Wheel robot zero speed detection method, system, medium and device
By calculating, standardizing, and performing Fourier transforms based on inertial sensor data, and combining one-dimensional convolutional neural networks and fully connected neural networks, the problem of low accuracy in zero-speed detection was solved, the detection capability under different motion modes was improved, and the impact of engine vibration was reduced.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WUXI A CARRIER INTELLIGENT EQUIP
- Filing Date
- 2023-06-06
- Publication Date
- 2026-06-19
AI Technical Summary
Existing zero-speed detection methods have low detection accuracy under different motion modes, and engine vibration further reduces detection accuracy.
By selecting inertial sensor data based on a preset length sliding window, performing modulus calculation, standardization, and discrete Fourier transform, a one-dimensional convolutional neural network is used to automatically extract time-domain and frequency-domain features, and a fully connected neural network is used for zero-speed detection, thus avoiding dependence on a preset threshold.
This improved the zero-speed detection capability of wheeled robots under different motion modes, reduced the impact of engine and other component vibrations on the detection effect, and achieved high-performance zero-speed detection.
Smart Images
Figure CN116753946B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of zero-speed detection technology, and in particular to a method, system, medium, and equipment for zero-speed detection of wheeled robots. Background Technology
[0002] Zero-speed detection refers to detecting periods when the vehicle's speed (for wheeled robots) is zero, serving as preliminary preparation for subsequent correction of inertial navigation system errors using zero-speed correction. Current zero-speed detection methods primarily involve inputting data from the vehicle's inertial measurement unit into a model for calculation. If the calculation result is below a set threshold, the vehicle is considered stationary.
[0003] Because traditional zero-speed detection methods rely on a predetermined threshold, they cannot reliably detect different motion modes simultaneously. In addition, engine vibration when the vehicle is stationary can also reduce the detection accuracy of methods based on fixed thresholds.
[0004] Therefore, to address the above problems, it is necessary to design a zero-speed detection method for wheeled robots to solve the problems of low detection accuracy under different motion modes and decreased detection accuracy caused by engine vibration in the existing technology. Summary of the Invention
[0005] The present invention provides a zero-speed detection method, system, medium and device for wheeled robots. It provides a high-performance zero-speed detection method that does not rely on a pre-set threshold, improves the zero-speed detection capability of wheeled robots in different motion modes, and reduces the impact of engine and other component vibrations on the zero-speed detection effect when the robot is parked.
[0006] Firstly, a zero-speed detection method for a wheeled robot is provided, specifically including the following steps:
[0007] Acquire inertial sensor data for wheeled robots;
[0008] The inertial sensor data is selected based on a preset length sliding window, and the modulus of the selected inertial sensor data is calculated to obtain one-dimensional inertial data.
[0009] The one-dimensional inertial data is standardized to obtain standardized data;
[0010] Perform a discrete Fourier transform on the standardized data to obtain the Fourier spectrum;
[0011] Based on a preset one-dimensional convolutional neural network, time-domain feature data and frequency-domain feature data are extracted from the standardized data and the Fourier spectrum.
[0012] Zero-speed detection is performed on the time-domain feature data and the frequency-domain feature data based on a preset fully connected neural network.
[0013] According to the first aspect, in a first possible implementation of the first aspect, the inertial sensor data includes triaxial specific force and triaxial angular velocity;
[0014] The formula for calculating the modulus f of triaxial specific force is as follows:
[0015]
[0016] The formula for calculating the magnitude ω of the triaxial angular velocity is as follows:
[0017]
[0018] In the formula, f x f y f z The specific force is the force along the three axes; ω x ω y ω z It represents the angular velocity of the three axes.
[0019] According to the first aspect, in a second possible implementation of the first aspect, the standardization calculation formula is as follows:
[0020]
[0021] In the formula, Standardized data obtained by standardizing one-dimensional inertial data x. Let x be the mean value of the one-dimensional inertial data, and σ(x) be the standard deviation of the one-dimensional inertial data x.
[0022] According to the first aspect, in a third possible implementation of the first aspect, the discrete Fourier transform formula is as follows:
[0023]
[0024] In the formula, X(k) is the (k+1)th value of the Fourier spectrum obtained after the discrete Fourier transform, x n This represents the (n+1)th value before the standardized data transformation, where e is the natural constant and j is the imaginary unit.
[0025] According to the first aspect, in the fourth possible implementation of the first aspect, the step of "extracting time-domain feature data and frequency-domain feature data from the standardized data and the Fourier spectrum based on a preset one-dimensional convolutional neural network" specifically includes the following steps:
[0026] The preset one-dimensional convolutional neural network includes a preset first one-dimensional convolutional neural network and a preset second one-dimensional convolutional neural network;
[0027] Temporal feature data is extracted from the standardized data based on a preset first one-dimensional convolutional neural network.
[0028] Frequency domain feature data is extracted from the Fourier spectrum based on a preset second one-dimensional convolutional neural network.
[0029] According to the first aspect, in the fifth possible implementation of the first aspect...
[0030] The preset fully connected neural network formula is as follows:
[0031] x output =max(A1x input +b1,0) Equation (5);
[0032] x input The time-domain feature data and the frequency-domain feature data; x output This represents the zero-speed output result; A1 is the transformation matrix; b1 is the deviation.
[0033] Secondly, a zero-speed detection system for wheeled robots is also provided, including:
[0034] The data acquisition module acquires inertial sensor data from the wheeled robot.
[0035] The sliding window module is communicatively connected to the data acquisition module. It is used to select the inertial sensor data based on a preset length sliding window and to perform modulus calculation on the inertial sensor data selected by the sliding window to obtain one-dimensional inertial data.
[0036] The standardization module, which is communicatively connected to the sliding window module, is used to standardize the one-dimensional inertial data and obtain standardized data.
[0037] The Fourier module, which is communicatively connected to the standardization module, is used to perform discrete Fourier transform on the standardized data to obtain the Fourier spectrum.
[0038] The feature extraction module is communicatively connected to the standardization module and the Fourier module, and is used to extract time-domain feature data and frequency-domain feature data from the standardized data and the Fourier spectrum based on a preset one-dimensional convolutional neural network.
[0039] The zero-speed detection module is communicatively connected to the feature extraction module and is used to perform zero-speed detection on the time-domain feature data and the frequency-domain feature data based on a preset fully connected neural network.
[0040] Thirdly, a storage medium is also provided, on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the zero-speed detection method for wheeled robots as described above.
[0041] Fourthly, an electronic device is also provided, including a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that the processor, when running the computer program, implements the zero-speed detection method for wheeled robots as described above.
[0042] Compared with the prior art, the advantages of the present invention are as follows: the inertial sensor data of the wheeled robot (read by the inertial measurement unit) is selected based on a sliding window of preset length. After the modulus is calculated, standardized and discrete Fourier transform, the data is input into the zero-speed detection deep learning model - a preset one-dimensional convolutional neural network to automatically extract time-domain features and frequency-domain features and detect them. Finally, the zero-speed detection result is obtained based on the preset fully connected neural network.
[0043] Therefore, by using neural networks to automatically extract and fuse the time and frequency domain features of the data, a high-performance zero-speed detection method that does not rely on a pre-set threshold is provided, which improves the zero-speed detection capability of wheeled robots in different motion modes and reduces the impact of engine and other component vibrations on the zero-speed detection effect when the robot is parked. Attached Figure Description
[0044] Figure 1 This is a flowchart illustrating an embodiment of a zero-speed detection method for a wheeled robot according to the present invention;
[0045] Figure 2 This is a schematic diagram of the connection structure between a preset one-dimensional convolutional neural network and a preset fully connected neural network according to an embodiment of the present invention;
[0046] Figure 3 This is a structural schematic diagram of a zero-speed detection system for a wheeled robot according to the present invention. Detailed Implementation
[0047] Referring now to specific embodiments of the invention, examples of which are illustrated in the accompanying drawings. Although the invention will be described in conjunction with specific embodiments, it will be understood that it is not intended to limit the invention to the described embodiments. Rather, it is intended to cover variations, modifications, and equivalents included within the spirit and scope of the invention as defined by the appended claims. It should be noted that the method steps described herein can be implemented by any functional block or functional arrangement, and any functional block or functional arrangement can be implemented as a physical entity or a logical entity, or a combination of both.
[0048] To enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0049] Note: The examples described below are merely specific examples and are not intended to limit the embodiments of the present invention to the specific steps, values, conditions, data, order, etc. Those skilled in the art can utilize the concept of the present invention to construct more embodiments not mentioned herein by reading this specification.
[0050] See Figure 1 As shown, this embodiment of the invention provides a zero-speed detection method for a wheeled robot, specifically including the following steps:
[0051] S100, acquires inertial sensor data of the wheeled robot;
[0052] S200, select the inertial sensor data based on a preset length sliding window, and perform modulus calculation on the inertial sensor data selected by the sliding window to obtain one-dimensional inertial data;
[0053] S300, standardize the one-dimensional inertial data to obtain standardized data;
[0054] S400, Perform a discrete Fourier transform on the standardized data to obtain the Fourier spectrum;
[0055] S500, based on a preset one-dimensional convolutional neural network, extract time-domain feature data and frequency-domain feature data from the standardized data and the Fourier spectrum;
[0056] S600, zero-speed detection is performed on the time-domain feature data and the frequency-domain feature data based on a preset fully connected neural network.
[0057] Specifically, in this embodiment, the present invention utilizes inertial sensor data (read by inertial measurement unit) of a wheeled robot selected based on a preset length sliding window. After modulus calculation, standardization processing and discrete Fourier transform, the data is input into a zero-speed detection deep learning model - a preset one-dimensional convolutional neural network to automatically extract time-domain features and frequency-domain features and detect them. Finally, the zero-speed detection result is obtained based on a preset fully connected neural network.
[0058] Therefore, by using a neural network to automatically extract and fuse the time-domain and frequency-domain features of the data, this invention provides a high-performance zero-speed detection method that does not rely on a pre-set threshold, thereby improving the zero-speed detection capability of wheeled robots in different motion modes and reducing the impact of engine and other component vibrations on the zero-speed detection effect when the robot is parked.
[0059] Preferably, in another embodiment of this application, the inertial sensor data includes triaxial specific force and triaxial angular velocity;
[0060] The formula for calculating the modulus f of triaxial specific force is as follows:
[0061]
[0062] The formula for calculating the magnitude ω of the triaxial angular velocity is as follows:
[0063]
[0064] In the formula, f x f y f z The specific force is the force along the three axes; ω x ω y ω z It represents the angular velocity of the three axes.
[0065] Specifically, in this embodiment, the inertial sensor data includes triaxial specific force and triaxial angular velocity. The modulus values of the specific force and angular velocity are taken to obtain two segments of one-dimensional inertial data of length N (i.e., the length of the preset sliding window), as follows:
[0066] f = {f0, f1, ..., f} N-1},ω={ω0,ω1,…ω N-1}
[0067] It should be noted that specific force is the difference between the absolute acceleration aI of the carrier relative to inertial space and the gravitational acceleration G. Specific force can be measured by an accelerometer; that is, the sensitive mass of the accelerometer is sensitive to the difference between absolute acceleration and gravitational acceleration.
[0068] Preferably, in another embodiment of this application, the standardization calculation formula is as follows:
[0069]
[0070] In the formula, Standardized data obtained by standardizing one-dimensional inertial data x. Let x be the mean value of the one-dimensional inertial data, and σ(x) be the standard deviation of the one-dimensional inertial data x.
[0071] Specifically, in this embodiment, in order to adapt the one-dimensional inertial data to the input requirements of the neural network, f and ω need to be standardized to obtain standardized data. and
[0072] Preferably, in another embodiment of this application, the discrete Fourier transform formula is as follows:
[0073]
[0074] In the formula, X(k) is the (k+1)th value of the Fourier spectrum obtained after the discrete Fourier transform, x nThis represents the (n+1)th value before the standardized data transformation, where e is the natural constant and j is the imaginary unit.
[0075] Specifically, in this embodiment, the two segments of one-dimensional inertial data are standardized. and The corresponding Fourier spectra obtained by Discrete Fourier Transform are as follows:
[0076] F = {F0, F1, ..., F} N-1},ω={W0,W1,…W N-1}
[0077] See also Figure 2 As shown, preferably, in another embodiment of this application, the step "S500, extracting time-domain feature data and frequency-domain feature data from the standardized data and the Fourier spectrum based on a preset one-dimensional convolutional neural network" specifically includes the following steps:
[0078] The preset one-dimensional convolutional neural network includes a preset first one-dimensional convolutional neural network and a preset second one-dimensional convolutional neural network;
[0079] Temporal feature data is extracted from the standardized data based on a preset first one-dimensional convolutional neural network.
[0080] Frequency domain feature data is extracted from the Fourier spectrum based on a preset second one-dimensional convolutional neural network.
[0081] Specifically, in this embodiment, and The N×2 matrix obtained after stacking is input into the first one-dimensional convolutional neural network of the 1DCNN to extract temporal features. The N×2 matrix obtained after stacking F and W is input into the second one-dimensional convolutional neural network of the 1DCNN to extract frequency domain features. These two 1DCNN networks are trained based on the historical inertial sensor data of the wheeled robot and its corresponding zero-velocity label data. The two networks can have different parameters, such as stride, number of network layers, number of convolutional kernels, etc.
[0082] The input length of the first one-dimensional convolutional neural network is N×2, and it has 4 convolutional CNN layers with strides of 16, 8, 3, and 3 respectively. Similarly, the first one-dimensional convolutional neural network can also use the same first one-dimensional convolutional neural network.
[0083] Preferably, in another embodiment of this application, the preset fully connected neural network formula is as follows:
[0084] x output =max(A1x input +b1,0) Equation (5);
[0085] x input The data consists of time-domain feature data and frequency-domain feature data, with a length of [length missing]. m is the number of 1DCNN filters in the last feature extraction layer, N is the preset sliding window length, n is the number of max pooling layers in the network, and x output The zero-speed output is a one-dimensional vector of length 20; A1 is the transformation matrix with dimension 1. b1 is a one-dimensional vector of length 20 representing the deviation.
[0086] Specifically, in this embodiment, the preset fully connected neural network includes a preset first fully connected neural network and a preset second fully connected neural network; both the first fully connected neural network and the preset second fully connected neural network can be represented by the above equation (v), only the vector dimensions are different; the preset first fully connected neural network (fully connected layer 1) searches for nonlinear relationships in the time-domain features and frequency-domain features extracted from the feature extraction layer. It has 20 neurons and uses the ReLU function as the activation function; the preset second fully connected neural network (fully connected layer 2) can obtain a 2*1 row vector, which represents the confidence levels of zero speed and non-zero speed respectively. For example, if (0.8, 0.2), the detection result is judged to be zero speed. Therefore, the preset second fully connected neural network outputs the detection result: in a zero-speed state or not in a zero-speed state.
[0087] It should be noted that the preset fully connected neural network was trained using historical inertial sensor data of the wheeled robot and its corresponding zero-velocity label data.
[0088] See also Figure 3 As shown, this embodiment of the invention also provides a zero-speed detection system for a wheeled robot, comprising:
[0089] The data acquisition module acquires inertial sensor data from the wheeled robot.
[0090] The sliding window module is communicatively connected to the data acquisition module. It is used to select the inertial sensor data based on a preset length sliding window and to perform modulus calculation on the inertial sensor data selected by the sliding window to obtain one-dimensional inertial data.
[0091] The standardization module, which is communicatively connected to the sliding window module, is used to standardize the one-dimensional inertial data and obtain standardized data.
[0092] The Fourier module, which is communicatively connected to the standardization module, is used to perform discrete Fourier transform on the standardized data to obtain the Fourier spectrum.
[0093] The feature extraction module is communicatively connected to the standardization module and the Fourier module, and is used to extract time-domain feature data and frequency-domain feature data from the standardized data and the Fourier spectrum based on a preset one-dimensional convolutional neural network.
[0094] The zero-speed detection module is communicatively connected to the feature extraction module and is used to perform zero-speed detection on the time-domain feature data and the frequency-domain feature data based on a preset fully connected neural network.
[0095] Therefore, by using a neural network to automatically extract and fuse the time-domain and frequency-domain features of the data, this invention provides a high-performance zero-speed detection method that does not rely on a pre-set threshold, thereby improving the zero-speed detection capability of wheeled robots in different motion modes and reducing the impact of engine and other component vibrations on the zero-speed detection effect when the robot is parked.
[0096] Specifically, this embodiment corresponds one-to-one with the above method embodiments. The functions of each module have been described in detail in the corresponding method embodiments, so they will not be repeated here.
[0097] Based on the same inventive concept, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements all or part of the method steps of the above method.
[0098] The present invention can implement all or part of the processes in the above methods, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.
[0099] Based on the same inventive concept, embodiments of this application also provide an electronic device, including a memory and a processor. The memory stores a computer program that runs on the processor. When the processor executes the computer program, it implements all or part of the method steps described above.
[0100] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the computer device, connecting all parts of the computer device through various interfaces and lines.
[0101] Memory can be used to store computer programs and / or modules. The processor performs various functions of the computer device by running or executing the computer programs and / or modules stored in the memory, and by accessing data stored in the memory. Memory can primarily include a program storage area and a data storage area. The program storage area can store the operating system and at least one application program required for a function (e.g., sound playback, image playback, etc.); the data storage area can store data created based on the use of the mobile phone (e.g., audio data, video data, etc.). Furthermore, memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, SmartMedia Cards (SMC), Secure Digital (SD) cards, Flash Cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.
[0102] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, servers, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.
[0103] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), servers, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0104] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0105] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0106] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
Claims
1. A method for detecting zero speed of a wheeled robot, characterized by, Specifically, the following steps are included: Acquire inertial sensor data for wheeled robots; The inertial sensor data is selected based on a preset length sliding window, and the modulus of the selected inertial sensor data is calculated to obtain one-dimensional inertial data. The one-dimensional inertial data is standardized to obtain standardized data; Perform a discrete Fourier transform on the standardized data to obtain the Fourier spectrum; Based on a preset one-dimensional convolutional neural network, time-domain feature data and frequency-domain feature data are extracted from the standardized data and the Fourier spectrum. Zero-speed detection is performed on the time-domain feature data and the frequency-domain feature data based on a preset fully connected neural network. The step of "extracting time-domain feature data and frequency-domain feature data from the standardized data and the Fourier spectrum based on a preset one-dimensional convolutional neural network" specifically includes the following steps: The preset one-dimensional convolutional neural network includes a preset first one-dimensional convolutional neural network and a preset second one-dimensional convolutional neural network; Temporal feature data is extracted from the standardized data based on a preset first one-dimensional convolutional neural network. Frequency domain feature data is extracted from the Fourier spectrum based on a preset second one-dimensional convolutional neural network. The preset fully connected neural network includes a preset first fully connected neural network and a preset second fully connected neural network; The first fully connected neural network is pre-set to find the nonlinear relationship between the time-domain features and frequency-domain features extracted from the feature extraction layer; The preset second fully connected neural network can obtain a 2*1 row vector, which represents the confidence of zero speed and non-zero speed respectively. The preset second fully connected neural network outputs the detection result: in the zero speed state or not in the zero speed state.
2. The wheeled robot zero velocity detection method of claim 1, wherein, The inertial sensor data includes triaxial specific force and triaxial angular velocity; The modulus of triaxial specific force The calculation formula is as follows: Formula (One); The modulus of the three-axis angular velocity The calculation formula is as follows: Formula (II); wherein , , is the triaxial ratio of forces; , , is the triaxial angular velocity.
3. The wheeled robot zero velocity detection method of claim 1, wherein, The standardized processing calculation formula is as follows: Formula (3); In the formula, One-dimensional inertial data Standardized data obtained through standardization processing One-dimensional inertial data The average value, One-dimensional inertial data The standard deviation.
4. The wheeled robot zero velocity detection method of claim 1, wherein, The formula for the Discrete Fourier Transform is as follows: Formula (IV); In the formula, The fourth term of the Fourier spectrum obtained after the discrete Fourier transform. A number, Represents the first data before standardized transformation. A number, It is a natural constant. It is the imaginary unit.
5. The zero-speed detection method for a wheeled robot as described in claim 1, characterized in that, The preset fully connected neural network formula is as follows: Formula (5); These are time-domain feature data and frequency-domain feature data; Output the result at zero speed; This is the transformation matrix; This is a deviation.
6. A wheeled robot zero speed detection system characterized by, include: The data acquisition module acquires inertial sensor data from the wheeled robot. The sliding window module is communicatively connected to the data acquisition module. It is used to select the inertial sensor data based on a preset length sliding window and to perform modulus calculation on the inertial sensor data selected by the sliding window to obtain one-dimensional inertial data. The standardization module, which is communicatively connected to the sliding window module, is used to standardize the one-dimensional inertial data and obtain standardized data. The Fourier module, which is communicatively connected to the standardization module, is used to perform discrete Fourier transform on the standardized data to obtain the Fourier spectrum. The feature extraction module is communicatively connected to the standardization module and the Fourier module, and is used to extract time-domain feature data and frequency-domain feature data from the standardized data and the Fourier spectrum based on a preset one-dimensional convolutional neural network. The zero-speed detection module is communicatively connected to the feature extraction module and is used to perform zero-speed detection on the time-domain feature data and the frequency-domain feature data based on a preset fully connected neural network. The step of "extracting time-domain feature data and frequency-domain feature data from the standardized data and the Fourier spectrum based on a preset one-dimensional convolutional neural network" specifically includes the following steps: The preset one-dimensional convolutional neural network includes a preset first one-dimensional convolutional neural network and a preset second one-dimensional convolutional neural network; Temporal feature data is extracted from the standardized data based on a preset first one-dimensional convolutional neural network. Frequency domain feature data is extracted from the Fourier spectrum based on a preset second one-dimensional convolutional neural network. The preset fully connected neural network includes a preset first fully connected neural network and a preset second fully connected neural network; The first fully connected neural network is pre-set to find the nonlinear relationship between the time-domain features and frequency-domain features extracted from the feature extraction layer; The preset second fully connected neural network can obtain a 2*1 row vector, which represents the confidence of zero speed and non-zero speed respectively. The preset second fully connected neural network outputs the detection result: in the zero speed state or not in the zero speed state.
7. A storage medium having stored thereon a computer program, characterized in that When the computer program is executed by the processor, it implements the zero-speed detection method for wheeled robots as described in any one of claims 1 to 5.
8. An electronic device comprising a storage medium, a processor, and a computer program stored in the storage medium and operable on the processor, characterized in that, When the processor runs the computer program, it implements the zero-speed detection method for wheeled robots as described in any one of claims 1 to 5.