Vehicle behavior recognition methods, devices, electronic equipment, vehicles and media

Abstract

This application provides a vehicle behavior recognition method, device, electronic device, vehicle, and medium. The method involves: determining basic behavioral units of the vehicle based on its basic characteristics. Each basic behavioral unit includes a basic behavior identifier and a behavior start time. The basic characteristics include at least one of obstacle features, vehicle features, traffic features, environmental features, and geographical location features. The basic behavioral units are matched with at least one complex behavior. If multiple sets of basic behavioral units of the vehicle successfully match a first complex behavior among the at least one complex behavior, the labels of the multiple sets of basic behavioral units are set as the label of the first complex behavior. This invention avoids the limitations of a single data source by using basic characteristics, covering multiple information dimensions and providing more comprehensive information support. Matching multiple sets of basic behavioral units with the first complex behavior achieves accurate recognition of complex behaviors, improving the accuracy of vehicle behavior recognition.

Description

Technical Field

[0001] This application relates to the field of vehicle-assisted driving, and more particularly to a vehicle behavior recognition method, device, electronic device, vehicle, and medium. Background Technology

[0002] Existing vehicle behavior recognition methods, on the one hand, rely on single or a few data types for analysis, resulting in an inability to fully capture the complexity and diversity of vehicle behavior; on the other hand, their recognition accuracy is insufficient when faced with high-dimensional, heterogeneous, and spatiotemporally characteristic multi-source data. Summary of the Invention

[0003] This application provides a vehicle behavior recognition method, device, electronic device, vehicle, and medium, aiming to improve the technical problems of existing technologies that cannot fully capture the complexity and diversity of vehicle behavior and have insufficient recognition accuracy.

[0004] A vehicle behavior recognition method, comprising: The basic behavioral units of a vehicle are determined based on its basic characteristics. Each basic behavioral unit includes a basic behavioral identifier and a start time of the behavior. The basic characteristics include at least one of obstacle characteristics, vehicle characteristics, traffic characteristics, environmental characteristics, and geographical location characteristics. The basic behavior unit of the vehicle is matched with at least one complex behavior, wherein each complex behavior contains multiple basic behaviors and time information of each basic behavior; If the multiple sets of basic behavioral units of the vehicle successfully match the first complex behavior in the at least one complex behavior, then the labels of the multiple sets of basic behavioral units are set as the labels of the first complex behavior.

[0005] In this embodiment, the vehicle behavior recognition method determines the basic behavioral units of the vehicle based on its fundamental characteristics. This avoids the limitations of a single data source, covers multiple information dimensions, and provides more comprehensive information support for behavior recognition, thereby capturing the complexity and diversity of vehicle behavior. The basic behavioral units of the vehicle are matched with at least one complex behavior. When a match is successful, the labels of multiple sets of basic behavioral units are set as the label of the first complex behavior. This achieves the matching of basic behavioral units with complex behaviors and the efficient fusion of multi-dimensional data, thereby enabling the determination of behavior labels and the accurate recognition of complex behaviors, thus improving the accuracy of vehicle behavior recognition.

[0006] Furthermore, the preset behavior state rules include at least one set of preset rule conditions and a maximum time constraint; The process of determining the basic behavioral units of a vehicle based on its fundamental characteristics, each basic behavioral unit including a basic behavioral identifier and a start time of the behavioral occurrence, includes: Based on all the aforementioned basic features, a state transition sequence corresponding to each of the aforementioned basic behaviors is determined; the state transition sequence includes at least one behavioral state; From the preset behavior state rules, obtain the preset rule conditions and maximum time constraints corresponding to each behavior state; the preset behavior state rules include at least one preset behavior identifier; When all behavioral states in the state transition sequence satisfy the corresponding preset rule conditions and maximum time constraints, the preset behavior identifier corresponding to the state transition sequence is determined as the basic behavior identifier of the basic behavior unit. The start timestamp of the first behavioral state and the end timestamp of the last behavioral state in the state transition sequence are determined as the start time of the behavior of the basic behavioral unit. In this embodiment, the state transition sequence of basic behaviors is determined through basic features, thereby enabling the determination of each behavior state. When all behavior states in the state transition sequence satisfy the corresponding preset rule conditions and maximum time constraints, the basic behavior identifier and behavior start time of each basic behavior are determined, thus achieving accurate identification of basic behaviors and improving the identification accuracy of basic behaviors.

[0007] Furthermore, the matching of the vehicle's basic behavioral units with at least one complex behavior includes: The basic behavior identifiers and behavior start times of multiple sets of basic behavior units of the vehicle are sorted to obtain a basic behavior time series. The dynamic programming algorithm is used to determine whether the time series of the basic behaviors matches the preset time series of the complex behaviors; the preset time series of behaviors includes multiple basic behaviors and the start and end times of each basic behavior. When the basic behavior time series matches the preset behavior time series, rule verification is performed on the basic behavior time series to obtain the rule verification result; When the rule verification result indicates that the basic behavior time series verification is passed, it is determined that the multiple sets of basic behavior units of the vehicle are successfully matched with the first complex behavior in the at least one complex behavior.

[0008] In this embodiment, the basic behavior time series is determined by sorting the basic behavior identifiers and the start time of the behavior occurrence, thereby enabling the sorting of the basic behavior occurrence events. A dynamic programming algorithm is used to confirm whether the basic behavior time series matches a preset behavior time series corresponding to the complex behavior, achieving complex behavior matching of the basic behavior time series. Furthermore, rule verification of the basic behavior time series reduces interference from random factors through a dual verification mechanism and solves the problem of single-logic misjudgment, thus improving the recognition accuracy of complex vehicle behaviors.

[0009] Further, the step of determining whether the basic behavior time series matches the preset behavior time series corresponding to the complex behavior using a dynamic programming algorithm includes: A distance matrix is ​​constructed based on the preset behavior time series corresponding to the complex behavior and the basic behavior time series corresponding to multiple sets of basic behavior units; Determine the local distance between any two elements in the preset behavior time series and the basic behavior time series; Based on all the local distances, a path is planned using a dynamic programming algorithm on the distance matrix to obtain the planned path distance; When the planned path distance is less than or equal to a preset distance threshold, the basic behavior time series and the preset behavior time series are confirmed to match.

[0010] In this embodiment, path planning for the distance matrix is ​​achieved through dynamic programming algorithm and local distance, thereby realizing the calculation of element similarity between preset behavior time series and basic behavior time series. At the same time, it confirms whether the basic behavior time series and preset behavior time series match, thus realizing the comprehensive capture of the complexity and diversity of vehicle behavior and improving the matching accuracy of basic behavior time series.

[0011] Further, the step of performing rule validation on the basic behavioral time series to obtain the rule validation result includes: Determine the time deviation between adjacent behaviors between the preset behavior time series and the basic behavior time series; When the time deviation of all adjacent behaviors is less than a preset deviation threshold, a rule verification result is obtained that indicates the basic behavior time series verification has passed.

[0012] In this embodiment, the time deviation between adjacent behaviors enables dual verification of the time series of basic behaviors, avoiding the randomness of matching and thus improving the matching accuracy of complex behaviors.

[0013] Furthermore, the labels of the multiple sets of basic behavioral units are set as the labels of the first complex behavior, including: After converting the format of the label of the first complex behavior corresponding to the preset behavior time series, the converted label of the first complex behavior is determined as the label of the multiple sets of basic behavior units.

[0014] In this embodiment, by converting the format of the label of the first complex behavior corresponding to the preset behavior time sequence, the converted label of the first complex behavior is determined as the label of multiple basic behavior units. This realizes the format conversion of the label of the first complex behavior and the determination of the vehicle's complex behavior, thereby ensuring the efficient transmission of the vehicle's complex behavior and improving the interaction efficiency with other modules of the vehicle.

[0015] A vehicle behavior recognition device, comprising: The basic behavior determination module is used to determine the basic behavior units of the vehicle based on the vehicle's basic characteristics. Each basic behavior unit includes a basic behavior identifier and a behavior occurrence start time. The basic characteristics include at least one of obstacle characteristics, vehicle characteristics, traffic characteristics, environmental characteristics, and geographical location characteristics. A complex behavior matching module is used to match the basic behavior units of the vehicle with at least one complex behavior, wherein each complex behavior includes multiple basic behaviors and time information of each basic behavior. The behavior label setting module is used to set the label of the multiple sets of basic behavior units of the vehicle to the label of the first complex behavior if the multiple sets of basic behavior units of the vehicle successfully match the first complex behavior in the at least one complex behavior.

[0016] An electronic device includes a controller and a memory, wherein, Memory, used to store computer programs; The controller is used to execute the program stored in the memory to implement the vehicle behavior recognition method described above.

[0017] A vehicle that includes the aforementioned electronic equipment.

[0018] A computer-readable storage medium storing a computer program, which, when executed by a controller, implements the above-described vehicle behavior recognition method. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a vehicle behavior recognition method provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating step S101 of a vehicle behavior recognition method provided in an embodiment of the present invention; Figure 3 This is a flowchart illustrating step S102 of a vehicle behavior recognition method provided in an embodiment of the present invention; Figure 4 This is a flowchart illustrating step S302 of a vehicle behavior recognition method provided in an embodiment of the present invention; Figure 5 This is a flowchart illustrating step S303 of a vehicle behavior recognition method provided in an embodiment of the present invention; Figure 6 This is a flowchart illustrating step S103 of a vehicle behavior recognition method provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the structure of a vehicle behavior recognition device provided in one embodiment of this application; Figure 8 This is a schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0021] To make the technical problems, technical solutions, and beneficial effects solved by this application clearer, the following detailed description is provided in conjunction with embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0022] 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, not all, of the embodiments of the present invention. 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.

[0023] In one embodiment, please refer to Figure 1 This paper provides a vehicle behavior recognition method, which can be applied to applications such as... Figure 8 The controller includes the following steps S101-S103: S101. Determine the basic behavioral units of the vehicle based on its basic characteristics. Each basic behavioral unit includes a basic behavioral identifier and a start time of the behavior. The basic characteristics include at least one of obstacle characteristics, vehicle characteristics, traffic characteristics, environmental characteristics, and geographical location characteristics.

[0024] S102. Match the basic behavior unit of the vehicle with at least one complex behavior, wherein each complex behavior includes multiple basic behaviors and time information of each basic behavior.

[0025] S103. If the multiple sets of basic behavior units of the vehicle successfully match the first complex behavior in the at least one complex behavior, then the labels of the multiple sets of basic behavior units are set as the labels of the first complex behavior.

[0026] Basic features refer to the core elements of fundamental attributes in multi-source driving data, including but not limited to obstacle features, vehicle features, traffic features, environmental features, and geographic location features. Multi-source driving data refers to data collected from different data sources during vehicle driving, such as obstacle information, vehicle information, and environmental information. Obstacle features include but are not limited to obstacle type (large vehicle / small vehicle / pedestrian), behavior (lane change / following), and ID. Vehicle features include but are not limited to acceleration, braking status, following status, starting status, ID, heading angle, and lane affiliation. Traffic features include but are not limited to congestion zone markings, traffic signs (speed limit / yield), and traffic light status. Environmental features include but are not limited to weather (sunny / rainy), daytime, and nighttime. Geographic location features include but are not limited to straight roads / curves, different types of intersections, road types, and river-dividing sections. Among them, obstacle features are used to clarify the attributes and dynamics of vehicles and their interactive objects; vehicle features are used to reflect the current motion and control state of the vehicle (autonomous vehicle); traffic features are used to judge the rationality of behavior in conjunction with road traffic conditions; environmental features are used to consider the impact of the environment on behavior perception and execution; and geographic location features are used to refine behavioral features by aggregating geographic scenes. The basic behavioral unit refers to the basic information of each simple behavior, including but not limited to the basic behavior identifier and the behavior's start time. The basic behavior identifier refers to the core attribute used to clearly identify the basic behavior, i.e., the type of basic behavior, such as lane changing, speeding, etc. The behavior's start time includes the time point when the basic behavior occurs and the time point when it ends.

[0027] Each complex behavior comprises multiple basic behaviors and time information for each basic behavior, including the start and end times of the behavior. The complex behavior is formed by the ordered combination of these basic behaviors according to their corresponding time sequences. In one embodiment, the vehicle complex behavior refers to the signal after the complex behavior recognition result has been encapsulated.

[0028] Here, the first complex behavior refers to a complex behavior that successfully matches multiple sets of basic behavioral units, such as continuous multi-vehicle entry. The label of the first complex behavior refers to an identifier used to characterize the behavior type and other information of the first complex behavior. The labels of multiple sets of basic behavioral units refer to information used to characterize the complex behavior formed by multiple basic behaviors, such as the behavior type.

[0029] As an example, in step S101, information during the driving process is collected by at least one in-vehicle device to obtain driving data from different data sources. Then, the driving data from all data sources is synchronized in time, and the last frame of data is obtained as the latest data, thus obtaining multi-source driving data. Next, the multi-source driving data of the vehicle is directly acquired through a signal subscription mode, and basic features are extracted from the multi-source driving data. Specifically, feature extraction is performed on the multi-source driving data according to predefined rules to obtain all basic features in the multi-source driving data. Then, the basic behavioral units of the vehicle are determined based on the basic features of the vehicle. Specifically, each behavioral state is determined first through all basic features, and then sorted according to the timestamp of each behavioral state to obtain a state transition sequence. When all behavioral states in the state transition sequence satisfy the preset behavioral state rules, the basic behavior identifier and the start time of the behavior occurrence are determined. In one example, a trigger-based recognition rule is constructed based on the behavior rule system and the state transition mechanism. When the basic features trigger the recognition rule, the basic behavior identifier and the start time of the behavior occurrence corresponding to all basic features are determined according to the behavior rule system and the state transition mechanism.

[0030] In one example, a rule-based approach is used to extract features from obstacle features, vehicle features, traffic features, environmental features, and geographic location features in multi-source driving data, thereby obtaining all basic features in the multi-source driving data. Then, by determining the preset behavioral states of basic behaviors that all basic features satisfy, the basic behavioral identifiers corresponding to all basic features are determined. The start time of the first recorded behavioral state and the end time of the last recorded behavioral state are then determined as the start time of the behavior.

[0031] As an example, in step S102, the basic behavior units of the vehicle are matched with at least one complex behavior. Specifically, based on the start time of each basic behavior identifier, all basic behaviors corresponding to each basic behavior identifier are sorted, that is, the basic behavior identifier and behavior timestamp of the basic behaviors are sorted to obtain a basic behavior time series. Then, the multiple sets of basic behavior units of the vehicle are sorted to obtain the basic behavior time series, and complex behavior matching is performed with the preset behavior time series of each complex behavior. That is, the similarity between the basic behaviors in the basic behavior time series and the behaviors in the preset behavior time series is calculated to determine whether the multiple sets of basic behavior units of the vehicle are successfully matched with the complex behavior.

[0032] In one example, a dynamic programming algorithm is used to align the timeline of the basic behavior time series and the preset behavior time series, calculate the DTW distance between the two series to quantify the similarity, and determine whether they match by setting a similarity threshold.

[0033] As an example, in step S103, if the multiple sets of basic behavior units of the vehicle successfully match the first complex behavior among at least one complex behavior, then the labels of the multiple sets of basic behavior units are set as the labels of the first complex behavior. Specifically, when the multiple sets of basic behavior units of the vehicle successfully match a certain complex behavior among at least one complex behavior, the complex behavior that successfully matches the multiple sets of basic behavior units of the vehicle is determined as the first complex behavior, and the label corresponding to the first complex behavior is obtained and the label is determined as the label of the multiple sets of basic behavior units.

[0034] In one example, the labels of the multiple sets of basic behavioral units output include, but are not limited to, the category of complex behavior, the first time interval of the complex behavior, and the category of each basic behavior contained in the complex behavior and its corresponding second time interval (the start time and end time of the basic behavior).

[0035] In this embodiment, the vehicle behavior recognition method determines the basic behavioral units of the vehicle based on its fundamental characteristics. This avoids the limitations of a single data source, covers multiple information dimensions, and provides more comprehensive information support for behavior recognition, thereby capturing the complexity and diversity of vehicle behavior. The basic behavioral units of the vehicle are matched with at least one complex behavior. When a match is successful, the labels of multiple sets of basic behavioral units are set as the label of the first complex behavior. This achieves the matching of basic behavioral units with complex behaviors and the efficient fusion of multi-dimensional data, thereby enabling the determination of behavior labels and the accurate recognition of complex behaviors, thus improving the accuracy of vehicle behavior recognition.

[0036] In one embodiment, please refer to Figure 2 The preset behavior state rules include at least one set of preset rule conditions and a maximum time constraint; in step S101, the determination of the vehicle's basic behavior units based on the vehicle's basic characteristics, each basic behavior unit including a basic behavior identifier and a behavior occurrence start time, includes: S201. Based on all the aforementioned basic features, determine the state transition sequence corresponding to each of the aforementioned basic behaviors; the state transition sequence includes at least one behavior state.

[0037] S202. Obtain the preset rule conditions and maximum time constraints corresponding to each of the preset behavior states from the preset behavior state rules; the preset behavior state rules include at least one preset behavior identifier.

[0038] 203. When all behavioral states in the state transition sequence satisfy the corresponding preset rule conditions and maximum time constraints, the preset behavior identifier corresponding to the state transition sequence is determined as the basic behavior identifier of the basic behavior unit.

[0039] 204. The start timestamp of the first behavioral state and the end timestamp of the last behavioral state in the state transition sequence are determined as the start time of the behavior of the basic behavioral unit.

[0040] Here, a state transition sequence refers to a sequence of behavioral states used to identify each basic behavior. Preset behavioral state rules refer to the pre-defined judgment rules and maximum time constraints for each behavioral state. A state transition sequence includes at least one behavioral state. A behavioral state refers to the definable and identifiable dynamic attributes and operational characteristics of a vehicle performing a certain behavior. Preset behavioral state rules include, but are not limited to, at least one set of preset rule conditions and maximum time constraints. Preset behavioral state rules include at least one preset behavior identifier. The preset behavior identifier is used to characterize the behavior type of each set of behavioral states.

[0041] As an example, in step S201, the state transition sequence of each basic behavior is determined based on all basic features. Specifically, the behavior state satisfied by each basic feature is detected, and all behavior states of the same basic behavior are sorted by time to obtain the state transition sequence of that basic behavior.

[0042] As an example, in step S202, preset behavior state rules are obtained, and preset rule conditions and maximum time constraints corresponding to each behavior state are obtained from the preset behavior state rules; the preset behavior state rules include at least one preset behavior identifier.

[0043] As an example, in step S203, it is detected whether each behavior state in the state transition sequence satisfies its corresponding preset rule conditions and maximum time constraints. When all behavior states in the state transition sequence satisfy their corresponding preset rule conditions and maximum time constraints, a preset behavior identifier corresponding to that group of behavior states is obtained from the preset behavior state rules and set as the preset behavior identifier corresponding to the state transition sequence. Then, the preset behavior identifier corresponding to the state transition sequence is determined as the basic behavior identifier of the basic behavior unit.

[0044] As an example, in step S204, the start time of the first behavioral state in the state transition sequence is determined as the behavioral start timestamp of the basic behavior, and the end time of the last behavioral state is determined as the behavioral end timestamp of the basic behavior, so that the start time of the behavior of the basic behavioral unit can be determined.

[0045] In one example, the rule systems for different behaviors differ. The triggering of each basic behavior can be decomposed into a multi-stage behavior state sequence (i.e., a state transition sequence), which can be represented as: in, For the i-th behavior state, Behavioral state The start time, For state The termination time. Each behavioral state must satisfy preset rule conditions and a maximum time constraint. When the state transition sequence is triggered completely in order and the verification of the last behavior state passes, the behavior is considered successfully recognized. The second time interval of the basic behavior can be defined as follows: That is, from the start timestamp of the first behavioral state to the end timestamp of the last behavioral state.

[0046] In this embodiment, the state transition sequence of basic behaviors is determined through basic features, thereby enabling the determination of each behavior state. When all behavior states in the state transition sequence satisfy the corresponding preset rule conditions and maximum time constraints, the basic behavior identifier and behavior start time of each basic behavior are determined, thus achieving accurate identification of basic behaviors and improving the identification accuracy of basic behaviors.

[0047] In one embodiment, please refer to Figure 3 In step S102, matching the basic behavioral unit of the vehicle with at least one complex behavior includes: S301. Sort the basic behavior identifiers and behavior start times of the multiple basic behavior units of the vehicle to obtain a basic behavior time sequence.

[0048] S302. Determine whether the time series of the basic behaviors matches the preset time series of the complex behaviors according to the dynamic programming algorithm; the preset time series of behaviors includes multiple basic behaviors and the start and end times of each basic behavior.

[0049] S303. When the basic behavior time series matches the preset behavior time series, perform rule verification on the basic behavior time series to obtain the rule verification result.

[0050] 304. When the rule verification result indicates that the basic behavior time series verification is passed, it is determined that the multiple sets of basic behavior units of the vehicle are successfully matched with the first complex behavior in the at least one complex behavior.

[0051] The basic behavior time series refers to the sequence obtained by sorting all basic behavior identifiers and their start times in chronological order. Dynamic programming refers to Dynamic Time Warping (DTW), an algorithm used to measure the similarity of time series. This algorithm constructs a distance matrix between two curves, applies path constraints, and finds the optimal path with the minimum cumulative distance. The predefined behavior time series refers to a sequence of complex behaviors defined using historical data. A predefined behavior time series includes multiple basic behaviors and the start and end times of each basic behavior.

[0052] As an example, in step S301, the basic behavior identifiers and behavior start times of multiple basic behavior units of the vehicle are sorted. Specifically, the basic behavior identifiers and behavior start times of multiple basic behavior units of the vehicle are sorted according to the behavior start time to obtain the basic behavior time sequence.

[0053] As an example, in step S302, the dynamic programming algorithm is used to determine whether the basic behavior time series matches the preset behavior time series corresponding to the complex behavior. Specifically, a distance matrix between the basic behavior time series and the preset behavior time series is constructed, and then the dynamic programming algorithm is used to search for the optimal path distance in the constructed distance matrix and determine whether the optimal path distance meets the preset distance threshold, thereby determining whether the basic behavior time series matches the preset behavior time series.

[0054] In one example, if the optimal path distance meets a preset distance threshold, the basic behavior time series is determined to match the preset behavior time series. If the optimal path distance does not meet the preset distance threshold, the basic behavior time series is determined to not match the preset behavior time series.

[0055] As an example, in step S303, when matching the basic behavior time series and the preset behavior time series, rule verification is performed on the basic behavior time series. Specifically, the basic behavior time series is verified again to determine the validity of the matching, that is, by detecting whether the time deviations of all adjacent behaviors meet the preset time deviation threshold, thereby obtaining the rule verification result. Wherein, if the time deviations of all adjacent behaviors meet the preset time deviation threshold, a rule verification result indicating that the basic behavior time series verification has passed is obtained. Otherwise, if the time deviation of at least one adjacent behavior does not meet the preset time deviation threshold, a rule verification result indicating that the basic behavior time series verification has failed is obtained.

[0056] As an example, in step S304, when the rule verification result indicates that the basic behavior time series verification has passed, it is determined that multiple sets of basic behavior units of the vehicle are successfully matched with the first complex behavior in at least one complex behavior.

[0057] In one example, the behavior category of the preset behavior time series is determined as the behavior category of the base behavior time series.

[0058] In one example, if the rule validation result fails to represent the basic behavior time series validation, the basic behavior time series is re-matched with the preset behavior time series until it is determined that the basic behavior time series matches the preset behavior time series, or all preset behavior time series fail to match.

[0059] In one example, the preset behavior time series is That is, sequence Q is typically composed of m time stamps of basic behaviors triggered in chronological order, where, The timestamp (end timestamp) of the i-th basic action in the sequence, satisfying the timing constraints: .

[0060] The basic behavioral time series is That is, sequence R consists of n time stamps of basic actions concatenated in chronological order, where... Represents the timestamp of the i-th basic action in the sequence, and satisfies the timing constraints: .

[0061] In this embodiment, the basic behavior time series is determined by sorting the basic behavior identifiers and the start time of the behavior occurrence, thereby enabling the sorting of the basic behavior occurrence events. A dynamic programming algorithm is used to confirm whether the basic behavior time series matches a preset behavior time series corresponding to the complex behavior, achieving complex behavior matching of the basic behavior time series. Furthermore, rule verification of the basic behavior time series reduces interference from random factors through a dual verification mechanism and solves the problem of single-logic misjudgment, thus improving the recognition accuracy of complex vehicle behaviors.

[0062] In one embodiment, please refer to Figure 4 In step S302, the step of confirming whether the basic behavior time series matches the preset behavior time series corresponding to the complex behavior according to the dynamic programming algorithm includes: S401. Construct a distance matrix based on the preset behavior time series corresponding to the complex behavior and the basic behavior time series corresponding to multiple sets of basic behavior units.

[0063] S402. Determine the local distance between any two elements in the preset behavior time series and the basic behavior time series.

[0064] S403. Based on all the local distances, perform path planning on the distance matrix using a dynamic programming algorithm to obtain the planned path distance.

[0065] S404. When the planned path distance is less than or equal to a preset distance threshold, confirm that the basic behavior time series and the preset behavior time series match.

[0066] The distance matrix is ​​a two-dimensional table that measures the pairwise distances between all elements in two sequences, its core function being to quantify the similarity or difference between elements. Local distance refers to the distance between individual elements in two sequences. Planned path distance refers to the distance of the optimal path found using a dynamic programming algorithm. Preset distance threshold refers to a pre-defined distance threshold used to evaluate the planned path distance.

[0067] As an example, in step S401, a distance matrix is ​​constructed based on the preset behavior time series corresponding to complex behaviors and the basic behavior time series corresponding to multiple sets of basic behavior units. Specifically, all behavior timestamps in the preset behavior time series and all behavior timestamps in the basic behavior time series are sorted in chronological order, and then a distance matrix is ​​constructed based on the two sorted timestamp sequences.

[0068] As an example, in step S402, the local distance between each pair of elements in the preset behavior time series and the basic behavior time series is determined. Specifically, the Euclidean distance between each pair of elements in the preset behavior time series and the basic behavior time series is calculated and determined as the local distance.

[0069] In one example, the formula for calculating the local distance is: ,in, For local distance, This represents the timestamp of the i-th basic behavior in the preset behavior time series. This represents the timestamp of the j-th basic behavior in the basic behavior time series.

[0070] As an example, in step S403, path planning is performed on the distance matrix using a dynamic programming algorithm based on all local distances. Specifically, the distances between any two elements in the distance matrix are determined based on all local distances and filled into the distance matrix. Then, path planning is performed on the filled distance matrix using a dynamic programming algorithm to minimize the sum of all local distances on the path, thus obtaining the planned path distance.

[0071] In one example, finding a path from the top left corner to the bottom right corner of the distance matrix that minimizes the sum of all local distances along the path is called the "optimal regularized path," or the planned path distance. The calculation of this distance relies on the recursive relationship of dynamic programming as follows: In the formula, the cumulative distance matrix D(i,j) represents the minimum cumulative distance from the starting point (1,1) to the point (i,j). The path must satisfy the following three constraints: (1) Boundary conditions, that is, the starting point is , the endpoint is (m, n); (2) Monotonicity, i.e. path: the path can only move to the right, down or right-down, to ensure that the temporal order is not reversed; (3) each move moves to the adjacent cell, such as from (i, j) to (i+1, j), (i, j+1) or (i+1, j+1), to avoid skipping too many elements.

[0072] As an example, in step S404, a preset distance threshold is obtained, and the planned path distance is compared with the preset distance threshold. When the planned path distance is less than or equal to the preset distance threshold, it is confirmed that the basic behavior time series and the preset behavior time series match.

[0073] In one example, when the planned path distance is greater than a preset distance threshold, it is confirmed that the basic behavior time series and the preset behavior time series do not match.

[0074] In this embodiment, path planning for the distance matrix is ​​achieved through dynamic programming algorithm and local distance, thereby realizing the calculation of element similarity between preset behavior time series and basic behavior time series. At the same time, it confirms whether the basic behavior time series and preset behavior time series match, thus realizing the comprehensive capture of the complexity and diversity of vehicle behavior and improving the matching accuracy of basic behavior time series.

[0075] In one embodiment, please refer to Figure 5 In step S303, the step of performing rule verification on the basic behavioral time series to obtain the rule verification result includes: S501. Determine the time deviation between adjacent behaviors between the preset behavior time series and the basic behavior time series.

[0076] S502. When the time deviation of all adjacent behaviors is less than the preset deviation threshold, a rule verification result is obtained that indicates that the time series verification of the basic behavior has passed.

[0077] The time deviation between adjacent behaviors refers to the time difference between adjacent behaviors in the preset behavior time series and the base behavior time series. The preset deviation threshold is a pre-set threshold used to evaluate whether the time deviation passes the verification.

[0078] As an example, in step S501, the time deviation between adjacent behaviors in the preset behavior time series and the basic behavior time series is determined. Specifically, the time deviation between adjacent behaviors in the preset behavior time series and the basic behavior time series is calculated to obtain the time deviation of each adjacent behavior.

[0079] As an example, in step S502, a preset deviation threshold is obtained, and the time deviations of all adjacent behaviors are compared with the preset deviation threshold to perform rule verification on the basic behavior time series. When the time deviations of all adjacent behaviors are less than the preset deviation threshold, a rule verification result indicating that the basic behavior time series verification has passed is obtained.

[0080] In one example, when the time deviation of at least one of the time deviations of all adjacent behaviors is greater than or equal to a preset deviation threshold, a rule verification result indicating that the time series verification of the basic behavior has failed is obtained.

[0081] In one example, the formula for calculating the time deviation between adjacent actions is as follows: ; in, The time difference between adjacent actions; This is the end timestamp of the i-th basic behavior in the preset behavior time series; This is the end timestamp of the (i-1)th basic behavior in the preset behavior time series; It is the behavior end timestamp of the i-th basic behavior in the basic behavior time series; It is the behavior end timestamp of the (i-1)th basic behavior in the basic behavior time series.

[0082] In this embodiment, the time deviation between adjacent behaviors enables dual verification of the time series of basic behaviors, avoiding the randomness of matching and thus improving the matching accuracy of complex behaviors.

[0083] In one embodiment, please refer to Figure 6 In step S103, setting the labels of the multiple sets of basic behavioral units as the labels of the first complex behavior includes: S601. After converting the format of the label of the first complex behavior corresponding to the preset behavior time series, the format-converted label of the first complex behavior is determined as the label of the multiple sets of basic behavior units.

[0084] The first complex behavior label refers to the behavior label of a complex behavior that successfully matches multiple sets of basic behavior units. This label characterizes the type and timing information of the complex behavior, such as a sudden braking label or a lane-changing label. The behavior label is the behavior type defined using historical data. The labels of the multiple sets of basic behavior units refer to the identified labels used to characterize the complex behavior of the vehicle.

[0085] As an example, in step S601, when multiple sets of basic behavior units of the vehicle successfully match the first complex behavior among the at least one complex behavior, a label corresponding to a preset behavior time sequence is obtained and determined as the label of the first complex behavior. Then, the label of the first complex behavior is format-converted. Specifically, the complex behavior category corresponding to the first complex behavior, the first time interval of the complex behavior (including the start and end timestamps of the complex behavior), and the categories of each basic behavior included in the complex behavior and their corresponding second time intervals (including the start and end timestamps of the basic behaviors) are standardized and encapsulated to obtain the converted label of the first complex behavior, and the format-converted label of the first complex behavior is determined as the label of multiple sets of basic behavior units.

[0086] In one example, the identified complex behaviors can be formatted and output, and the output of complex vehicle behaviors is represented as follows: in, For complex behavior categories; This is the starting time of the complex behavior, that is, the starting timestamp of the first basic behavior. ; This refers to the termination time of the complex behavior, specifically the timestamp indicating the end of the last basic behavior. ; A set of k basic behaviors that constitute a complex behavior.

[0087] The i-th basic behavior contained in complex behavior C Represented as: in, Types of basic behaviors; This is the start timestamp of the i-th basic behavior; This is the timestamp of the end of the i-th basic behavior.

[0088] In this embodiment, by converting the format of the label of the first complex behavior corresponding to the preset behavior time sequence, the converted label of the first complex behavior is determined as the label of multiple basic behavior units. This realizes the format conversion of the label of the first complex behavior and the determination of the vehicle's complex behavior, thereby ensuring the efficient transmission of the vehicle's complex behavior and improving the interaction efficiency with other modules of the vehicle.

[0089] In one embodiment, an incremental data backhaul strategy is adopted, in which a portion of the collected data is periodically uploaded to the cloud platform to iteratively optimize the preset behavior time series corresponding to each complex behavior, and to iteratively optimize all the preset thresholds.

[0090] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0091] This application also provides a vehicle behavior recognition device 70, please refer to... Figure 7 ,include: The basic behavior determination module 701 is used to determine the basic behavior units of the vehicle based on the basic characteristics of the vehicle. Each basic behavior unit includes a basic behavior identifier and a behavior occurrence start time. The basic characteristics include at least one of obstacle characteristics, vehicle characteristics, traffic characteristics, environmental characteristics, and geographical location characteristics. The complex behavior matching module 702 is used to match the basic behavior unit of the vehicle with at least one complex behavior, wherein each complex behavior includes multiple basic behaviors and time information of each basic behavior. The behavior label setting module 703 is used to set the label of the multiple sets of basic behavior units of the vehicle to the label of the first complex behavior if the multiple sets of basic behavior units of the vehicle successfully match the first complex behavior in the at least one complex behavior.

[0092] This application also provides an electronic device 80, please refer to... Figure 8 It includes a memory 801 and a controller 802, wherein the memory 801 is used to store computer programs; and the controller 802 is used to execute the programs stored in the memory 801 to implement the vehicle behavior recognition method described in any embodiment of this application.

[0093] In one embodiment, the present invention provides a vehicle including the aforementioned electronic device 80.

[0094] This application also provides a computer-readable storage medium storing a computer program that, when executed by a controller, implements the vehicle behavior recognition method described in any embodiment of this application.

[0095] In this application, "multiple" refers to two or more.

[0096] The terms “first,” “second,” “third,” “fourth,” etc., used in this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0097] In this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, in this application, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0098] Unless otherwise specified, all steps in this application may be performed sequentially or randomly. For example, if the method includes steps A and B, it means that the method may include steps A and B performed sequentially, or it may include steps B and A performed sequentially. For example, if the method may also include step C, it means that step C may be added to the method in any order. For example, the method may include steps A, B, and C, or it may include steps A, C, and B, or it may include steps C, A, and B, etc.

[0099] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A vehicle behavior recognition method, characterized in that, include: The basic behavioral units of a vehicle are determined based on its basic characteristics. Each basic behavioral unit includes a basic behavioral identifier and the start time of the behavior. The basic features include At least one of the following: obstacle features, vehicle features, traffic features, environmental features, and geographical location features; The basic behavior unit of the vehicle is matched with at least one complex behavior, wherein each complex behavior contains multiple basic behaviors and time information of each basic behavior; If the multiple sets of basic behavioral units of the vehicle successfully match the first complex behavior in the at least one complex behavior, then the labels of the multiple sets of basic behavioral units are set as the labels of the first complex behavior.

2. The vehicle behavior recognition method as described in claim 1, characterized in that, The preset behavior state rules include at least one set of preset rule conditions and a maximum time constraint; The process of determining the basic behavioral units of a vehicle based on its fundamental characteristics, each basic behavioral unit including a basic behavioral identifier and a start time of the behavioral occurrence, includes: Based on all the aforementioned basic features, a state transition sequence corresponding to each of the aforementioned basic behaviors is determined; the state transition sequence includes at least one behavioral state; From the preset behavior state rules, obtain the preset rule conditions and maximum time constraints corresponding to each behavior state; the preset behavior state rules include at least one preset behavior identifier; When all behavioral states in the state transition sequence satisfy the corresponding preset rule conditions and maximum time constraints, the preset behavior identifier corresponding to the state transition sequence is determined as the basic behavior identifier of the basic behavior unit. The start timestamp of the first behavioral state and the end timestamp of the last behavioral state in the state transition sequence are determined as the start time of the behavior of the basic behavioral unit.

3. The vehicle behavior recognition method as described in claim 1, characterized in that, The matching of the vehicle's basic behavioral units with at least one complex behavior includes: The basic behavior identifiers and behavior start times of multiple sets of basic behavior units of the vehicle are sorted to obtain a basic behavior time series. The dynamic programming algorithm is used to determine whether the time series of the basic behaviors matches the preset time series of the complex behaviors; the preset time series of behaviors includes multiple basic behaviors and the start and end times of each basic behavior. When the basic behavior time series matches the preset behavior time series, rule verification is performed on the basic behavior time series to obtain the rule verification result; When the rule verification result indicates that the basic behavior time series verification is passed, it is determined that the multiple sets of basic behavior units of the vehicle are successfully matched with the first complex behavior in the at least one complex behavior.

4. The vehicle behavior recognition method as described in claim 3, characterized in that, The step of determining whether the basic behavior time series matches the preset behavior time series corresponding to the complex behavior based on the dynamic programming algorithm includes: A distance matrix is ​​constructed based on the preset behavior time series corresponding to the complex behavior and the basic behavior time series corresponding to multiple sets of basic behavior units; Determine the local distance between any two elements in the preset behavior time series and the basic behavior time series; Based on all the local distances, a path is planned using a dynamic programming algorithm on the distance matrix to obtain the planned path distance; When the planned path distance is less than or equal to a preset distance threshold, the basic behavior time series and the preset behavior time series are confirmed to match.

5. The vehicle behavior recognition method as described in claim 3, characterized in that, The step of performing rule validation on the basic behavioral time series to obtain the rule validation result includes: Determine the time deviation between adjacent behaviors between the preset behavior time series and the basic behavior time series; When the time deviation of all adjacent behaviors is less than a preset deviation threshold, a rule verification result is obtained that indicates the basic behavior time series verification has passed.

6. The vehicle behavior recognition method as described in claim 1, characterized in that, Setting the labels of the multiple sets of basic behavioral units as the labels of the first complex behavior includes: After converting the format of the label of the first complex behavior corresponding to the preset behavior time series, the converted label of the first complex behavior is determined as the label of the multiple sets of basic behavior units.

7. A vehicle behavior recognition device, characterized in that, include: The basic behavior determination module is used to determine the basic behavior units of the vehicle based on its basic characteristics. Each basic behavior unit includes a basic behavior identifier and a behavior occurrence start time. The basic characteristics include... At least one of the following: obstacle features, vehicle features, traffic features, environmental features, and geographical location features; A complex behavior matching module is used to match the basic behavior units of the vehicle with at least one complex behavior, wherein each complex behavior includes multiple basic behaviors and time information of each basic behavior. The behavior label setting module is used to set the label of the multiple sets of basic behavior units of the vehicle to the label of the first complex behavior if the multiple sets of basic behavior units of the vehicle successfully match the first complex behavior in the at least one complex behavior.

8. An electronic device, characterized in that, Includes controller and memory, among which, Memory, used to store computer programs; A controller is used to execute a program stored in a memory to implement the vehicle behavior recognition method according to any one of claims 1 to 6.

9. A vehicle, characterized in that, Including the electronic device as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by the controller, implements the vehicle behavior recognition method according to any one of claims 1 to 6.

Images