A vehicle control method, apparatus, device, and storage medium
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING DESAY SV AUTOMOTIVE CO LTD
- Filing Date
- 2025-06-30
- Publication Date
- 2026-07-24
Smart Images
Figure CN120481919B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of vehicle control technology, and in particular to a vehicle control method, device, equipment, and storage medium. Background Technology
[0002] As the automotive industry rapidly evolves towards intelligence and automation, consumers' expectations for convenient car operation functions are growing. Traditional methods of opening the car trunk, whether manually pressing a tailgate switch or using a remote key, have certain limitations.
[0003] In everyday use, such as when a user's hands are full of shopping bags, luggage, or other items, manual operation is extremely inconvenient. Often, the items must be put down before the trunk can be opened, which is not only cumbersome but can also lead to damage or loss of items during the process. While remote key operation is relatively convenient, it can also cause problems in certain situations, such as when the key is inside a bag and difficult to retrieve, or when the battery is low.
[0004] Currently, to meet users' needs for convenience, several foot-activated trunk technologies have emerged. However, most of these technologies use traditional threshold algorithms to detect foot kicks. Fixed thresholds cannot dynamically adapt to complex and diverse external environments, easily leading to false positives or false negatives, significantly impacting the user experience. Summary of the Invention
[0005] This invention provides a vehicle control method, apparatus, device, and storage medium to solve the problem of poor vehicle control accuracy when controlling vehicles without contact.
[0006] In a first aspect, the present invention provides a vehicle control method, comprising:
[0007] The system acquires environmental radar data collected by the vehicle-mounted radar in the current vehicle, and determines multiple sets of channel impulse response tap data from the environmental radar data. The time delay corresponding to each set of channel impulse response tap data is within a preset time delay range, and the preset time delay range of each set of channel impulse response tap data is different.
[0008] The channel impulse response tap data is spliced together to obtain user action feature data;
[0009] The user's action feature data is processed using a preset neural network model to obtain action categories, and the current vehicle is controlled according to the action categories.
[0010] In a second aspect, the present invention provides a vehicle control device, comprising:
[0011] The tap data determination module is used to acquire environmental radar data collected by the vehicle-mounted radar in the current vehicle, and determine multiple sets of channel impulse response tap data from the environmental radar data. The time delay corresponding to each set of channel impulse response tap data is within a preset time delay range, and the preset time delay range of each set of channel impulse response tap data is different.
[0012] The splicing module is used to splice the channel impulse response tap data to obtain user action feature data;
[0013] The control module is used to process the user action feature data using a preset neural network model to obtain the action category, and control the current vehicle according to the action category.
[0014] Thirdly, the present invention provides an electronic device comprising:
[0015] At least one processor;
[0016] and memory that is communicatively connected to at least one processor;
[0017] The memory stores a computer program that can be executed by at least one processor, which is executed by at least one processor to enable the at least one processor to perform the vehicle control method of the first aspect described above.
[0018] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a processor to execute the vehicle control method of the first aspect described above.
[0019] The vehicle control scheme provided by this invention employs a multi-channel impulse response tap data splicing strategy. The spliced data contains a wider range of temporal and spatial information, enabling a more detailed depiction of the user's actions. From the subtle signal changes caused by the initial minute movements, to the signal fluctuations during the movement, and finally to the signal attenuation at the end of the movement, all are more completely represented, further enhancing the information content of the action features and providing higher-quality data input for the neural network model. The spliced data is then processed by a neural network to achieve accurate recognition of user actions, effectively solving the problems of low sensing accuracy, poor environmental adaptability, and inaccurate action recognition in existing technologies. This significantly improves the accuracy, stability, and intelligence of action detection, creating a convenient, efficient, and reliable vehicle control operation experience for users.
[0020] It should be understood that the description in this section is not intended to identify key or essential features of the invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description
[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.
[0022] Figure 1 This is a flowchart of a vehicle control method provided according to Embodiment 1 of the present invention;
[0023] Figure 2 This is a streamlined schematic diagram of a neural network model structure provided in Embodiment 1 of the present invention;
[0024] Figure 3 This is environmental radar data before noise reduction provided in Embodiment 1 of the present invention;
[0025] Figure 4 This is a noise-reduced environmental radar data provided in Embodiment 1 of the present invention;
[0026] Figure 5 This is a flowchart of a vehicle control method provided according to Embodiment 2 of the present invention;
[0027] Figure 6 This is a schematic diagram of user action feature data provided in Embodiment 2 of the present invention;
[0028] Figure 7 This is a schematic diagram of a prediction effect provided according to Embodiment 2 of the present invention;
[0029] Figure 8 This is a schematic diagram of the structure of a vehicle control device according to Embodiment 3 of the present invention;
[0030] Figure 9 This is a schematic diagram of the structure of an electronic device provided according to Embodiment 4 of the present invention. Detailed Implementation
[0031] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0032] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. In the description of this invention, unless otherwise stated, "a plurality of" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist; for example, A and / or B can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.
[0033] Example 1
[0034] Figure 1 The flowchart of a vehicle control method is provided in Embodiment 1 of the present invention. This embodiment is applicable to the control of vehicles. The method can be executed by a vehicle control device, which can be implemented in hardware and / or software. The vehicle control device can be configured in an electronic device, which can be composed of two or more physical entities or a single physical entity.
[0035] like Figure 1 As shown, the vehicle control method provided in Embodiment 1 of the present invention specifically includes the following steps:
[0036] S101. Obtain environmental radar data collected by the vehicle-mounted radar in the current vehicle, and determine multiple sets of channel impulse response tap data from the environmental radar data, wherein the time delay corresponding to each set of channel impulse response tap data is within a preset time delay range, and the preset time delay range of each set of channel impulse response tap data is different.
[0037] Specifically, Channel Impulse Response (CIR) is a way to describe the multipath effects of a channel. In wireless communication, signals propagate along different paths with varying lengths and propagation times, resulting in time delays and phase shifts when the signal reaches the receiver. The Channel Impulse Response describes the energy distribution of the signal at different points in time after it has passed through the channel.
[0038] In this embodiment, radar data of the surrounding environment outside the vehicle can be acquired using radar deployed in the vehicle. Then, multiple sets of channel impulse response (CIR) tap data are determined from the radar data. CIR tap data refers to the response values at various discrete time points in the CIR. The CIR is sampled as a series of discrete time points, each corresponding to a Tap. The CIR tap data includes the time delay and amplitude information of the radar signal reaching the vehicle-mounted radar receiver via different paths during propagation. The time position of each Tap represents the delay of the signal along different paths from the transmitter to the receiver. The amplitude of each Tap represents the signal strength along that path, reflecting characteristics such as signal reflection, refraction, and scattering. The time delay of each set of CIR tap data falls within its corresponding preset time delay range, and the preset time delay range is different for each set of CIR tap data. The vehicle-mounted radar can be deployed at the rear of the vehicle.
[0039] S102. The channel impulse response tap data is spliced together to obtain user action feature data.
[0040] Specifically, the temporal differences and signal strength variations in user action information across different CIR Taps can reflect the changes in the propagation speed and direction of radar signals in space caused by user actions. For example, if one set of CIR Taps contains data of a stronger signal received earlier, and another set contains data of a weaker signal received later, by analyzing the time and signal strength differences across multiple CIR Taps, it can be inferred that the user action was a movement from a position close to the location where the CIR Tap signal was received earlier to a location where the CIR Tap signal was received later. Furthermore, the speed of the action and signal attenuation can be inferred, thus adding crucial speed dimension information to the feature description of the user action. The stitched multiple CIR Taps significantly enhance the feature representation capability of user actions. Compared to a single CIR Tap, the data from stitched multiple CIR Taps contains broader temporal and spatial information, enabling a more detailed depiction of the entire user action, from the subtle signal changes caused by the initial small action, to signal fluctuations during the action, and finally to signal attenuation at the end of the action. Secondly, it effectively improves the model's ability to distinguish interference signals. In complex vehicle environments, interference signals often exhibit randomness and locality. Multi-CIR Tap stitched data integrates information from multiple locations and time segments, as well as the continuity, regularity, and velocity variation patterns of real user actions displayed in multi-CIR Taps. This contrasts sharply with the disorder and isolation of interference signals, enabling the model to more accurately extract the distinguishing features between interference and real user actions. Data stitching provides high-quality feature information input for subsequent models, greatly improving the performance and stability of user action recognition.
[0041] In this embodiment, user action feature data can be obtained by splicing multiple sets of channel impulse response tap data, which contains key information about the action speed dimension.
[0042] S103. Process the user action feature data using a preset neural network model to obtain the action category, and control the current vehicle according to the action category.
[0043] In this embodiment, a neural network model can be pre-built. User action feature data is input into the pre-built neural network model, which can output action categories, such as kicking categories, including front kicks or side kicks. If the action category is a pre-built category, the current vehicle can be controlled by issuing control commands corresponding to the pre-built category.
[0044] The technical solution of this invention employs a multi-channel impulse response tap data splicing strategy. The spliced data contains a wider range of temporal and spatial information, enabling a more detailed depiction of the user's actions. From the subtle signal changes caused by the initial minute movements, to the signal fluctuations during the movement, and finally to the signal attenuation at the end of the movement, all are more completely represented, further enhancing the information content of the action features and providing higher-quality data input for the neural network model. The spliced data is then processed by the neural network to achieve accurate recognition of user actions, effectively solving the problems of low sensing accuracy, poor environmental adaptability, and inaccurate action recognition in existing technologies. This significantly improves the accuracy, stability, and intelligence of action detection, creating a convenient, efficient, and reliable vehicle control operation experience for users.
[0045] Optionally, the preset neural network model includes an input layer, a convolutional network layer, and a fully connected network layer; the convolutional network layer includes a first convolutional layer with a first preset kernel size, a max pooling layer, a first activation function layer, a second convolutional layer, a second activation function layer, a third convolutional layer, a third activation function layer, a fourth convolutional layer, and a fourth activation function layer; the fully connected network layer includes an average pooling layer, a Dropout layer, a first fully connected layer, a fifth activation function layer, a second fully connected layer, and a Softmax layer; wherein the second, third, and fourth convolutional layers each include multiple convolutional kernels with a second preset kernel size, and the first preset kernel size is different from the second preset kernel size.
[0046] Specifically, the convolutional layers in the preset neural network model can perform convolution operations on the data. By using convolutional kernels of different sizes and depths, various key features in the data are extracted, such as local signal fluctuation features, spatial location correlation features, and dynamic change features of actions. The pooling layers in the preset neural network model can adaptively downsample the convolution results, automatically adjusting the pooling kernel size and stride according to the distribution of signal features, reducing the amount of data while retaining the main feature information. The fully connected layers in the preset neural network model can comprehensively analyze the highly abstract features after multiple convolutions and pooling, accurately determining whether user actions exist and their specific types, and providing a preliminary evaluation of relevant action parameters (such as force and speed).
[0047] in, Figure 2 This is a streamline diagram of a neural network model structure, such as... Figure 2 As shown, the preset neural network model used in this solution includes:
[0048] Input layer ( Figure 2 The input layer (inputlayer) is used to determine the data dimensions and perform normalization processing, providing a standardized data foundation for subsequent network processing.
[0049] Convolutional network layers: consisting of multiple convolutional layers ( Figure 2 The convolutional layer consists of conv layers. The first convolutional layer uses a kernel size (i.e., the first preset kernel size) of 5*1, with 8 kernels, using "same" padding and a stride of 1. This layer is followed by a max-pooling layer. Figure 2 The pooling kernel size is 3*1, and the stride is 2*1. Subsequent convolutional layers have a 3*1 kernel size (i.e., the first preset kernel size) and 8 kernels, also using the same padding method and a stride of 1. Some convolutional layers are followed by max-pooling layers with a 3*1 kernel size and a stride of 2*1. Through these convolutional layer settings, local signal variation features are captured along the time series direction, while also considering single-dimensional spatial features. Different numbers of kernels can generate multiple feature maps, enriching feature extraction capabilities. The "same" padding method maintains the data spatial dimension, and is further enhanced by the ReLU activation function. Figure 2 The ReLU algorithm introduces non-linearity, highlighting effective feature changes and accelerating training. Pooling layers retain the main feature change trends while reducing data volume and do not excessively lose spatial information. Through layers of convolution and pooling, more abstract and high-level user action features are gradually extracted.
[0050] Fully connected network layer: the average pooling layer in this layer ( Figure 2 fc_avgpool in the context of data dimensionality reduction can decrease computational cost while preserving global features. Dropout layers ( Figure 2 The `drop` operator in the model can randomly set the neuron output to 0 with a certain probability, preventing overfitting and improving model generalization. The number of neurons in the fully connected layer gradually changes, starting with a fully connected layer containing 32 neurons. Figure 2 The first fully connected layer (fc1) integrates local feature information, learns complex global relationships, and then passes through the ReLU activation function (…). Figure 2 After processing by fc_relu, it is connected to the last fully connected layer. Figure 2 fc2), the second fully connected layer, has the number of neurons corresponding to the number of user action categories. It maps the previously processed features to the category space and finally passes them through the Softmax layer. Figure 2The softmax function in the kernel transforms the output into a relative probability distribution for each category, thereby determining the user action category corresponding to the signal. This enables accurate classification and recognition of different user action types, providing a reliable basis for vehicle control. The number of neurons in the first fully connected layer can be greater than the number of neurons in the second fully connected layer. Same convolution mode involves adding appropriate zero-value padding around the input data during convolution to keep the size of the output feature map close to the size of the input data. Specifically, the padding size of Same convolution is automatically calculated based on the kernel size and stride to ensure that the size of the output feature map is as close as possible to the size of the input data.
[0051] Optionally, acquiring the environmental radar data collected by the vehicle-mounted radar in the current vehicle includes: acquiring the environmental radar signal collected by the vehicle-mounted radar in the current vehicle, and performing wavelet noise reduction on the environmental radar signal to obtain the noise-reduced environmental radar data.
[0052] Specifically, Figure 3 This is environmental radar data before noise reduction. Figure 4 This is a noise-reduced environmental radar data. Figure 3 and Figure 4 The horizontal axis represents the radar data frame count, and the vertical axis represents the radar data intensity. The blue line represents the polygonal line plotted from the environmental radar data points. Wavelet denoising technology can be used to preprocess the raw signals acquired by the vehicle-mounted radar. Wavelet denoising parameters can be pre-set, with the threshold selection rule using the Sqtwolog rule. This rule determines the threshold based on the signal variance, effectively adapting to different signal characteristics. A soft threshold h is used for threshold processing; compared to hard thresholding, soft thresholding better preserves signal details while suppressing noise. A scale-dependent threshold adjustment strategy dynamically adjusts the threshold according to the signal characteristics at different scales, enhancing the denoising effect. The number of decomposition layers can be set to 7. By appropriately setting the number of decomposition layers, effective decomposition and denoising processing of the signal can be performed at different frequency levels. The selected wavelet basis function can be Coif3. Coif3 basis function has good time-frequency localization characteristics in signal processing, which can accurately capture local feature changes in the signal, thereby effectively removing noise components in the signal and extracting relatively pure signal information related to the kicking action, providing a high-quality data foundation for subsequent model input.
[0053] Optionally, the vehicle-mounted radar includes an ultra-wideband radar.
[0054] Specifically, vehicle-mounted radar includes ultra-wideband (UWB) radar.
[0055] Example 2
[0056] Figure 5 This is a flowchart of a vehicle control method provided in Embodiment 2 of the present invention. The technical solution of the present invention is further optimized based on the above optional technical solutions, and a specific method for controlling the vehicle is given.
[0057] Optionally, determining multiple sets of channel impulse response tap data from the environmental radar data includes: when the number of frames of the environmental radar data is a preset number, determining multiple sets of channel impulse response tap data for each frame of environmental radar data.
[0058] Optionally, the step of splicing the channel impulse response tap data to obtain user action feature data includes: determining groups of channel impulse response tap data within the same preset time delay range as target data according to their chronological order, to obtain multiple groups of target data, wherein each group of target data is a preset number of channel impulse response tap data corresponding to environmental radar data with time delays within the same preset time delay range; splicing the multiple groups of target data to obtain user action feature data.
[0059] Optionally, controlling the current vehicle according to the action category includes: if the action category is a preset kick category, then controlling the trunk of the current vehicle to open.
[0060] like Figure 5 As shown, the vehicle control method provided in Embodiment 2 of the present invention specifically includes the following steps:
[0061] S201. Obtain the environmental radar signal collected by the vehicle-mounted radar in the current vehicle, and perform wavelet noise reduction on the environmental radar signal to obtain the noise-reduced environmental radar data.
[0062] S202. When the number of frames of the environmental radar data is a preset number, multiple sets of channel impulse response tap data are determined for each frame of environmental radar data.
[0063] Specifically, a preset number of frames, such as 200 frames, of environmental radar data can be acquired, and multiple sets of channel impulse response tap data can be determined for each frame of environmental radar data, such as 3 sets of channel impulse response tap data. The time delay in each set of channel impulse response tap data is within the corresponding preset time delay range.
[0064] S203. According to the chronological order, the channel impulse response tap data groups within the same preset time delay range are determined as target data to obtain multiple sets of target data. Each set of target data consists of a preset number of channel impulse response tap data corresponding to environmental radar data, and the time delay is within the same preset time delay range.
[0065] For example, if there are 3 groups of channel impulse response tap data and 200 frames of environmental radar data, the channel impulse response tap data in the 200 frames of environmental radar data can be classified according to the preset time delay range corresponding to each group of channel impulse response tap data to obtain multiple groups of target data. Each group of target data consists of 200 frames of channel impulse response tap data arranged in chronological order, and the time delays in these channel impulse response tap data are within the same preset time delay range.
[0066] S204. The multiple sets of target data are spliced together to obtain user action feature data.
[0067] Specifically, Figure 6 This is a schematic diagram of user action feature data. Figure 6 The horizontal axis represents the radar data frame count, the vertical axis represents the radar data intensity, and the blue broken line is a broken line drawn from user action feature data points.
[0068] S205. Process the user action feature data using a preset neural network model to obtain the action category.
[0069] S206. If the action category is a preset kick category, then control the opening of the trunk of the current vehicle.
[0070] Specifically, if the action category is the first preset kick category, the trunk of the current vehicle will be opened; if the action category is the second preset kick category, the trunk of the current vehicle will be closed.
[0071] Figure 7 This is a diagram illustrating a prediction effect. Figure 7 The horizontal axis represents the count of action samples (i.e., the number of user action feature data sets), and the vertical axis represents the action category as follows: 1 indicates a front kick, 2 indicates a left 30cm front kick, 3 indicates a right 30cm front kick, and 4 indicates a horizontal sweep. The red circles represent the action categories predicted by the model, and the purple horizontal lines represent the actual action categories, coinciding with the red circles. Experiments have shown that the method described in this scheme achieves an accuracy rate of 99.70% in action category recognition.
[0072] The vehicle control method provided in this invention, through wavelet denoising and multi-CIR Tap stitching, achieves accurate extraction of effective signal features related to user actions in complex vehicle environments. This includes subtle differences in signal characteristics across different action types and feature variation patterns under different environments. It filters out noise interference and signal fluctuations caused by environmental factors, providing high-quality data for model prediction and enhancing the ability to distinguish between interference and genuine user actions. Furthermore, a powerful neural network model is constructed that can fully learn and understand the action features in the stitched data, achieving high-precision action classification and recognition, effectively handling unknown interference, reducing the false judgment rate, and creating a convenient, efficient, and reliable trunk operation experience for users.
[0073] Example 3
[0074] Figure 8 This is a schematic diagram of a vehicle control device provided in Embodiment 3 of the present invention. Figure 8 As shown, the device includes: a tap data determination module 301, a splicing module 302, and a control module 303, wherein:
[0075] The tap data determination module is used to acquire environmental radar data collected by the vehicle-mounted radar in the current vehicle, and determine multiple sets of channel impulse response tap data from the environmental radar data. The time delay corresponding to each set of channel impulse response tap data is within a preset time delay range, and the preset time delay range of each set of channel impulse response tap data is different.
[0076] The splicing module is used to splice the channel impulse response tap data to obtain user action feature data;
[0077] The control module is used to process the user action feature data using a preset neural network model to obtain the action category, and control the current vehicle according to the action category.
[0078] The vehicle control device provided in this invention employs a multi-channel impulse response tap data stitching strategy. The stitched data contains a wider range of temporal and spatial information, enabling a more detailed depiction of the user's actions. From the subtle signal changes caused by the initial minute movements, to the signal fluctuations during the movement, and finally to the signal attenuation at the end of the movement, all are more completely represented, further enhancing the information content of the action features and providing higher-quality data input for the neural network model. The stitched data is then processed by the neural network to achieve accurate recognition of user actions, effectively solving the problems of low sensing accuracy, poor environmental adaptability, and inaccurate action recognition in existing technologies. This significantly improves the accuracy, stability, and intelligence of action detection, creating a convenient, efficient, and reliable vehicle control operation experience for users.
[0079] Optionally, the tap data determination module includes:
[0080] The tap data determination unit is used to determine multiple sets of channel impulse response tap data for each frame of environmental radar data when the number of frames of the environmental radar data is a preset number.
[0081] Optional, the splicing modules include:
[0082] The target data unit is used to determine the channel impulse response tap data group within the same preset time delay range as target data according to the time sequence, so as to obtain multiple sets of target data. Each set of target data is a preset number of channel impulse response tap data whose time delay is within the same preset time delay range, corresponding to environmental radar data.
[0083] The splicing unit is used to splice the multiple sets of target data to obtain user action feature data.
[0084] Optionally, the preset neural network model includes an input layer, a convolutional network layer, and a fully connected network layer; the convolutional network layer includes a first convolutional layer with a first preset kernel size, a max pooling layer, a first activation function layer, a second convolutional layer, a second activation function layer, a third convolutional layer, a third activation function layer, a fourth convolutional layer, and a fourth activation function layer; the fully connected network layer includes an average pooling layer, a Dropout layer, a first fully connected layer, a fifth activation function layer, a second fully connected layer, and a Softmax layer; wherein the second, third, and fourth convolutional layers each include multiple convolutional kernels with a second preset kernel size, and the first preset kernel size is different from the second preset kernel size.
[0085] Optionally, the tap data determination module includes:
[0086] The noise reduction unit is used to acquire the environmental radar signal collected by the vehicle-mounted radar in the current vehicle, and to perform wavelet noise reduction on the environmental radar signal to obtain the noise-reduced environmental radar data.
[0087] Optionally, the vehicle-mounted radar includes an ultra-wideband radar.
[0088] Optionally, the control module includes:
[0089] The trunk control unit is used to control the opening of the trunk of the current vehicle if the action category is a preset kick category.
[0090] The vehicle control device provided in the embodiments of the present invention can execute the vehicle control method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method.
[0091] Example 4
[0092] Figure 9 A schematic diagram of an electronic device 40 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as in-vehicle computers, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.
[0093] like Figure 9 As shown, the electronic device 40 includes at least one processor 41 and a memory, such as a read-only memory (ROM) 42 or a random access memory (RAM) 43, communicatively connected to the at least one processor 41. The memory stores computer programs executable by the at least one processor. The processor 41 can perform various appropriate actions and processes based on the computer program stored in the ROM 42 or loaded into the RAM 43 from storage unit 48. The RAM 43 may also store various programs and data required for the operation of the electronic device 40. The processor 41, ROM 42, and RAM 43 are interconnected via a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.
[0094] Multiple components in electronic device 40 are connected to I / O interface 45, including: input unit 46, such as keyboard, mouse, etc.; output unit 47, such as various types of monitors, speakers, etc.; storage unit 48, such as disk, optical disk, etc.; and communication unit 49, such as network card, modem, wireless transceiver, etc. Communication unit 49 allows electronic device 40 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0095] Processor 41 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 41 performs the various methods and processes described above, such as vehicle control methods.
[0096] In some embodiments, the vehicle control method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 40 via ROM 42 and / or communication unit 49. When the computer program is loaded into RAM 43 and executed by processor 41, one or more steps of the vehicle control method described above may be performed. Alternatively, in other embodiments, processor 41 may be configured to perform the vehicle control method by any other suitable means (e.g., by means of firmware).
[0097] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoC) systems, complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transferring data and instructions to the storage system, the at least one input device, and the at least one output device.
[0098] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0099] The computer equipment provided above can be used to execute the vehicle control method provided in any of the above embodiments, and has corresponding functions and beneficial effects.
[0100] Example 5
[0101] In the context of this invention, a computer-readable storage medium may be a tangible medium, wherein the computer-executable instructions, when executed by a computer processor, are used to perform a vehicle control method, the method comprising:
[0102] The system acquires environmental radar data collected by the vehicle-mounted radar in the current vehicle, and determines multiple sets of channel impulse response tap data from the environmental radar data. The time delay corresponding to each set of channel impulse response tap data is within a preset time delay range, and the preset time delay range of each set of channel impulse response tap data is different.
[0103] The channel impulse response tap data is spliced together to obtain user action feature data;
[0104] The user's action feature data is processed using a preset neural network model to obtain action categories, and the current vehicle is controlled according to the action categories.
[0105] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by, or in conjunction with, an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.
[0106] The computer equipment provided above can be used to execute the vehicle control method provided in any of the above embodiments, and has corresponding functions and beneficial effects.
[0107] It is worth noting that in the above embodiments of the vehicle control device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.
[0108] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.
Claims
1. A vehicle control method, characterized in that, include: The system acquires environmental radar data collected by the vehicle-mounted radar in the current vehicle, and determines multiple sets of channel impulse response tap data from the environmental radar data. The time delay corresponding to each set of channel impulse response tap data is within a preset time delay range, and the preset time delay range of each set of channel impulse response tap data is different. The channel impulse response tap data is spliced together to obtain user action feature data; The user's action feature data is processed using a preset neural network model to obtain action categories, and the current vehicle is controlled according to the action categories.
2. The method according to claim 1, characterized in that, The step of determining multiple sets of channel impulse response tap data from the environmental radar data includes: When the number of frames of the environmental radar data is a preset number, multiple sets of channel impulse response tap data are determined for each frame of environmental radar data.
3. The method according to claim 2, characterized in that, The process of splicing the channel impulse response tap data to obtain user action feature data includes: According to the chronological order, channel impulse response tap data groups within the same preset time delay range are identified as target data to obtain multiple sets of target data. Each set of target data consists of a preset number of corresponding environmental radar data and channel impulse response tap data within the same preset time delay range. The multiple sets of target data are spliced together to obtain user action feature data.
4. The method according to any one of claims 1-3, characterized in that, The preset neural network model includes an input layer, a convolutional network layer, and a fully connected network layer; the convolutional network layer includes a first convolutional layer with a first preset kernel size, a max pooling layer, a first activation function layer, a second convolutional layer, a second activation function layer, a third convolutional layer, a third activation function layer, a fourth convolutional layer, and a fourth activation function layer; The fully connected network layer includes an average pooling layer, a Dropout layer, a first fully connected layer, a fifth activation function layer, a second fully connected layer, and a Softmax layer; The second, third, and fourth convolutional layers each include multiple convolutional kernels with a second preset kernel size, and the first preset kernel size is different from the second preset kernel size.
5. The method according to claim 1, characterized in that, The acquisition of environmental radar data collected by the vehicle-mounted radar in the current vehicle includes: The environmental radar signal collected by the vehicle-mounted radar in the current vehicle is acquired, and wavelet denoising is performed on the environmental radar signal to obtain the denoised environmental radar data.
6. The method according to claim 1, characterized in that, The vehicle-mounted radar includes an ultra-wideband radar.
7. The method according to claim 1, characterized in that, The control of the current vehicle based on the action category includes: If the action category is the preset kick category, then control the opening of the trunk of the current vehicle.
8. A vehicle control device, characterized in that, include: The tap data determination module is used to acquire environmental radar data collected by the vehicle-mounted radar in the current vehicle, and determine multiple sets of channel impulse response tap data from the environmental radar data. The time delay corresponding to each set of channel impulse response tap data is within a preset time delay range, and the preset time delay range of each set of channel impulse response tap data is different. The splicing module is used to splice the channel impulse response tap data to obtain user action feature data; The control module is used to process the user action feature data using a preset neural network model to obtain the action category, and control the current vehicle according to the action category.
9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the vehicle control method according to any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the vehicle control method according to any one of claims 1-7.