Method and device for identifying vehicle cut-in behavior, electronic equipment and storage medium
By filtering target vehicles within the vehicle's radius during autonomous driving, acquiring environmental and driving state features, and using machine learning to predict the probability of vehicle entry, this method solves the problems of wasted computational resources and insufficient adaptability in the prediction of vehicle entry behavior in existing technologies, and achieves efficient and accurate vehicle entry behavior recognition.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SUZHOU QINGZHOU ZHIHANG INTELLIGENT TECH CO LTD
- Filing Date
- 2024-12-31
- Publication Date
- 2026-06-30
Smart Images
Figure CN122300545A_ABST
Abstract
Description
Technical Field
[0001] This disclosure belongs to the field of data processing technology, and specifically relates to a method, device, electronic device and storage medium for identifying vehicle entry behavior. Background Technology
[0002] Autonomous driving refers to the technology that enables vehicles to navigate and drive automatically without the active intervention of a human driver. During vehicle operation, autonomous driving systems predict whether vehicles in adjacent lanes will merge into the lane currently in which the vehicle is located, typically based on methods such as rules, probability statistics, machine learning, and deep learning.
[0003] However, the relevant technologies cannot adapt to complex scenarios when making predictions based on rules and traditional machine learning methods. Using probabilistic and statistical predictions relies too heavily on parameter selection, while deep learning methods require a large amount of data, which in turn consumes a lot of computing resources. Summary of the Invention
[0004] This disclosure proposes a method, apparatus, electronic device, and storage medium for identifying vehicle cutting-in behavior.
[0005] The first aspect of this disclosure provides a method for recognizing vehicle cutting-in behavior, the method comprising:
[0006] Identify target vehicles within the current vehicle screening radius;
[0007] Obtain the vehicle environment characteristics and vehicle driving status characteristics of the target vehicle;
[0008] Based on the vehicle environment characteristics and the vehicle driving state characteristics, predict the probability of the target vehicle entering the channel corresponding to the current vehicle after a preset time period.
[0009] Based on the cut-in probability, predict whether the target vehicle will make a cut-in behavior after a preset time period.
[0010] In this embodiment of the disclosure, determining the target vehicle within the current vehicle screening radius includes:
[0011] Based on the current speed of the vehicle, determine the screening radius for selecting target vehicles after a preset time period;
[0012] Centered on the current vehicle, vehicles within the screening radius of the current vehicle that meet a preset standard in terms of the angle between them and the current vehicle are identified as target vehicles.
[0013] In this embodiment of the disclosure, determining the screening radius for selecting target vehicles after a preset time period based on the current vehicle speed includes:
[0014] The product of the speed and the preset time is used as the first screening radius;
[0015] If the first screening radius is greater than or equal to the preset second screening radius, then the first screening radius is used as the screening radius of the target vehicle; the preset second screening radius represents the screening radius when the current vehicle speed is zero.
[0016] If the first screening radius is less than or equal to the preset second screening radius, then the preset second screening radius is used as the screening radius of the target vehicle.
[0017] In this embodiment of the disclosure, determining vehicles within the screening radius of the current vehicle and at a preset angle to the current vehicle as target vehicles includes:
[0018] The vehicles within a circle centered on the current vehicle and with the filtering radius as the radius will be used as the initial vehicle set.
[0019] In the initial vehicle set, vehicles whose absolute value of the angle between their vehicle and the direction of the current vehicle's front is less than or equal to a preset angle are selected as target vehicles.
[0020] In this embodiment of the disclosure, the channel corresponding to the current vehicle includes a first channel, a second channel, a third channel, a fourth channel, and a fifth channel;
[0021] The third channel is the channel where the current vehicle is located, and the width of the third channel is equal to the width of the current vehicle.
[0022] The second channel and the fourth channel are symmetrical about the third channel, and the width of the second channel and the fourth channel is half the width of the vehicle.
[0023] The first channel and the fifth channel are symmetrical about the third channel, and the width of the first channel and the fifth channel is greater than the width of the vehicle.
[0024] In this embodiment of the disclosure, predicting the cut-in probability of the target vehicle into the channel corresponding to the current vehicle after a preset time period based on the vehicle environmental features and the vehicle driving state features includes:
[0025] The vehicle environment features are converted into the target vehicle position data in the current vehicle's coordinate system;
[0026] Based on the target vehicle location data and the vehicle driving status characteristics, the cut-in probability of the target vehicle in each channel after a preset time period is predicted.
[0027] In this embodiment of the disclosure, the coordinate system of the current vehicle is a coordinate system established based on the position of the current vehicle at the previous moment.
[0028] In this embodiment of the disclosure, predicting the cut-in probability of the target vehicle in each channel after a preset time period based on the target vehicle location data and the vehicle driving state characteristics includes:
[0029] The vehicle location data is fused with the vehicle driving state features to obtain fused features;
[0030] Based on the fusion features, predict the cut-in probability of the target vehicle in each channel after a preset time period.
[0031] In this embodiment of the disclosure, it also includes:
[0032] The vehicle location data and the vehicle driving status features are concatenated and / or weighted summed to obtain the fused features.
[0033] In this embodiment of the disclosure, predicting whether the target vehicle will engage in a cut-in behavior after a preset time period based on the cut-in probability includes:
[0034] Channels with a cut-in probability greater than a preset probability threshold are identified as cut-in channels;
[0035] Based on the position of the cutting-in channel in each channel, it is predicted whether the target vehicle will engage in cutting-in behavior.
[0036] In this embodiment of the disclosure, predicting whether the target vehicle has engaged in cutting-in behavior based on the position of the cutting-in channel in each channel includes:
[0037] If the cutting-in channel is any one of the second channel, the third channel, or the fourth channel, then the target vehicle has engaged in cutting-in behavior.
[0038] If the cutting-in channel is the first channel or the fifth channel, then the target vehicle does not engage in cutting-in behavior.
[0039] An embodiment of the second aspect of this disclosure provides a vehicle cutting-in behavior recognition device, the device comprising:
[0040] The target vehicle filtering module is used to identify target vehicles within the current vehicle filtering radius.
[0041] The target vehicle feature acquisition module is used to acquire the vehicle environment features and vehicle driving state features of the target vehicle.
[0042] The probability prediction module is used to predict the probability of the target vehicle entering the channel corresponding to the current vehicle after a preset time period, based on the vehicle environment characteristics and the vehicle driving state characteristics.
[0043] The cut-in behavior prediction module is used to predict whether the target vehicle will engage in cut-in behavior after a preset time period based on the cut-in probability.
[0044] An embodiment of the third aspect of this disclosure provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in the first aspect or any optional embodiment of the first aspect.
[0045] An embodiment of the fourth aspect of this disclosure provides a computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the method described in the first aspect or any optional embodiment of the first aspect.
[0046] The technical solutions provided in this disclosure have at least the following technical effects or advantages:
[0047] This embodiment of the disclosure determines target vehicles within the current vehicle's screening radius, which can filter out target vehicles that may engage in cutting-in behavior, avoiding treating all vehicles around the current vehicle as target vehicles and reducing computational load; it only acquires the vehicle environment characteristics and vehicle driving state characteristics of the target vehicle, avoiding calculations based on all data related to the target vehicle, further reducing the computational load, and at the same time, based on the vehicle environment characteristics, it can be applied to varying road environments to a certain extent; based on this, based on the vehicle environment characteristics and vehicle driving state characteristics, it predicts the probability of the target vehicle reaching the current vehicle's corresponding channel after a preset time period to cut in, and finally, based on the cutting-in probability, it predicts whether the target vehicle will engage in cutting-in behavior after the preset time period.
[0048] Additional aspects and advantages of this disclosure will be set forth in part in the description which follows, and in part will be obvious from the description or may be learned by practice of this disclosure. Attached Figure Description
[0049] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this disclosure. Furthermore, the same reference numerals denote the same parts throughout the drawings.
[0050] In the attached diagram:
[0051] Figure 1A flowchart of a vehicle cutting behavior recognition method provided in an embodiment of this disclosure is shown;
[0052] Figure 2 A schematic diagram of a vehicle cutting behavior recognition method provided in an embodiment of this disclosure is shown;
[0053] Figure 3 A schematic diagram of a method for recognizing vehicle cutting-in behavior according to an embodiment of the present disclosure is shown;
[0054] Figure 4 A schematic diagram of a vehicle cutting behavior recognition method provided in an embodiment of this disclosure is shown;
[0055] Figure 5 A schematic diagram of a vehicle cutting behavior recognition method provided in an embodiment of this disclosure is shown;
[0056] Figure 6 A schematic diagram of the structure of a vehicle cutting behavior recognition device provided in another embodiment of this disclosure is shown;
[0057] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure is shown;
[0058] Figure 8 A schematic diagram of a storage medium provided according to an embodiment of the present disclosure is shown. Detailed Implementation
[0059] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
[0060] It should be noted that, unless otherwise stated, the technical or scientific terms used in this disclosure shall have the ordinary meaning as understood by one of ordinary skill in the art to which this disclosure pertains.
[0061] The following describes the implementation scenarios and related technologies involved in the embodiments of this disclosure.
[0062] Autonomous driving refers to the technology that enables vehicles to navigate and drive automatically without active human driver intervention. During vehicle operation, a key issue for ensuring driving safety is predicting whether vehicles in adjacent lanes will cut in front of the current vehicle. This predictive task is particularly complex on highways due to the involvement of various vehicle interaction methods and potential game-theoretic behaviors.
[0063] The following methods are commonly used in related technologies for prediction:
[0064] Rule-based methods, at their core, determine the likelihood of a cutting-in action based on predefined rules and vehicle behavior patterns. Specifically, they consider factors such as vehicle orientation, turn signal signals, speed and acceleration changes, and whether the vehicle is crossing lane lines. While easy to implement and computationally inexpensive, this method is poorly suited to complex scenarios and cannot handle non-regular behaviors.
[0065] The core idea of probabilistic statistical methods is to infer the likelihood of a cut-in by statistically analyzing the probability distribution of vehicle motion parameters. Specifically, Bayes' theorem is used to calculate the probability distribution; Hidden Markov Models (HMMs) model the vehicle state as a sequence of hidden states, predicting future behavior based on observations. This method can handle a certain degree of uncertainty, but manually selected parameters are sensitive to the results.
[0066] Traditional machine learning algorithms rely on handcrafted features for prediction. Specifically, they use algorithms such as Support Vector Machines (SVM), Decision Trees, and Random Forests, considering features like vehicle speed, acceleration, and relative position. This approach depends on feature engineering and can handle non-linear relationships, but it struggles with highly complex non-linear relationships.
[0067] Deep learning methods, at their core, utilize neural network structures for training to produce multimodal predictions. Specifically, they employ structures such as RNNs, CNNs, and Transformers, trained on large datasets. While capable of handling highly complex nonlinear relationships, this approach requires substantial data, can be time-consuming for inference, and consumes significant resources.
[0068] In view of the above, this disclosure provides a method, apparatus, electronic device, and storage medium for identifying vehicle cutting-in behavior. The technical solutions of this disclosure are described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of this disclosure will now be described with reference to the accompanying drawings.
[0069] like Figure 1 As shown, this disclosure provides a method for recognizing vehicle entry behavior. This method can be applied to the processor of the current vehicle or the cloud server corresponding to the current vehicle. The method includes:
[0070] In step S11, target vehicles within the current vehicle screening radius are determined.
[0071] For example, in this embodiment of the disclosure, the screening radius can be obtained by statistically analyzing experimental data of various types of vehicles. The specific value of the screening radius is not limited in this embodiment, and those skilled in the art can adjust it according to actual conditions. Target vehicles within the current vehicle's screening radius can be identified by acquiring corresponding data from the vehicle's onboard sensors. Alternatively, target vehicles within the current vehicle's screening radius can be obtained through vehicle network data, navigation data, or other methods.
[0072] Identifying target vehicles using data from onboard sensors can be achieved through the following methods: First, onboard sensors such as radar, LiDAR, and cameras are used to collect data about the surrounding environment. Second, the sensor data is preprocessed, including filtering, noise reduction, and calibration, to improve data quality. Third, deep learning models (such as YOLO, SSD, Faster R-CNN, etc.) or traditional computer vision algorithms are used to detect target vehicles in the preprocessed data. Finally, for each detected target vehicle, its relative distance to the current vehicle is calculated. For example, radar can directly measure the target distance; LiDAR calculates the distance by measuring the round-trip time or phase difference of the laser pulse; and cameras can use stereo vision or depth estimation algorithms to infer the distance from the image.
[0073] Another way to determine the target vehicle is through vehicle-to-everything (V2X) data, which can be based directly on the distance between the target vehicle and the current vehicle.
[0074] Among them, the target vehicle can also be determined by distance and angle: based on the current vehicle speed, the screening radius for selecting target vehicles after a preset time period is determined; with the current vehicle as the center, vehicles within the current vehicle's screening radius and whose angle with the current vehicle meets the preset standard are identified as target vehicles.
[0075] For example, a preset time period can first be defined to predict possible future entry points. For instance, the preset time period could be 3 seconds, 5 seconds, etc. The filtering radius is then calculated using the current vehicle speed and the preset time period. The filtering radius can be calculated using the following formula: Filtering radius = Current speed × Preset time period.
[0076] For each detected target vehicle, calculate its distance and azimuth angle relative to the current vehicle. The azimuth angle can be determined by the angle between the current vehicle's and the target vehicle's headlines at the corresponding times. Target vehicles within the filtering radius that have an angle with the current vehicle within a preset angle range are selected. The preset angle can be set according to road conditions and safety considerations, such as 30 degrees or 45 degrees.
[0077] Specifically, in some embodiments, the determination of the target vehicle differs depending on whether the current vehicle is stationary or traveling at different speeds. Therefore, considering the vehicle's operating state, the screening radius can also be determined in the following way: the product of speed and preset time is used as the first screening radius; if the first screening radius is greater than or equal to a preset second screening radius, then the first screening radius is used as the screening radius of the target vehicle; the preset second screening radius represents the screening radius when the current vehicle speed is zero; if the first screening radius is less than or equal to the preset second screening radius, then the preset second screening radius is used as the screening radius of the target vehicle.
[0078] The initial vehicle set is formed by selecting vehicles within a circle centered on the current vehicle and with a radius equal to the selected radius. Within the initial vehicle set, vehicles whose absolute value of the angle between their vehicle and the direction of the current vehicle's front is less than or equal to a preset angle are selected as target vehicles.
[0079] like Figure 2 As shown, av_speed*PredictionTime represents the filtering radius. Vehicles within the range of av_speed*PredictionTime constitute the initial vehicle set. Figure 2 Vehicles 1, 2, 3, 4, and 5 form the initial vehicle set. The target vehicle is determined from the initial vehicle set based on the angle between each vehicle and the current vehicle. Specifically, vehicles 1, 3, 4, and 5 are identified as the target vehicles.
[0080] The above process ensures that vehicles within a certain range are still set when driving at low speeds, while vehicles within a larger range are set when driving at high speeds, because vehicles can cover a greater distance within the predicted time when driving at high speeds.
[0081] In step S12, the vehicle environment characteristics and vehicle driving status characteristics of the target vehicle are obtained.
[0082] For example, vehicle environmental characteristics can include traffic conditions, weather conditions, road type, distribution of surrounding vehicles, vehicle history information, map information, etc. Vehicle driving state characteristics can include the target vehicle's speed, acceleration, heading angle, heading angular velocity, distance to the current vehicle, and rate of change of relative distance, etc.
[0083] In step S13, based on vehicle environmental characteristics and vehicle driving state characteristics, the probability of the target vehicle entering the channel corresponding to the current vehicle after a preset time period is predicted.
[0084] For example, feature data can be preprocessed before predicting the cut-in probability, including normalization, missing value handling, outlier detection, etc., to improve the prediction results of the cut-in probability.
[0085] Choose a suitable machine learning model to predict the cutoff probability. Possible models include logistic regression, decision trees, random forests, support vector machines, and neural networks. Train the model using historical data and adjust model parameters to optimize predictive performance. Evaluate the model's predictive accuracy through cross-validation or holding a validation set.
[0086] Using a trained model, based on the characteristics of the target vehicle at the current moment, predict the probability that the target vehicle will enter the current vehicle lane after a preset time period.
[0087] In some embodiments, the channel corresponding to the current vehicle includes a first channel, a second channel, a third channel, a fourth channel, and a fifth channel; the third channel is the channel where the current vehicle is located, and the width of the third channel is equal to the width of the current vehicle; the second channel and the fourth channel are symmetrical about the third channel, and the width of the second channel and the fourth channel is half the width of the vehicle; the first channel and the fifth channel are symmetrical about the third channel, and the width of the first channel and the fifth channel is greater than the width of the vehicle.
[0088] For example, such as Figure 3 The diagram shown is a schematic representation of each channel in an embodiment of this disclosure. Channels 0-4 represent the first to fifth channels, respectively, and channel 2 represents the third channel. Figure 3 In this context, `prev_target_lane_path` is the target route selected by the current vehicle at the previous moment, that is, the route that the current vehicle planned to go to at the previous moment. This route is assembled from the center lines of the lanes.
[0089] In some embodiments, based on vehicle environmental features and vehicle driving state features, predicting the entry probability of the target vehicle in the channel corresponding to the current vehicle after a preset time period includes: converting the vehicle environmental features into the target vehicle position data in the coordinate system of the current vehicle; and predicting the entry probability of the target vehicle in each channel after the preset time period based on the target vehicle position data and vehicle driving state features.
[0090] For example, the center position and heading of the current vehicle are determined. Using the center position and heading of the current vehicle, the world coordinates of the target vehicle are transformed into coordinates relative to the current vehicle's Spherical Layer Coordinate System (SL). Here, the coordinate system of the current vehicle is established based on the current vehicle's position at the previous moment. After transformation to the current vehicle's coordinate system, the probability of the target vehicle cutting into each channel is predicted.
[0091] Based on the target vehicle's location data and driving status characteristics, predict the cut-in probability of the target vehicle in each channel after a preset time period, including:
[0092] Vehicle location data is fused with vehicle driving status features to obtain fused features; based on the fused features, the probability of a target vehicle entering each channel after a preset time period is predicted.
[0093] For example, location data and driving status features can be fused into a new feature set. This can be achieved in several ways, such as directly concatenating different feature vectors into a single long vector, assigning different weights based on the importance of the features, or using methods such as Principal Component Analysis (PCA) to transform the feature space to reduce dimensionality and highlight important features.
[0094] In step S14, based on the cut-in probability, it is predicted whether the target vehicle will cut in after a preset time period.
[0095] For example, for each target vehicle, its cut-in probability across all lanes is checked. If the cut-in probability in any lane exceeds a set threshold, the target vehicle is predicted to have engaged in cut-in behavior. In some cases, it may be necessary to consider the cut-in probability of a target vehicle across multiple lanes, not just a single lane. A rule can be set, such as considering cut-in behavior as occurring if the cut-in probabilities of multiple lanes simultaneously exceed a threshold, or if the total probability exceeds a certain value. Since the vehicle's driving state and environmental characteristics change in real time, these characteristics need to be monitored in real time, and the predicted cut-in probability needs to be updated accordingly.
[0096] In some embodiments, step S14 can be implemented by identifying channels with a cut-in probability greater than a preset probability threshold as cut-in channels; and predicting whether the target vehicle has cut-in behavior based on the position of the cut-in channels in each channel.
[0097] For example, a probability threshold can be set to determine whether a lane intrusion is likely. This threshold can be set based on the system's security requirements and historical data. Lanes with an intrusion probability greater than the preset probability threshold are identified as intrusion lanes. The location of intrusion lanes among all lanes is analyzed to determine whether they are in critical areas, such as lanes in front of or to the side of vehicles.
[0098] Based on the location and number of entry channels, predict whether a target vehicle will engage in entry behavior. Rules can be set, such as predicting entry behavior if a channel in a key area is identified as an entry channel. If multiple channels are identified as entry channels, further analysis of their locations and relative importance may be needed. Since vehicle driving states and environmental characteristics change in real time, these characteristics need to be monitored in real time, and the predicted entry probability needs to be updated accordingly.
[0099] In this embodiment of the disclosure, based on Figure 3The indicated channel positions, based on the position of the cutting-in channel in each channel, predict whether a target vehicle will cut in, including: if the cutting-in channel is any one of the second, third, or fourth channels, then the target vehicle will cut in; if the cutting-in channel is the first or fifth channel, then the target vehicle will not cut in. Figure 4 As shown, if the target vehicle (agent) is in the first channel corresponding to the current vehicle after a preset time period, it means that the target vehicle has not engaged in any cutting-in behavior.
[0100] The following is a detailed flowchart illustrating the method for recognizing vehicle cutting-in behavior. Figure 5 As shown, cut-in features refer to features related to the cut-in action, which may include the vehicle's relative position, speed, acceleration, etc. Regular features may include the vehicle's historical information, map information, etc.
[0101] Embedding: Converts features into processable numerical vectors. For cut-in features and regular features, the embedding layer converts them into fixed-dimensional vectors. Attention: A self-attention mechanism is used to capture dependencies between features. When processing sequence data, the attention layer helps the model focus on more important features. Cut and fusion: This refers to cutting and fusing different feature vectors to combine them into a unified representation. A classifier is used to classify the fused features. In the context of cut-in prediction, the classifier might be used to predict whether a vehicle will perform a cut-in action. Channel probability: Based on the model's output, predicts the probability that the obstacle vehicle might reach the channel after 3 seconds. Here, "channel" might refer to the vehicle's possible location on the road. Generating tensor channell: Generates a tensor representing the probability distribution of each channel based on the model's output. Generating cut-in probability based on channel: Calculates the probability of a cut-in action based on the channel probability distribution. If the target vehicle is in the middle three channels (high-risk areas) after 3 seconds, a cut-in action is considered to have occurred.
[0102] Embedding is typically implemented using fully connected layers or pre-trained embedding layers. Attention can utilize multi-head self-attention mechanisms, where each head focuses on different aspects of the features. Cut and fusion involve feature selection and fusion techniques, such as feature concatenation or weighted summation. The classifier can be one or more fully connected layers, with the last layer using a softmax function for multi-class classification. Channel probabilities are converted from the classifier's output into a probability distribution using the softmax function. Post-processing involves probability calculation and the determination of the cut-in behavior.
[0103] In summary, the embodiments of this disclosure determine target vehicles within the current vehicle's screening radius, thus filtering out target vehicles that may engage in cutting-in behavior. This avoids considering all vehicles around the current vehicle as target vehicles, reducing computational load. By acquiring only the target vehicle's environmental and driving state characteristics, calculations based on all data related to the target vehicle are avoided, further reducing computational load. Furthermore, the vehicle environmental characteristics allow for applicability to varying road environments. Based on these characteristics, the probability of a target vehicle entering the current vehicle's corresponding lane after a preset time period is predicted. Finally, based on the entry probability, the presence of cutting-in behavior by the target vehicle after the preset time period is predicted.
[0104] correspond Figure 1 The illustrated method for recognizing vehicle cutting-in behavior, in this disclosure, also provides a device for recognizing vehicle cutting-in behavior, such as... Figure 6 As shown, the device includes:
[0105] The target vehicle filtering module 601 is used to determine target vehicles within the current vehicle filtering radius.
[0106] The target vehicle feature acquisition module 602 is used to acquire the vehicle environment features and vehicle driving state features of the target vehicle.
[0107] The probability prediction module 603 is used to predict the probability of the target vehicle entering the channel corresponding to the current vehicle after a preset time period, based on the vehicle environment characteristics and the vehicle driving state characteristics.
[0108] The cut-in behavior prediction module 604 is used to predict whether the target vehicle will engage in cut-in behavior after a preset time period based on the cut-in probability.
[0109] In one alternative implementation, the target vehicle screening module is further configured to:
[0110] Based on the current speed of the vehicle, determine the screening radius for selecting target vehicles after a preset time period;
[0111] Centered on the current vehicle, vehicles within the screening radius of the current vehicle that meet a preset standard in terms of the angle between them and the current vehicle are identified as target vehicles.
[0112] In one alternative implementation, the target vehicle screening module is further configured to:
[0113] The product of the speed and the preset time is used as the first screening radius;
[0114] If the first screening radius is greater than or equal to the preset second screening radius, then the first screening radius is used as the screening radius of the target vehicle; the preset second screening radius represents the screening radius when the current vehicle speed is zero.
[0115] If the first screening radius is less than or equal to the preset second screening radius, then the preset second screening radius is used as the screening radius of the target vehicle.
[0116] In one alternative implementation, the target vehicle screening module is further configured to:
[0117] The vehicles within a circle centered on the current vehicle and with the filtering radius as the radius will be used as the initial vehicle set.
[0118] In the initial vehicle set, vehicles whose absolute value of the angle between their vehicle and the direction of the current vehicle's front is less than or equal to a preset angle are selected as target vehicles.
[0119] In one optional implementation, the channel corresponding to the current vehicle includes a first channel, a second channel, a third channel, a fourth channel, and a fifth channel;
[0120] The third channel is the channel where the current vehicle is located, and the width of the third channel is equal to the width of the current vehicle.
[0121] The second channel and the fourth channel are symmetrical about the third channel, and the width of the second channel and the fourth channel is half the width of the vehicle.
[0122] The first channel and the fifth channel are symmetrical about the third channel, and the width of the first channel and the fifth channel is greater than the width of the vehicle.
[0123] In an optional implementation, the prediction cut-in probability module is further configured to:
[0124] The vehicle environment features are converted into the target vehicle position data in the current vehicle's coordinate system;
[0125] Based on the target vehicle location data and the vehicle driving status characteristics, the cut-in probability of the target vehicle in each channel after a preset time period is predicted.
[0126] In one optional implementation, the coordinate system of the current vehicle is a coordinate system established based on the position of the current vehicle at the previous moment.
[0127] In an optional implementation, the prediction cut-in probability module is further configured to:
[0128] The vehicle location data is fused with the vehicle driving state features to obtain fused features;
[0129] Based on the fusion features, predict the cut-in probability of the target vehicle in each channel after a preset time period.
[0130] In an optional implementation, the prediction cut-in probability module is further configured to:
[0131] The vehicle location data and the vehicle driving status features are concatenated and / or weighted summed to obtain the fused features.
[0132] In one alternative implementation, the prediction of the cut-in behavior module is further configured to:
[0133] Channels with a cut-in probability greater than a preset probability threshold are identified as cut-in channels;
[0134] Based on the position of the cutting-in channel in each channel, it is predicted whether the target vehicle will engage in cutting-in behavior.
[0135] In one alternative implementation, the prediction of the cut-in behavior module is further configured to:
[0136] If the cutting-in channel is any one of the second channel, the third channel, or the fourth channel, then the target vehicle has engaged in cutting-in behavior.
[0137] If the cutting-in channel is the first channel or the fifth channel, then the target vehicle does not engage in cutting-in behavior.
[0138] The vehicle cutting behavior recognition device and the vehicle cutting behavior recognition method provided in the above embodiments of this disclosure are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.
[0139] This disclosure also provides an electronic device for performing the above-described method for recognizing vehicle cutting-in behavior. Please refer to... Figure 7 This illustrates a schematic diagram of an electronic device provided by some embodiments of the present disclosure. For example... Figure 7As shown, the electronic device 7 includes: a processor 700, a memory 701, a bus 702, and a communication interface 703. The processor 700, the communication interface 703, and the memory 701 are connected via the bus 702. The memory 701 stores a computer program that can run on the processor 700. When the processor 700 runs the computer program, it executes the vehicle cutting behavior recognition method provided in any of the foregoing embodiments of this disclosure.
[0140] It should be pointed out that, Figure 7 The illustrated electronic device can be implemented as Figure 1 The electronic device can be a server-side device or an access point device. When the electronic device is implemented as a server-side device, the memory 701 can also store the aforementioned monitoring model. When the electronic device is implemented as an access point device, the electronic device can also include a display.
[0141] The memory 701 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between the virtual devices in the system is achieved through at least one communication interface 703 (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network.
[0142] Bus 702 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory 701 is used to store programs. After receiving an execution instruction, the processor 700 executes the program. The vehicle cutting behavior recognition method disclosed in any of the foregoing embodiments of this disclosure can be applied to the processor 700, or implemented by the processor 700.
[0143] The processor 700 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 700 or by instructions in software form. The processor 700 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 701. Processor 700 reads the contents of memory 701 and, in conjunction with its hardware, completes the steps of the above method.
[0144] The electronic device provided in this disclosure and the vehicle cutting behavior recognition method provided in this disclosure are based on the same inventive concept and have the same beneficial effects as the methods they employ, operate, or implement.
[0145] This disclosure also provides a computer-readable storage medium corresponding to the vehicle cutting-in behavior recognition method provided in the foregoing embodiments. Please refer to... Figure 8 The computer-readable storage medium shown is an optical disc 30, on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it executes the vehicle cutting behavior recognition method provided in any of the foregoing embodiments.
[0146] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.
[0147] The computer-readable storage medium provided in the above embodiments of this disclosure and the vehicle cutting behavior identification method provided in the embodiments of this disclosure are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.
[0148] Although alternative embodiments of this disclosure have been described, those skilled in the art, upon learning the basic inventive concept, can make further changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this disclosure.
[0149] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this disclosure. It should be understood that the above description is only a specific embodiment of this disclosure and is not intended to limit the scope of protection of this disclosure. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of this disclosure should be included within the scope of protection of this invention.
Claims
1. A method of identifying a cut-in behavior of a vehicle, characterized by, The method includes: Identify target vehicles within the current vehicle screening radius; Obtain the vehicle environment characteristics and vehicle driving status characteristics of the target vehicle; Based on the vehicle environment characteristics and the vehicle driving state characteristics, predict the probability of the target vehicle entering the channel corresponding to the current vehicle after a preset time period. Based on the cut-in probability, predict whether the target vehicle will make a cut-in behavior after a preset time period.
2. The method of claim 1, wherein, The determination of target vehicles within the current vehicle screening radius includes: Based on the current speed of the vehicle, determine the screening radius for selecting target vehicles after a preset time period; Centered on the current vehicle, vehicles within the screening radius of the current vehicle that meet a preset standard in terms of the angle between them and the current vehicle are identified as target vehicles.
3. The method of claim 2, wherein, The step of determining the filtering radius for selecting target vehicles after a preset time period based on the current vehicle speed includes: The product of the speed and the preset time is used as the first screening radius; If the first screening radius is greater than or equal to the preset second screening radius, then the first screening radius is used as the screening radius of the target vehicle; the preset second screening radius represents the screening radius when the current vehicle speed is zero. If the first screening radius is less than or equal to the preset second screening radius, then the preset second screening radius is used as the screening radius of the target vehicle.
4. The method of claim 3, wherein, The step of identifying vehicles within the current vehicle's screening radius and whose angle with the current vehicle meets a preset standard as target vehicles includes: The vehicles within a circle centered on the current vehicle and with the filtering radius as the radius will be used as the initial vehicle set. In the initial vehicle set, vehicles whose absolute value of the angle between their vehicle and the direction of the current vehicle's front is less than or equal to a preset angle are selected as target vehicles.
5. The method according to claim 1, characterized in that, The lanes corresponding to the current vehicle include the first lane, the second lane, the third lane, the fourth lane, and the fifth lane; The third channel is the channel where the current vehicle is located, and the width of the third channel is equal to the width of the current vehicle. The second channel and the fourth channel are symmetrical about the third channel, and the width of the second channel and the fourth channel is half the width of the vehicle. The first channel and the fifth channel are symmetrical about the third channel, and the width of the first channel and the fifth channel is greater than the width of the vehicle.
6. The method according to claim 5, characterized in that, The step of predicting the entry probability of the target vehicle into the channel corresponding to the current vehicle after a preset time period based on the vehicle environment characteristics and the vehicle driving state characteristics includes: The vehicle environment features are converted into the target vehicle position data in the current vehicle's coordinate system; Based on the target vehicle location data and the vehicle driving status characteristics, the cut-in probability of the target vehicle in each channel after a preset time period is predicted.
7. The method according to claim 6, characterized in that, The coordinate system of the current vehicle is a coordinate system established based on the position of the current vehicle at the previous moment.
8. The method according to claim 6, characterized in that, The step of predicting the cut-in probability of the target vehicle in each channel after a preset time period based on the target vehicle location data and the vehicle driving state characteristics includes: The vehicle location data is fused with the vehicle driving state features to obtain fused features; Based on the fusion features, predict the cut-in probability of the target vehicle in each channel after a preset time period.
9. The method according to claim 8, characterized in that, The process of fusing the vehicle location data with the vehicle driving state features to obtain fused features includes: The vehicle location data and the vehicle driving status features are concatenated and / or weighted summed to obtain the fused features.
10. The method according to claim 5, characterized in that, The step of predicting whether the target vehicle will engage in a cut-in behavior after a preset time period based on the cut-in probability includes: Channels with a cut-in probability greater than a preset probability threshold are identified as cut-in channels; Based on the position of the cutting-in channel in each channel, it is predicted whether the target vehicle will engage in cutting-in behavior.
11. The method according to claim 10, characterized in that, The prediction of whether the target vehicle will engage in cutting-in behavior based on the position of the cutting-in channel in each channel includes: If the cutting-in channel is any one of the second channel, the third channel, or the fourth channel, then the target vehicle has engaged in cutting-in behavior. If the cutting-in channel is the first channel or the fifth channel, then the target vehicle does not engage in cutting-in behavior.
12. A device for recognizing vehicle cutting-in behavior, characterized in that, The device includes: The target vehicle filtering module is used to identify target vehicles within the current vehicle filtering radius. The target vehicle feature acquisition module is used to acquire the vehicle environment features and vehicle driving state features of the target vehicle. The probability prediction module is used to predict the probability of the target vehicle entering the channel corresponding to the current vehicle after a preset time period, based on the vehicle environment characteristics and the vehicle driving state characteristics. The cut-in behavior prediction module is used to predict whether the target vehicle will engage in cut-in behavior after a preset time period based on the cut-in probability.
13. An electronic device comprising a memory, a processor, and a computer program stored in the memory, wherein the processor, when executing the computer program, implements the method of any one of claims 1 to 11.
14. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method of any one of claims 1 to 11.