Method and device for determining the behavior of persons
Fuzzy logic-based behavioral categorization addresses the limitations of fixed zone methods by accurately predicting passenger behavior and traffic routes in complex airport environments, enhancing operational efficiency.
Patent Information
- Application Number
- DE102018100220
- Authority / Receiving Office
- DE · DE
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2018-01-08
- Publication Date
- 2024-11-07
- Estimated Expiration
- 2038-01-08
AI Technical Summary
Existing methods for determining passenger behavior in areas like airports are limited by the use of fixed, sharply defined zones, leading to misinterpretations and inadequate representation of passenger behavior, particularly in complex environments with multiple sub-areas and transition zones.
A method utilizing fuzzy logic to assign behavioral categories based on movement characteristics, such as average walking speed, stop-walk ratio, and path variability, allowing for more accurate determination of passenger behavior by assigning discrete values to blur designations and membership functions, enabling reliable prediction of passenger volumes and traffic routes.
Enables reliable prediction of passenger behavior and traffic routes by accurately categorizing behavior using fuzzy logic, even in large, complex airport environments with multiple sub-areas, improving operational planning and capacity management.
Smart Images

Figure 00000005_0000 
Figure 00000006_0000 
Figure 00000007_0000
Abstract
Description
[0001] The invention relates to a method and a device for determining the behavior of at least one person within a defined area, wherein the defined area comprises a plurality of interaction possibilities.
[0002] Within confined spaces, such as airports, shopping centers, or other areas where large numbers of people regularly move, the people present there often have very different intentions. Knowledge of this can be useful for defining the traffic routes within these confined spaces and thus improving the overall concept of the interaction options arranged within the space.
[0003] Airports generally strive for the most accurate knowledge possible about future passenger volumes. Long-term forecasts safeguard airport operators' investments, while medium-term forecasts support capacity planning, and short-term forecasts help respond to operational disruptions. To further improve airport operations, commercial passenger flow forecasting programs are used.
[0004] In practice, to determine the need for various service points, only the external state of passengers, such as position and speed, is considered. Okamoto, K. et al., "Classification of Pedestrian Behavior in a Shopping Mall based on LRF and Camera Observations," MVA 2011 IAPR Conference on Machine Vision Applications, June 13-15, Nara, Japan, 2011, describes how to determine the internal state of visitors to a shopping mall as part of an investigation. Visitor behavior was categorized. A weakness is the use of fixed, sharply defined areas, which frequently leads to misinterpretations.
[0005] From “Classifying human activity patterns from smartphone collected GPS data: A fuzzy classification and aggregation approach” it is known to classify human activities using fuzz logic based on GPS data.
[0006] It is therefore an object of the present invention to provide an improved method and an improved device with which the behavior of at least one person within a defined area can be predicted more reliably.
[0007] The object is achieved according to the invention with the method according to claim 1 and the device according to claim 6.
[0008] According to claim 1, a method is proposed for determining the behavior of at least one person within a demarcated area, wherein the demarcated area comprises a plurality of interaction options. According to the invention, such a demarcated area can be an airport complex consisting of a plurality of sub-areas. The individual sub-areas are connected by transitions where passengers are guided from one sub-area to the next (for example, security or check-in).
[0009] The behavior of the people is determined by assigning a behavioral category from a plurality of different behavioral categories using fuzzy logic. For this purpose, discrete values for various movement characteristics are first determined from the person's movements within the defined area, as recorded by a recording system. Consequently, several different movement characteristics exist, for each of which one or more values are determined based on the recording system. The values of the individual movement characteristics can be taken natively from the recording system or derived as ratios or equivalents from the recorded values. Over time, a multitude of such discrete values are thus determined for a movement characteristic.
[0010] Based on fuzzy logic, each motion characteristic is assigned a plurality of blur labels, with each blur label having at least one membership function that defines the degree of membership of a discrete value to the respective blur label. For example, if a motion characteristic has three blur labels, such as small, medium, and large, each blur label is assigned a membership function. By inserting a discrete value, resulting from the detection of the movements using the detection system, into each of the membership functions, a degree of membership of this discrete value to the respective blur label results for each blur label. In the simplest case, the blur label whose membership function results in the highest degree of membership can then be selected for the specific discrete value.
[0011] For each of the motion characteristics, a specific blur label is determined by inserting the discrete value(s) for the respective motion characteristic into the membership functions and determining the degree of membership of the discrete value to the respective blur label. Following this step, a specific blur label is thus available for each motion characteristic underlying the method, which corresponds to the discrete value for the respective motion characteristic. Based on these specific blur labels for the motion characteristics thus determined, the corresponding behavioral category is then determined, for example, by selecting it from a set of behavioral categories for which the corresponding combination of blur labels is defined.
[0012] This makes it possible to use fuzzy logic to select appropriate behavioral categories and thus determine the behavior of people within a defined area. Based on fuzzy logic, appropriate levels of uncertainty are defined to better represent people's behavior. By determining people's behavior, predictions can be made about passenger volumes and traffic routes in such defined areas, such as airports.
[0013] It was recognized that, especially in the area of airports, meaningful grouping is possible using fuzzy logic and the present concept of selecting behavioral categories, even if an airport area is usually defined by a multitude of sub-areas defined by corresponding transition zones. These transition zones generally influence people's behavior, which can have a more or less negative impact on determining the behavioral category. It was recognized that a corresponding behavioral category can nevertheless be reliably determined.
[0014] According to the invention, the movement characteristics are in particular the average walking speed, the stop-walk ratio, in particular within a predetermined period of time, and / or the variability of the walking path.
[0015] According to the invention, each of the motion properties is assigned at least the following blur designations: low, medium, and high. Each motion property thus has such a blur designation, with each of these blur designations being assigned a corresponding membership function that defines the degree to which a specific value belongs to the respective blur designation.
[0016] According to the invention, one of the following behavior categories is selected based on the determined concrete blur designations of the motion properties: - goal-directed, where it is assumed that the person has chosen a specific interaction option that he or she wants to access and knows the way there; - path-searching, where it is assumed that the person wants to seek out an interaction option but has not yet decided which interaction option he or she wants to seek out or that the person does not know the way to an interaction option that has already been selected; - aimless, where it is assumed that the person does not pursue a specific goal.
[0017] According to the invention, the behavior categories are selected when a specific combination of the previously determined uncertainty labels of the individual movement characteristics is present. Accordingly, the behavior category "goal-directed" is selected when the average walking speed is high, the stop-walk ratio is low, and the path variability is also low. In contrast, the behavior category "path-searching" is selected when the average walking speed is medium, the stop-walk ratio is medium, and the path variability is also medium. However, the behavior category "aimless" is selected when the average walking speed is low, the stop-walk ratio is high, and the path variability is also high.
[0018] Preferably, all combinations of possible blur labels for the individual movement characteristics are covered and assigned a corresponding behavioral category. Thus, it is conceivable that, in the simplest case, all combinations not yet mentioned above are classified into the behavioral category "aimless" if a corresponding behavioral category cannot be clearly derived from the respective combination of blur labels for the movement characteristics. However, it is also conceivable that the remaining combinations of blur labels are approximately classified into the other behavioral categories, so that, for example, the behavioral category "goal-directed" is assumed even if the average walking speed is medium, the stop-walk ratio is low, and the path variability is also low.
[0019] In a further advantageous embodiment, the variability of the path is determined using a movement model, whereby, in particular, an angle or radius of a walking path is determined, and based on this, the degree of path variability can then be determined. The higher the angle or radius of the path, the greater the path's variability. The number of changes of direction can also be an indicator here.
[0020] The existing method can also be applied to a model motion. In this case, the radius of the circular motion is specified, with the variability decreasing with larger radiuses.
[0021] The object is also achieved with the device for determining the behavior of at least one person within a delimited area according to claim 3, wherein the delimited area comprises a plurality of interaction options with which the at least one person can interact.
[0022] The invention is explained in more detail by way of example with reference to the accompanying figures. They show: Fig. 1 - Schematic representation of the device according to the invention for detecting the behavior of persons; Fig. 2a-2c - Examples for determining the behavior category.
[0023] Fig. Figure 1 schematically shows the device 1, in which a defined area 3 is recorded using a camera system 2. Using the camera data from camera 2, the movements of individual persons within the defined area 3 can then be determined using a detection system 4, which includes camera 2. The detection system 4 forwards its results to an evaluation unit 5, which then performs the appropriate evaluation to determine the behavioral categories.
[0024] Fig. Figure 2a shows a diagram for determining the behavior category "goal-directed." On the left, an example of a person's path is shown, characterized in particular by a relatively straight path with few curves. On the right, the uncertainty labels "low," "medium," and "high" are shown for the respective movement characteristics: average walking speed, stop-walk ratio, and path variability, including the specific, discrete value in which the respective diagrams are plotted. It can be seen that, using the membership function, the average walking speed can be considered high, while the stop-walk ratio and path variability have the uncertainty label "low." In this combination, the behavior category "goal-directed" can be assumed.
[0025] For this purpose, Fig. Figure 2b illustrates the recognition of the behavior category "path-searching." The average walking speed is defined by the blur label "medium," as are the stop-walk ratio and path variability.
[0026] Fig. Figure 2c finally shows the determination of the behavioral category “aimless,” where the mean walking speed is low, while the stop-walk ratio and the variability of the path are determined to be high. List of reference symbols 1 device 2 cameras 3 demarcated area 4 Recording system 5 Evaluation unit
Claims
[1] A method for determining the behavior of at least one person within a defined area (3), wherein the defined area (3) comprises a plurality of interaction possibilities with which the at least one person can interact, and wherein the behavior of the person is determined by assigning a behavior category from a plurality of different behavior categories by means of fuzzy logic, comprising the steps: - Determining discrete values relating to various movement characteristics from the movements of the persons within the defined area (3) recorded by a detection system (4), - where each motion property has a plurality of blur labels, each of which indicates the degree of membership of a discrete value to the respective blur label via a membership function, - determining a specific blur designation for each movement property as a function of the determined discrete values of the respective person and the membership functions of the blur designations of the movement properties by means of an evaluation unit (5), and - Determining a behavior category based on the determined concrete blur designations of the movement characteristics by means of the evaluation unit (5), - where the movement characteristics are the average walking speed, the stop-walk ratio and the variability of the walking path, and each of the movement characteristics is assigned at least the following uncertainty labels: low, medium and high, - based on the determined specific blur labels of the movement properties, a behavioral category is selected from the following: ◯ goal-directed, where it is assumed that the person has chosen a specific interaction option that they want to access and knows the route there; ◯ path-searching, where it is assumed that the person wants to seek out an interaction option but has not yet decided which interaction option they want to seek out, or that the person does not know the way to an interaction option they have already chosen; ◯ aimless, which assumes that the person does not have a specific goal; - where the behavior category ◯ is selected purposefully when the average walking speed is high, the stop-walk ratio is low and the path variability is low; ◯ path-searching is selected when the average walking speed is medium, the stop-walk ratio is medium and the path variability is medium; ◯ is selected randomly when the average walking speed is low, the stop-walk ratio is high, and the path variability is high. [2] Method according to one of the preceding claims, characterized by that the variability of the path is determined using a movement model. [3] Device (1) for determining the behavior of at least one person within a defined area (3), wherein the defined area (3) comprises a plurality of interaction possibilities with which the at least one person can interact, characterized by that the device (1) is designed to carry out the method according to one of the preceding claims and comprises at least one detection system (4) and one evaluation unit (5).
