable window condition if the following conditions are simultaneously met: Crowd density Not exceeding the preset traffic interference threshold; Average movement speed Not exceeding the preset speed threshold; Mainstream stability The preset stability conditions are met; Percentage of personnel working on advertising display components Not lower than the preset attention threshold; Short stay percentage Not lower than the preset dwell time threshold.

[0045] The above five conditions are not simply a combination of each other, but rather correspond to five prerequisites required for the advertising launch process: Pedestrian density constraints are used to limit the degree of channel occupancy; Average movement speed constraints are used to exclude high-speed traversal scenarios; Mainstream stability constraints are used to ensure the generation of stable lateral buffer release bits; The percentage of people viewing the advertising display components is used to reflect whether there is a basis for forming a viewing relationship during the current time period; The short-stay percentage is used to reflect whether there is enough dwell time within the ad unit to accommodate the advertising content.

[0046] When multiple candidate publishable units simultaneously meet the publishable window criteria, the controller compares them in the following order: First, compare the percentage of people facing the ad display component in multiple candidate publishing units. If one candidate publishing unit has the largest percentage of people facing the ad display component, it is directly identified as the current publishing unit. If two or more candidate publishing units have the same percentage, continue to compare the percentage of people who stay in these tied units. If one unit has the largest percentage of people who stay in these tied units, it is directly identified. If there are still ties, continue to compare the flow density of the tied units and select the candidate publishing unit with the lowest flow density. If there are still ties, continue to compare the path distance between these tied units and the current location of the mobile advertising robot and select the candidate publishing unit with the shortest path distance. If the path distance is still the same, select the candidate publishing unit with the smaller number as the current publishing unit.

[0047] With the above-mentioned terminating comparison rule, the selection logic when multiple candidate release units simultaneously meet the conditions is deterministic, and there will be no problem of no clear landing point during the execution process.

[0048] Once the current publishable unit is determined, the controller generates a lateral buffer publishing position located next to the main flow line based on the main flow line, channel boundaries, the body size of the mobile advertising robot, and preset safety gaps, instead of parking the mobile advertising robot directly at the geometric center of the current publishable unit or any location with a large number of viewers.

[0049] Let the maximum body width of the mobile advertising robot be The preset safety gap is The current publishable unit's main workflow is The controller calculates the normal distances from the main flow line to the left and right lane boundaries, respectively. and When one side satisfies: In the formula, If the left or right side is selected, that side is considered a candidate side for forming a lateral buffer placement position. If both the left and right sides meet the conditions, the controller further compares the path lengths of the mobile advertising robot from its current position to the target positions on the left and right sides, and selects the side with the shorter path length as the actual stopping side. After determining the actual stopping side, the controller takes the projection center point of the current publishable unit on the main flow line as... Take the unit normal vector pointing to the actual parking side as Then the center point of the lateral buffer release position for:

[0050] The lateral buffer placement position obtained from the above formula satisfies the following conditions: First, the outer contour of the mobile advertising robot is located inside the channel boundary; second, the minimum distance between the outer contour of the mobile advertising robot and the main flow line is not less than the preset safety gap; third, the mobile advertising robot is located beside the main flow line rather than within the main flow zone. The orientation angle of the advertising display component is determined according to its display normal pointing towards the direction of the main flow. Since the main flow direction vector is... The main flow of people is from which The controller aligns the display normal of the advertising display component with... Alignment ensures that people entering the currently publishable unit can see the displayed content before approaching the mobile advertising robot, while the mobile advertising robot itself remains to the side of the main traffic flow, without obstructing the main passageway.

[0051] After solving for the center point and orientation angle of the lateral buffer publishing position, the controller plans a collision-free driving path from the current position to the lateral buffer publishing position and controls the mobile advertising robot to move along the collision-free driving path to the lateral buffer publishing position.

[0052] Once the mobile ad robot reaches the lateral buffer placement position, the controller first puts the ad display component into a pre-release state. In the pre-release state, the ad display component outputs guidance information, which is a brief prompt screen used to establish an initial viewing relationship, without immediately playing the full target ad content. The purpose of this setup is to first determine, with low placement cost, whether the currently publishable unit truly has a stable viewing relationship before deciding whether to enter the main release state.

[0053] In this embodiment, the preset continuous time period consists of a time window covered by multiple consecutive frames of images. Within each continuous time period, the controller counts the number of people in a set that meets the condition of being "located within the current publishable unit and facing the advertising display component". and the average movement speed of the group of people. .in, The statistical objects are limited to the set of personnel who meet the above conditions, and do not include all personnel in the current candidate release unit.

[0054] The controller determines that the valid viewing criteria are met and switches from the pre-release state to the main release state when all of the following conditions are met: The number of people located within the currently publishable unit and facing the ad display component within a preset continuous time period. Not less than the preset number of people; The average movement speed of this group of people No more than the preset viewing speed value.

[0055] In the main publishing state, the ad display component outputs the target ad content. This target ad content is pre-bound based on the patrol task file or time-segment task file and does not switch based on the identity attributes of individual users. Therefore, entering the main publishing state is based on the formation of effective viewing relationships at the unit level, rather than the matching results of individual user profiles.

[0056] If the pre-release state lasts for the preset pre-release duration and the controller still fails to determine that the valid viewing criteria are met, the controller will end the current release state and control the mobile advertising robot to exit the current lateral buffer release position and return to the preset cruise path to continue cruise. Therefore, the mobile advertising robot will not linger in positions in scenarios where there is no basis for viewing.

[0057] In the main publishing state, the controller continuously monitors the number of effective viewers, the duration of continuous viewing, changes in local congestion, and changes in the safe distance between the mobile advertising robot and surrounding people.

[0058] Continuous orientation duration refers to the continuous duration during which a person simultaneously meets the following conditions in consecutive image frames: Located within the currently publishable unit; Located within the effective viewing distance of the advertising display component; The angle between the head orientation and the display normal of the advertising display component is not greater than the preset orientation angle threshold.

[0059] The controller accumulates the continuous facing time for each person and counts the corresponding time lengths of people who meet the continuous facing condition within the current time period.

[0060] The controller extends a preset safety distance outward from the outer contour of the mobile advertising robot to form a localized congestion monitoring area. Let the number of people within the monitoring area be... The monitored area is Then the local congestion for: Local congestion is used to characterize whether excessive crowding occurs in the near-field area around a mobile advertising robot after it publishes an advertisement, thus serving as an important basis for exit control.

[0061] The controller terminates the current release state when any of the following conditions are met: The number of valid viewers is lower than the preset threshold. The duration of continuous exposure is below the preset duration threshold; The local congestion level is higher than the preset congestion threshold; The safe distance between the mobile advertising robot and people in the vicinity is less than the preset safe distance threshold.

[0062] After ending the current publishing state, the controller does not immediately turn around and enter the cruise path in the current pose. Instead, it first controls the mobile advertising robot to retreat from the current lateral buffer publishing position in a direction away from the main traffic flow. Then, it selects the first reachable path point in the current path segment that is ahead of the robot's current position and has the shortest path distance as the re-entry point. Subsequently, the controller generates a collision-free transition path from the current retreat position to the re-entry point and controls the mobile advertising robot to return to the preset cruise path along the collision-free transition path to continue cruising.

[0063] By adopting an exit method of first leaving and then rejoining, the mobile advertising robot is prevented from directly cutting across the main traffic flow area, and the exit action and cruise recovery action have a continuous and clear geometric connection.

[0064] In order to enable subsequent cruise control to use historical execution results to correct the current task, the controller records the following data according to the candidate release unit and time period: number of effective viewers, duration of continuous viewing, duration of release, local congestion change value and number of interruptions, and forms a release record corresponding to the candidate release unit and time period.

[0065] The controller performs corrections based on the published records according to the following rules: When the number of interruptions of the same candidate publishing unit reaches the first correction condition in the same time period of multiple consecutive cruise cycles, the preset passage interference threshold of the candidate publishing unit in that time period is reduced. When the number of effective viewers for the same candidate release unit is consistently lower than the second correction condition within the same time period of multiple consecutive cruise cycles, the preset attention threshold for that candidate release unit within that time period is increased. When the short-term dwell percentage of the same candidate release unit is consistently lower than the third correction condition in the same time period of multiple consecutive cruise cycles, the preset dwell threshold of the candidate release unit in that time period is increased. When multiple candidate release units exist in the same path segment, the controller updates the candidate release order of the multiple candidate release units in the path segment in descending order of the number of successful main releases, descending order of the average duration of the main release, and ascending order of the number of aborts.

[0066] When the controller calculates the lateral buffer publishing position based on the main flow line, channel boundary, body size of the mobile advertising robot, and preset safety gap of the current candidate publishing unit, if neither the left nor the right side satisfies the condition that the normal distance from the main flow line to the channel boundary is not less than the sum of the maximum body width and the preset safety gap, the controller marks the candidate publishing unit as an unpublishable unit for the current time period.

[0067] For unpublishable units, the controller keeps the mobile advertising robot passing along the preset cruise path, without executing the pre-publishing state or the main publishing state. If a candidate publishing unit is marked as an unpublishable unit for multiple consecutive cruise cycles, or exits due to local congestion exceeding the threshold for multiple consecutive cruise cycles, the controller sets the candidate publishing unit as a low-priority candidate publishing unit for the corresponding time period, and tries other candidate publishing units in the same path segment first in the next cruise cycle.

[0068] Example 2 This embodiment is another specific operating mode based on Embodiment 1, mainly used to illustrate the collaborative control process of the mobile advertising robot when there are multiple consecutive candidate publishing units in the same path segment.

[0069] When the mobile advertising robot runs along a certain path segment, the controller first extracts pedestrian flow features and determines the publishable window for each of the multiple candidate publishing units arranged sequentially within that path segment. If the previous candidate publishing unit meets the publishable window determination conditions and the current path length is sufficient to form a lateral buffer publishing position, the controller prioritizes publishing in the previous candidate publishing unit; if the previous candidate publishing unit meets the pedestrian flow feature conditions, but cannot generate a lateral buffer publishing position that meets the preset safety gap requirements due to insufficient space on the side of the channel, the controller does not publish the advertisement in that unit, but continues to detect the next candidate publishing unit.

[0070] When a subsequent candidate release unit meets the release window determination criteria and can form a lateral buffer release position, the controller executes the entry, pre-release, main release, exit, and backtracking processes in Embodiment 1 within the subsequent candidate release unit. In this way, the candidate release units are not independent, unrelated regions, but rather form a sequential release candidate sequence within the same path segment. The controller always controls the release in the order of first satisfying the window, then satisfying the lateral space, and finally executing the release, thereby avoiding the situation where a release is forcibly executed simply because a certain unit has good pedestrian flow characteristics.

[0071] During operation, the controller first establishes multiple candidate advertising units based on the preset cruise path and the channel boundaries of the work area. As the mobile advertising robot travels along the preset cruise path, it continuously collects personnel distribution and spatial distance information within each candidate advertising unit through environmental perception components. After synchronously processing the collected continuous images and distance data, the controller extracts the personnel density, mainstream stability, average movement speed, percentage of personnel facing the advertising display component, and percentage of short-term dwellings corresponding to the current candidate advertising unit. Based on this, it determines whether the current candidate advertising unit meets the criteria for an advertising window. Only when the current candidate advertising unit simultaneously exhibits minimal traffic interference, relatively concentrated personnel movement directions, low passing speed, and a certain number of people facing the advertising display component and short-term dwellings, will the controller designate the candidate advertising unit as the current advertising unit. This ensures that the advertising publishing action is no longer based on individual personnel identification or a single close-range contact, but rather on the unit-level personnel flow status meeting the publishing conditions.

[0072] After the current publishable unit is determined, the controller further generates a lateral buffer publishing position located beside the main flow line based on the main flow line, channel boundary, body size of the mobile advertising robot, and preset safety gap of the current publishable unit. The controller then controls the mobile advertising robot to move along a collision-free entry path to the lateral buffer publishing position, ensuring that the outer contour of the mobile advertising robot remains inside the channel boundary and maintains a distance of not less than the preset safety gap between it and the main flow line. Simultaneously, the advertising display component faces the direction of the main flow. Therefore, when performing advertising, the mobile advertising robot does not occupy the main flow area, but rather cooperates with the flow of people in the current candidate publishing unit by parking to the side and displaying the advertisement, allowing people entering the current publishable unit to receive advertising information without being obstructed.

[0073] When the mobile advertising robot reaches the lateral buffer publishing position, the controller first controls the advertising display component to enter the pre-publishing state to output guidance information and establish an initial viewing relationship. Within a preset continuous time period, if the number of people in the current publishable unit facing the advertising display component reaches a preset number, and the average movement speed of these people is not greater than a preset viewing speed value, the controller determines that the valid viewing judgment condition is met and controls the advertising display component to switch from the pre-publishing state to the main publishing state, outputting the target advertising content. If the valid viewing judgment condition is not met within the preset pre-publishing time, it means that although the current publishable unit has the initial publishing conditions, a stable viewing relationship has not yet been formed. At this time, the controller ends the current publishing state and controls the mobile advertising robot to exit the current lateral buffer publishing position and return to the preset cruising path to continue cruising.

[0074] In the main publishing state, the controller continuously monitors the number of effective viewers, continuous facing duration, changes in local congestion, and changes in the safe distance between the mobile advertising robot and surrounding people. When the number of effective viewers drops below a preset threshold, or the continuous facing duration is less than a preset duration threshold, or the local congestion is greater than a preset congestion threshold, or the safe distance between the mobile advertising robot and surrounding people is less than a preset safe distance threshold, the controller determines that the current publishing state should not be maintained. It then controls the mobile advertising robot to retreat from the current lateral buffer publishing position in a direction away from the main traffic flow, and then selects the first reachable path point ahead as the re-entry point, returning to the preset cruising path along a collision-free transition path to continue cruising. In this way, the exit process of the mobile advertising robot after publishing also avoids the main traffic flow, preventing secondary interference to on-site traffic caused by directly switching back to the cruising path.

[0075] The controller also records the number of effective viewers, continuous viewing duration, posting duration, local congestion changes, and number of interruptions according to the candidate posting units and time periods, forming corresponding posting records. During subsequent patrols, the controller adjusts the preset traffic interference threshold, preset attention threshold, preset dwell threshold, and candidate posting order for each candidate posting unit within the corresponding time period based on the posting records. This allows the mobile advertising robot to select candidate posting units within the same path at different time periods to better reflect the actual pedestrian flow. Thus, this application forms a closed-loop control process based on candidate posting units, encompassing perception, judgment, posting, exit, and correction. This enables the timing of advertising posting to coordinate with the spatial pose of the mobile advertising robot, achieving effective advertising information posting while ensuring traffic safety and continuity.

[0076] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for robot advertising based on pedestrian flow characteristic analysis, characterized in that, Includes the following steps: S1: Control the mobile advertising robot to travel along a preset cruising path and divide the work area into multiple candidate publishing units; S2: Collect personnel distribution information and spatial distance information in each candidate publishing unit through the environmental perception component, and extract the personnel density, mainstream stability, average moving speed, personnel proportion facing the advertising display component, and short-term dwell proportion corresponding to each candidate publishing unit; S3: Determine the publishable window based on pedestrian density, mainstream stability, average movement speed, proportion of people facing the advertising display component, and short-term dwelling proportion. When pedestrian density is not higher than the preset traffic interference threshold, average movement speed is not higher than the preset speed threshold, proportion of people facing the advertising display component is not lower than the preset attention threshold, short-term dwelling proportion is not lower than the preset dwelling threshold, and mainstream stability meets the preset stability condition, the corresponding candidate publishable unit is determined as the current publishable unit. S4: For the currently publishable unit, generate a lateral buffer publishing position located next to the main flow line based on the main flow line, channel boundary, body size of the mobile advertising robot and preset safety gap, and control the mobile advertising robot to move to the lateral buffer publishing position so that the minimum distance between the outer contour of the mobile advertising robot body and the main flow line is not less than the preset safety gap, and the advertising display component faces the direction of the main flow. S5: After the mobile advertising robot reaches the lateral buffer publishing position, it executes the advertising publishing according to the pre-publishing state and the main publishing state. The pre-publishing state is used to output guiding information, and the main publishing state is used to output the target advertising content after the valid viewing judgment conditions are met. S6: During the ad publishing process, continuously monitor the number of effective viewers, continuous facing time, local congestion changes, and changes in the safe distance between the mobile ad robot and surrounding people. When the number of effective viewers is lower than the preset number threshold, the continuous facing time is lower than the preset time threshold, the local congestion is higher than the preset congestion threshold, or the safe distance is lower than the preset safe distance threshold, end the current publishing state, and control the mobile ad robot to exit the current lateral buffer publishing position and return to the preset cruise path to continue cruise.

2. The robot advertising method based on pedestrian flow characteristic analysis according to claim 1, characterized in that, The environmental perception component includes an image acquisition unit and a distance measurement unit; The image acquisition unit is used to acquire images of people within the candidate publishing unit, and the distance measurement unit is used to acquire distance data within the candidate publishing unit. The personnel distribution information is obtained from personnel images, and the spatial distance information is obtained from distance data.

3. The robot advertising method based on pedestrian flow characteristic analysis according to claim 2, characterized in that, In step S1, a front boundary and a rear boundary are set along the extension direction of the preset cruise path, and a left and right boundary is set in combination with the channel boundary of the work area, so that multiple candidate release units are enclosed by the front boundary, the rear boundary and the left and right boundaries; adjacent candidate release units are arranged continuously in the extension direction of the preset cruise path.

4. The robot advertising method based on pedestrian flow characteristic analysis according to claim 3, characterized in that, In step S2, the mainstream stability is determined based on the consistency of the direction of movement trajectories of each person within a preset sampling period; The average moving speed is determined based on the displacement and time interval of each person's movement trajectory within a preset sampling period; The proportion of people facing the advertising display component is determined based on the proportion of the number of people whose heads or faces fall within the observation angle range of the advertising display component during the preset sampling period to the total number of people in the current candidate publishing unit. The short-stay percentage is determined based on the proportion of the number of people who enter the current candidate release unit within a preset sampling period and whose stay time is not less than a preset stay time threshold to the total number of people entering the current candidate release unit.

5. The robot advertising method based on pedestrian flow characteristic analysis according to claim 4, characterized in that, In step S3, when multiple candidate publishing units meet the criteria for publishing windows, the current publishing unit is determined by comparing them in the following order: the proportion of people facing the advertising display component from high to low, the proportion of short-term stay from high to low, the flow density from low to high, and the distance from the current location of the mobile advertising robot from near to far.

6. The robot advertising method based on pedestrian flow characteristic analysis according to claim 5, characterized in that, In step S4, the lateral buffer release position is offset from the main flow line to the side of the channel according to the normal direction of the main flow line. The offset endpoint satisfies that the outer contour of the mobile advertising robot is located inside the channel boundary, and the minimum distance between the outer contour of the mobile advertising robot and the main flow line is not less than the preset safety gap. The orientation angle of the advertising display component is determined according to the projection center point of the currently publishable unit on the main traffic line, with its display normal pointing towards the center point of the projection.

7. The robot advertising method based on pedestrian flow characteristic analysis according to claim 6, characterized in that, In step S5, when the number of people in the current publishable unit facing the advertising display component is not less than the preset number of people within a preset continuous time period, and the average moving speed of these people is not greater than the preset viewing speed value, it is determined that the valid viewing determination condition is met, and the state is switched from the pre-publishing state to the main publishing state. When the pre-release state lasts for the preset pre-release duration and the valid viewing criteria are still not met, the current release state ends, and the mobile advertising robot exits the current lateral buffer release position and returns to the preset cruise path to continue cruise.

8. The robot advertising method based on pedestrian flow characteristic analysis according to claim 7, characterized in that, In step S6, the local congestion level is determined based on the number of people and the area of ​​the monitoring area formed by extending a preset safety distance outward from the outer contour of the mobile advertising robot's body. After ending the current publishing state, control the mobile advertising robot to first retreat away from the current lateral buffer publishing position in a direction away from the main flow line, and then merge into the preset cruise path.

9. A robot advertising method based on pedestrian flow characteristic analysis according to claim 8, characterized in that, The method further includes: recording the number of valid viewers, duration of continuous viewing, duration of publication, local congestion change value and number of interruptions according to the candidate publication unit and time period, and generating publication records corresponding to the candidate publication unit and time period; During subsequent patrols, the preset passage interference threshold, preset attention threshold, preset dwell threshold, and candidate release order of each candidate release unit are adjusted based on the release records.

Images