Obstacle height identification method and device for autonomous parking, electronic equipment and medium

Abstract

The invention discloses an obstacle height identification method and device for autonomous parking, electronic equipment and a medium. The obstacle height identification method comprises the following steps: acquiring ultrasonic reflection echo information of an obstacle to be identified; performing down-sampling on the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain echo height data of the ultrasonic reflection echo information; and inputting the echo height data into a trained neural network model for identification so as to obtain a high-low classification result of the to-be-identified obstacle. According to the method, the obstacle echo information is collected, and then the data points of the ultrasonic reflection echo information are subjected to downsampling according to the preset distance interval to extract echo height data so as to train the neural network model to classify the heights of the obstacles, so that the complexity of manual calibration and feature selection is avoided, the accuracy is ensured, the calculated amount is small, and the efficiency is high. The method is suitable for engineering implementation of the vehicle-mounted ultrasonic radar.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a method, device, electronic equipment and medium for identifying the height of obstacles for autonomous parking. Background Art

[0002] In the autonomous parking system, the effective judgment of the height of obstacles directly affects the overall parking performance. Especially in the close area of ​​5 meters around the vehicle, efficient and accurate judgment of the height of obstacles can significantly improve the accuracy of parking space search, optimize the planning of parking path, and ensure the safety and efficiency of autonomous parking. At present, the use of ultrasonic detection technology to identify the height of obstacles has become an important part of the field of parking technology. The table lookup method in the prior art calibrates the echo information of typical obstacles such as curbs and walls to form a data record table, and then obtains the height attributes of obstacles by looking up the table. This method has two problems. First, since it is manually calibrated and records information of multiple dimensions to form a data table, if too much dimensional information is used, the calibration workload will increase exponentially, and the table lookup process will also become complicated; if a small amount of dimensional information is used for calibration, although it can reduce the workload and increase the feasibility of calibration, it will lead to a decrease in the accuracy of judging the height of obstacles. Second, when calibrating, each dimensional interval is manually divided, and these intervals often cannot fully and accurately reflect the real change law of the height attribute of obstacles. In addition, the SVM (Support Vector Machine) method relies heavily on expert experience in feature extraction. Since actual obstacle echoes often contain noise, manually selected features are difficult to fully reflect the characteristics of obstacle echoes, which will reduce the correlation between features and labels. Summary of the invention

[0003] The technical problem to be solved by the present invention is to overcome the defects of the prior art that the method for identifying the height of obstacles relies on manual experience to calibrate the data, resulting in low efficiency and accuracy, and to provide a method, device, electronic device and medium for identifying the height of obstacles for autonomous parking.

[0004] The present invention solves the above technical problems through the following technical solutions:

[0005] The present invention provides a method for identifying the height of an obstacle for autonomous parking, the method comprising:

[0006] Collect ultrasonic reflection echo information of obstacles to be identified;

[0007] Downsampling the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain echo height data of the ultrasonic reflection echo information;

[0008] The echo height data is input into a trained neural network model for identification to obtain a height classification result of the obstacle to be identified.

[0009] Preferably, the step of downsampling the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain the echo height data of the ultrasonic reflection echo information comprises:

[0010] Downsampling the data points of the ultrasonic reflection echo information at a first distance according to a preset first distance interval to obtain a first number of echo height data of the ultrasonic reflection echo information;

[0011] Downsampling the data points of the ultrasonic reflection echo information at a second distance according to a preset second distance interval to obtain a second number of echo height data of the ultrasonic reflection echo information;

[0012] The first distance is smaller than the second distance; the first distance interval is smaller than the second distance interval; the first number is larger than the second number.

[0013] Preferably, the step of downsampling the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain the echo height data of the ultrasonic reflection echo information specifically includes:

[0014] Downsampling the data points of the ultrasonic reflection echo information at a first distance interval at a first wave peak distance to obtain echo height data of five data points of the ultrasonic reflection echo information;

[0015] Downsampling the data points of the ultrasonic reflection echo information at a second distance interval at twice the distance of the first wave peak to obtain echo height data of three data points of the ultrasonic reflection echo information;

[0016] The second distance interval is twice the first distance interval; the first peak distance and the echo heights of the obtained 8 data points together represent the obstacle to be identified.

[0017] Preferably, the neural network model includes a multi-layer perceptron neural network model;

[0018] The multi-layer perceptron neural network model includes an input layer, a hidden layer and an output layer; the number of layers of the hidden layer is 2, and the number of nodes of the hidden layer is determined according to the dimension of the echo height data obtained by downsampling.

[0019] Preferably, the obstacle height identification method further includes:

[0020] Collect ultrasonic reflection echo information of several typical high and low obstacles;

[0021] Downsampling the data points of the ultrasonic reflection echo information of the several typical high and low obstacles to obtain the echo height data of the several typical high and low obstacles, and marking the echo height data according to the height types of the obstacles;

[0022] The constructed multi-layer perceptron neural network model is trained using the labeled echo height data based on a cross-validation method to obtain the trained neural network model.

[0023] Preferably, the several typical high and low obstacles include at least one of water barriers, large walls, obstacle vehicles and curbs, limit blocks and limit rods.

[0024] Preferably, the step of training the constructed multi-layer perceptron neural network model using the labeled echo height data based on the cross-validation method includes:

[0025] During the training process, a hyperparameter combination of the multilayer perceptron neural network model is determined based on a cross-validation method; the hyperparameter combination includes at least one of the number of hidden layer units, a learning rate, and a regularization parameter.

[0026] The present invention also provides an obstacle height recognition device for autonomous parking, the obstacle height recognition device comprising:

[0027] An information collection module is used to collect ultrasonic reflection echo information of obstacles to be identified;

[0028] An information downsampling module, used for downsampling the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain echo height data of the ultrasonic reflection echo information;

[0029] The obstacle identification module is used to input the echo height data into the trained neural network model for identification, so as to obtain the height classification results of the obstacles to be identified.

[0030] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for identifying the height of obstacles in autonomous parking when executing the computer program.

[0031] The present invention also provides a computer-readable medium having computer instructions stored thereon, and when the computer instructions are executed by a processor, the obstacle height identification method for autonomous parking as described above is implemented.

[0032] The positive and progressive effects of the present invention are:

[0033] The obstacle height recognition method for autonomous parking provided by the present invention collects obstacle echo information, and then downsamples the data points of the ultrasonic reflection echo information according to a preset distance interval to extract the echo height data to train a neural network model for obstacle height classification, thereby avoiding the complexity of manual calibration and feature selection, ensuring accuracy while having a small amount of calculation and high efficiency, and being suitable for the engineering implementation of vehicle-mounted ultrasonic radar. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0035] Figure 1 Schematic diagram of the flow of the obstacle height recognition method for autonomous parking in Embodiment 1 of the present invention.

[0036] Figure 2 Schematic diagram of the structure of the multi-layer perceptron neural network model in Example 1 of the present invention.

[0037] Figure 3 Schematic diagram of the structure of the obstacle height recognition device for autonomous parking in Example 2 of the present invention.

[0038] Figure 4 Schematic diagram of the structure of an electronic device in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0039] The present invention is further described below by way of examples, but the present invention is not limited to the scope of the examples.

[0040] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various places in the text does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0041] It should be understood that the "device", "system", "unit" and / or "module" used herein is a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0042] As shown in this document, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural, unless the context clearly indicates an exception. Generally speaking, the terms "include" and "comprise" only indicate that the steps and elements that have been clearly identified are included, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0043] The definitions of inclusion in this document, such as the terms "having", "may have", "include" or "may include" used herein, indicate the existence of the corresponding functions, operations, elements, etc. herein, and do not limit the existence of one or more other functions, operations, elements, etc. In addition, it should be understood that the terms "including" or "having" used herein indicate the existence of the features, numbers, steps, operations, elements, components, or a combination thereof described in the specification, without excluding the existence or addition of one or more other features, numbers, steps, operations, elements, components, or a combination thereof.

[0044] Flowcharts are used herein to illustrate the operations performed by the systems of the embodiments of this invention. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the various steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0045] Example 1

[0046] In the automatic parking system, radar probes distributed around the vehicle measure the distance and angle between the vehicle and surrounding objects, and then the on-board computer calculates the operation process to adjust the steering wheel rotation in accordance with the vehicle speed. The system includes an environmental data acquisition system, a central processing unit, and a vehicle strategy control system. The environmental data acquisition system includes an image acquisition system and an on-board distance detection system, which can collect image data and the distance data of surrounding objects from the vehicle body, and transmit it to the central processing unit through a data line. The central processing unit can analyze and process the collected data to obtain the current position, target position, and surrounding environmental parameters of the vehicle, make an automatic parking strategy based on the above parameters, and convert it into an electrical signal. After receiving the electrical signal, the vehicle strategy control system controls the vehicle's driving, such as angle and direction, according to the instructions until it parks.

[0047] The following embodiments of the present application provide a method, device, electronic device and medium for automatic parking obstacle recognition, wherein the automatic parking obstacle recognition device can be specifically integrated in an electronic device, and the electronic device can be a terminal, a server and other devices. Among them, the terminal can be a car machine, a mobile phone, a tablet computer, a smart Bluetooth device, a laptop computer, or a personal computer (PC) and other devices; the server can be a single server or a server cluster composed of multiple servers. In some embodiments, the automatic parking obstacle recognition device can also be integrated in multiple electronic devices, for example, the automatic parking obstacle recognition device can be integrated in multiple servers, and the automatic parking obstacle recognition method of the present application is implemented by multiple servers. In some embodiments, the server can also be implemented in the form of a terminal.

[0048] Please refer to Figure 1 , which is a flow chart of the obstacle height recognition method for autonomous parking in this embodiment. Specifically, Figure 1 As shown, the obstacle height identification method includes:

[0049] S11. Collect ultrasonic reflection echo information of obstacles to be identified; specifically, ultrasonic reflection echo information of obstacles to be identified can be collected by ultrasonic sensors. Ultrasonic sensors are sensors that convert ultrasonic signals into other energy signals (usually electrical signals). Ultrasonic waves are mechanical waves with a vibration frequency higher than 20KHz (kilohertz). Ultrasonic waves have the characteristics of high frequency, short wavelength, small diffraction phenomenon, especially good directionality, and the ability to become rays and propagate in a directional manner. In this embodiment, the ultrasonic sensor can be an ultrasonic radar.

[0050] S12. Downsample the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain echo height data of the ultrasonic reflection echo information; specifically, the ultrasonic echo is described in the form of a waveform on a horizontal axis with distance as a unit and a vertical axis with height as a unit, and the echo information is extracted from the waveform; considering that the model is to be deployed on a vehicle embedded platform, such as SOC (System on Chip) or MCU (Microcontroller Unit), it is necessary to minimize the memory and computing power occupied, and by downsampling the original input, the dimension of the model input is reduced, thereby improving the recognition efficiency.

[0051] S13, input the echo height data into the trained neural network model for identification, so as to obtain the height classification result of the obstacle to be identified. Preferably, the neural network model includes a multi-layer perceptron neural network model.

[0052] The network model structure of the multilayer perceptron used in this embodiment is composed of multiple neuron layers, including an input layer, a hidden layer, and an output layer. Each neuron layer is fully connected to the previous layer, and processes the input signal through weighted summation and nonlinear activation functions to finally generate an output. This multi-layer structure enables the multilayer perceptron to learn complex nonlinear mapping relationships, thereby solving various regression and classification tasks. The MLP (Multilayer Perceptron) model can adapt to and capture complex relationships in the data through nonlinear activation functions, which is more direct and convenient than SVM using Gaussian kernels. Therefore, this embodiment omits the process of manually calibrating different echo feature values, reduces the calibration complexity, and does not require feature engineering of traditional machine learning (such as SVM).

[0053] Further, the multilayer perceptron neural network model includes an input layer, a hidden layer and an output layer; the number of hidden layers is 2, and the number of nodes in the hidden layer is determined according to the dimension of the echo height data obtained by downsampling. In addition to downsampling the original input in the input layer, reducing the dimension of the model input, a smaller number of hidden layers and hidden layer nodes should be used in the hidden layer to reduce model parameters, reduce computational complexity and time consumption. Since the input data is processed by dimensionality reduction, the multilayer perceptron neural network model does not require too many and too complex hidden layers and structures. In this embodiment, the model is pre-trained and evaluated using 2 and 3 hidden layers respectively. After testing, when the hidden layer is 2 layers, good fitting performance can be obtained, and its classification accuracy and recall rate meet the standards, so 2 hidden layers can be used. The number of nodes in the hidden layer matches the dimension of the echo height data obtained by downsampling. It ensures that the model parameter amount is effectively reduced without losing the key input information (i.e., the echo height data at the first peak distance).

[0054] In an optional implementation, step S12 includes:

[0055] S121, downsampling data points of the ultrasonic reflection echo information at a first distance according to a preset first distance interval to obtain a first number of echo height data of the ultrasonic reflection echo information;

[0056] S122, downsampling the data points of the ultrasonic reflection echo information at a second distance according to a preset second distance interval to obtain a second number of echo height data of the ultrasonic reflection echo information;

[0057] The first distance is smaller than the second distance; the first distance interval is smaller than the second distance interval; the first number is larger than the second number. Since the noise of the echo is relatively smaller when the obstacle is closer, the confidence of the data point is higher, so more data points should be sampled at a close distance.

[0058] Specifically, at the first wave crest distance, the data points of the ultrasonic reflection echo information are downsampled according to the first distance interval to obtain echo height data of 5 data points of the ultrasonic reflection echo information; at the double distance of the first wave crest, the data points of the ultrasonic reflection echo information are downsampled according to the second distance interval to obtain echo height data of 3 data points of the ultrasonic reflection echo information; the second distance interval is twice the first distance interval; the first wave crest distance and the obtained echo heights of the 8 data points together represent the obstacle to be identified. For example, in this embodiment, the highest point of each segment is taken at interval a in 5 segments near the first wave crest distance, and the highest point of each segment is taken at interval b in 3 segments near the double distance of the first wave crest, wherein interval a is the spacing of 6 data points, and interval b is the spacing of 12 data points, and the interval value is obtained by compromise by analyzing ultrasonic echo data of multiple obstacles. Specifically, the intervals between the front and rear points are 6 and 12, respectively, which are obtained by compromise when analyzing ultrasonic echo data of various obstacles; if the interval between the segments is too small, the five segments near the distance of the first peak and the three segments near the distance twice the distance of the first peak will cover a smaller distance range, which will cause the points to be too concentrated and miss some valid information; if the interval between the segments is too large, noise points outside the echo may be included in the point range, which is also inappropriate. Therefore, the value is the result of a comprehensive compromise. The distance of the first peak and the echo height of 8 downsampled data points are used to represent an echo sample, which realizes downsampling of the original input and reduces the dimension of the model input. It should be noted that although the characteristics of the echo height are mainly focused on during the model training process, since the echo height is affected by the distance, the distance of the first peak needs to be used as input, and the trained model will also be a multi-segment model.

[0059] In another optional implementation, the obstacle height identification method further includes:

[0060] S21. Collect ultrasonic reflection echo information of several typical high and low obstacles; specifically, the several typical high and low obstacles include at least one of water barriers, large walls, obstacle vehicles and curbs, limit blocks, and limit rods.

[0061] S22, downsampling the data points of ultrasonic reflection echo information of several typical high and low obstacles to obtain echo height data of several typical high and low obstacles, and marking the echo height data according to the height types of the obstacles;

[0062] S23. Based on the cross-validation method, the labeled echo height data is used to train the constructed multi-layer perceptron neural network model to obtain a trained neural network model.

[0063] Specifically, during the training process, the hyperparameter combination of the multilayer perceptron neural network model is determined based on the cross-validation method; the hyperparameter combination may include the number of hidden layer units, learning rate, and regularization parameter. Please refer to Figure 2 , which is a schematic diagram of the structure of the multi-layer perceptron neural network model in this embodiment. Specifically, Figure 2 As shown, taking the above example as an example, a five-fold cross validation is performed for each hyperparameter combination. Specifically, the data set is divided into 5 parts with almost equal sample sizes (i.e., 5 folds), and five cycles are performed. In each round, 1 fold is used as a validation set, and the other 4 folds are used as training sets to train the model, and the precision and recall rate indicators of the model in the validation set are recorded. After the end, the average value of the performance indicators of the five cycles is calculated as the cross-validation result. Based on the cross-validation results, the best performing model hyperparameter combination is selected. Using the test set that does not participate in the cross-validation, the final selected model is evaluated to verify the generalization ability of the model. If it meets expectations, the training is completed. The number of hidden layer units is finally determined by hyperparameter tuning. After training and evaluation, the number of hidden layer units of the MLP model in this embodiment is (9, 8) to obtain better results.

[0064] In the actual simulation process, after bag playback verification, in the full distance segment of 0.65m (meters) to 5m, the recognition accuracy of this embodiment for low obstacles such as curbs and limit blocks is an average of 85%, and the recognition accuracy for high obstacles such as water barriers, vehicle fronts, and large walls is an average of 70%. Among them, in the close distance segment of 0.65~2.5m, the judgment accuracy of typical low and high obstacles reached 95% and 85% respectively. In addition, the current SVM method has a low recognition accuracy of about 40%~50% for irregularly shaped obstacles such as vehicle fronts, while the present invention can reach about 75%. Therefore, the classification performance of this embodiment is significantly improved compared to traditional methods, and the overall performance is good.

[0065] The obstacle height recognition method for autonomous parking provided by the present invention collects obstacle echo information, and then downsamples the data points of the ultrasonic reflection echo information according to a preset distance interval to extract the echo height data to train a neural network model to classify the obstacle height, thereby avoiding the complexity of manual calibration and feature selection; appropriately downsamples the original echo, and controls the number of hidden layers and nodes at the same time, effectively reducing the amount of model parameters without losing key input information, ensuring accuracy while having a small amount of calculation and high efficiency, and being suitable for engineering implementation of vehicle-mounted ultrasonic radar.

[0066] Example 2

[0067] Please refer to Figure 3 , which is a schematic diagram of the structure of the obstacle height recognition device for autonomous parking in this embodiment. Specifically, Figure 3As shown, the obstacle height identification device includes:

[0068] Information collection module 1, used to collect ultrasonic reflection echo information of obstacles to be identified;

[0069] An information downsampling module 2 is used to downsample the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain echo height data of the ultrasonic reflection echo information;

[0070] The obstacle identification module 3 is used to input the echo height data into the trained neural network model for identification, so as to obtain the height classification result of the obstacle to be identified.

[0071] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can refer to the partial description of the method embodiments. The device embodiments described above are only schematic, wherein the modules described as separate components may or may not be physically separated, and the modules displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art may understand and implement it without creative work.

[0072] Example 3

[0073] Figure 4 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the method of Embodiment 1 is implemented when the processor executes the program. Figure 4 The electronic device 30 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0074] like Figure 4 As shown, the electronic device 30 may be in the form of a general-purpose computing device, for example, it may be a server device. The components of the electronic device 30 may include, but are not limited to: at least one processor 31, at least one memory 32, and a bus 33 connecting different system components (including the memory 32 and the processor 31).

[0075] The bus 33 includes a data bus, an address bus, and a control bus.

[0076] The memory 32 may include a volatile memory, such as a random access memory (RAM) 321 and / or a cache memory 322 , and may further include a read-only memory (ROM) 323 .

[0077] The memory 32 may also include a program / utility 325 having a set (at least one) of program modules 324, such program modules 324 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0078] The processor 31 executes various functional applications and data processing, such as the method of embodiment 1 of the present invention, by running the computer program stored in the memory 32 .

[0079] The electronic device 30 may also communicate with one or more external devices 34 (e.g., keyboards, pointing devices, etc.). Such communication may be performed via an input / output (I / O) interface 35. Furthermore, the model-generated device 30 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 36. As shown, the network adapter 36 communicates with other modules of the model-generated device 30 via a bus 33. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the model-generated device 30, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems.

[0080] It should be noted that although several units / modules or sub-units / modules of the electronic device are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to an embodiment of the present invention, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be embodied.

[0081] Example 4

[0082] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method of Embodiment 1 is implemented.

[0083] The readable storage medium may include but is not limited to: a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device or any suitable combination of the above.

[0084] In a possible implementation manner, the present invention may also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the method for implementing Embodiment 1.

[0085] The program code for executing the present invention may be written in any combination of one or more programming languages, and may be executed entirely on a user device, partially on a user device, as an independent software package, partially on a user device and partially on a remote device, or entirely on a remote device.

[0086] Although the specific embodiments of the present invention are described above, it should be understood by those skilled in the art that this is only for illustration and the protection scope of the present invention is defined by the appended claims. Those skilled in the art may make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, but these changes and modifications all fall within the protection scope of the present invention.

Claims

1. A method for identifying the height of obstacles in autonomous parking, characterized in that: The obstacle height identification method comprises: Collect ultrasonic reflection echo information of obstacles to be identified; Downsampling the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain echo height data of the ultrasonic reflection echo information; The echo height data is input into a trained neural network model for identification to obtain a height classification result of the obstacle to be identified.

2. The obstacle height identification method according to claim 1, characterized in that: The step of downsampling the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain echo height data of the ultrasonic reflection echo information comprises: Downsampling the data points of the ultrasonic reflection echo information at a first distance according to a preset first distance interval to obtain a first number of echo height data of the ultrasonic reflection echo information; Downsampling the data points of the ultrasonic reflection echo information at a second distance according to a preset second distance interval to obtain a second number of echo height data of the ultrasonic reflection echo information; The first distance is smaller than the second distance; the first distance interval is smaller than the second distance interval; the first number is larger than the second number.

3. The obstacle height identification method according to claim 2, characterized in that: The step of downsampling the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain the echo height data of the ultrasonic reflection echo information specifically includes: Downsampling the data points of the ultrasonic reflection echo information at a first distance interval at a first wave peak distance to obtain echo height data of five data points of the ultrasonic reflection echo information; Downsampling the data points of the ultrasonic reflection echo information at a second distance interval at twice the distance of the first wave peak to obtain echo height data of three data points of the ultrasonic reflection echo information; The second distance interval is twice the first distance interval; the first peak distance and the echo heights of the obtained 8 data points together represent the obstacle to be identified.

4. The obstacle height identification method according to claim 1, characterized in that: The neural network model includes a multi-layer perceptron neural network model; The multi-layer perceptron neural network model includes an input layer, a hidden layer and an output layer; the number of layers of the hidden layer is 2, and the number of nodes of the hidden layer is determined according to the dimension of the echo height data obtained by downsampling.

5. The obstacle height identification method according to claim 4, characterized in that: The obstacle height identification method further includes: Collect ultrasonic reflection echo information of several typical high and low obstacles; Downsampling the data points of the ultrasonic reflection echo information of the several typical high and low obstacles to obtain the echo height data of the several typical high and low obstacles, and marking the echo height data according to the height types of the obstacles; The constructed multi-layer perceptron neural network model is trained using the labeled echo height data based on a cross-validation method to obtain the trained neural network model.

6. The obstacle height identification method according to claim 5, characterized in that: The several typical high and low obstacles include at least one of water barriers, large walls, obstacle vehicles and curbs, limit blocks, and limit rods.

7. The obstacle height identification method according to claim 5, characterized in that: The steps of training the constructed multi-layer perceptron neural network model using the labeled echo height data based on the cross-validation method include: During the training process, a hyperparameter combination of the multilayer perceptron neural network model is determined based on a cross-validation method; the hyperparameter combination includes at least one of the number of hidden layer units, a learning rate, and a regularization parameter.

8. An obstacle height recognition device for autonomous parking, characterized in that: The obstacle height recognition device comprises: An information collection module is used to collect ultrasonic reflection echo information of obstacles to be identified; An information downsampling module, used for downsampling the data points of the ultrasonic reflection echo information according to a preset distance interval to obtain echo height data of the ultrasonic reflection echo information; The obstacle identification module is used to input the echo height data into the trained neural network model for identification, so as to obtain the height classification results of the obstacles to be identified.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the obstacle height recognition method for autonomous parking is implemented as described in any one of claims 1 to 7.

10. A computer readable medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the obstacle height recognition method for autonomous parking is implemented as described in any one of claims 1 to 7.

Images