WO2021258698A1 - 扫地机数据处理方法、装置、设备及计算机可读介质 - Google Patents

扫地机数据处理方法、装置、设备及计算机可读介质 Download PDF

Info

Publication number
WO2021258698A1
WO2021258698A1 PCT/CN2020/140491 CN2020140491W WO2021258698A1 WO 2021258698 A1 WO2021258698 A1 WO 2021258698A1 CN 2020140491 W CN2020140491 W CN 2020140491W WO 2021258698 A1 WO2021258698 A1 WO 2021258698A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
support vector
vector machine
material type
sweeper
Prior art date
Application number
PCT/CN2020/140491
Other languages
English (en)
French (fr)
Inventor
谭泽汉
朱莹莹
Original Assignee
珠海格力电器股份有限公司
珠海联云科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 珠海格力电器股份有限公司, 珠海联云科技有限公司 filed Critical 珠海格力电器股份有限公司
Publication of WO2021258698A1 publication Critical patent/WO2021258698A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
    • A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
    • A47L11/00—Machines for cleaning floors, carpets, furniture, walls, or wall coverings
    • A47L11/24—Floor-sweeping machines, motor-driven
    • A—HUMAN NECESSITIES
    • A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
    • A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
    • A47L11/00—Machines for cleaning floors, carpets, furniture, walls, or wall coverings
    • A47L11/40—Parts or details of machines not provided for in groups A47L11/02 - A47L11/38, or not restricted to one of these groups, e.g. handles, arrangements of switches, skirts, buffers, levers
    • A—HUMAN NECESSITIES
    • A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
    • A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
    • A47L11/00—Machines for cleaning floors, carpets, furniture, walls, or wall coverings
    • A47L11/40—Parts or details of machines not provided for in groups A47L11/02 - A47L11/38, or not restricted to one of these groups, e.g. handles, arrangements of switches, skirts, buffers, levers
    • A47L11/4011—Regulation of the cleaning machine by electric means; Control systems and remote control systems therefor
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines

Definitions

  • the present disclosure relates to the technical field of sweeping robots, and in particular to a sweeping machine data processing method, device, equipment and computer readable medium.
  • the present disclosure provides a data processing method, device, equipment, and computer readable medium of a sweeper to solve the above-mentioned technical problem of "software obstacles cannot be identified”.
  • the present disclosure provides a data processing method for a sweeper, including: acquiring a reflection signal received by a millimeter-wave radar sensor installed on the sweeper, and the reflected signal is formed by reflecting on an object after the millimeter-wave radar sensor sends a transmission signal.
  • the signal extract the signal feature in the reflected signal, the signal feature is set to characterize the fluctuation information of the reflected signal; use the first support vector machine to identify the signal feature; determine the object's identity according to the recognition result of the signal feature by the first support vector machine Material type, the first support vector machine is obtained by training the second support vector machine with the training data with marking information, the marking information is set as the material type of the marking training data, and the recognition result is set as indicating the object and each material Type of association.
  • extracting the signal characteristics in the reflected signal includes: extracting the signal-to-noise ratio in the reflected signal; and performing time-domain conversion and/or frequency-domain conversion on the signal-to-noise ratio to obtain the signal characteristics.
  • determining the material type of the object according to the recognition result of the signal feature of the first support vector machine includes: obtaining the recognition result output by the first support vector machine, and the recognition result includes the predicted value of the object belonging to each material type; The maximum value among the values is used as the final recognition result, and the material type indicated by the maximum value is used as the final material type of the object.
  • the method before determining the material type of the object according to the recognition result of the signal feature of the first support vector machine, the method further includes: initializing various parameters in the second support vector machine through the training data to obtain the third support vector Machine; in the case that the recognition accuracy of the test data by the third support vector machine reaches the target threshold, the third support vector machine is used as the first support vector machine; the recognition accuracy of the test data by the third support vector machine is not reached In the case of the target threshold, continue to use the training data to train the third support vector machine to adjust the values of the parameters in the third support vector machine until the recognition accuracy of the test data by the third support vector machine reaches the target threshold.
  • the method further includes: in the case that the material type of the object is a preset material type, controlling the sweeper to follow the predetermined material type. Set the target mode action that matches the material type.
  • controlling the sweeper to act in a target manner matching the preset material type includes: extracting reflection signals
  • the distance information, angle information and radial velocity resolution of the signal features include distance information, angle information and radial velocity resolution; the position of the object is determined according to the distance information, angle information and radial velocity resolution, and the object’s position is identified
  • establishing a virtual wall at the position of the object further includes: obtaining the first size of the first shape; obtaining the product of the first size and the expansion coefficient to obtain the second size; Build a virtual wall.
  • controlling the sweeper to act in a target manner matching the preset material type further includes: controlling the sweeper Crash the object so that the sweeper sweeps along the edge of the object.
  • the present disclosure provides a data processing device for a sweeper, including: a signal acquisition module configured to obtain a reflection signal received by a millimeter-wave radar sensor installed on the sweeper, and the reflected signal is sent and transmitted by the millimeter-wave radar sensor The signal is reflected on the object after the signal; the feature extraction module is set to extract signal features in the reflected signal, and the signal feature is set to characterize the fluctuation information of the reflected signal; the recognition module is set to use the first support vector machine to pair The signal feature is recognized; the material discrimination module is set to determine the material type of the object according to the recognition result of the signal feature by the first support vector machine.
  • the first support vector machine uses the training data with label information to perform the second support vector machine. After training, the marking information is set as the material type of the marking training data, and the recognition result is set to indicate the association relationship between the object and each material type.
  • the present disclosure provides a computer device including a memory and a processor.
  • the memory stores a computer program that can run on the processor.
  • the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented. .
  • the present disclosure also provides a computer-readable medium having non-volatile program code executable by the processor, the program code causing the processor to execute any method of the first aspect described above.
  • the present disclosure obtains the reflected signal received by the millimeter wave radar sensor installed on the sweeper, extracts the signal characteristics in the reflected signal, uses the first support vector machine to identify the signal characteristics, and recognizes the signal characteristics according to the first support vector machine
  • the technical solution for determining the material type of the object can detect objects of different materials through millimeter wave radar, identify soft materials such as water surface, plastic bags, rags, and pet feces, and avoid obstacles in time.
  • there is no need to carry a visual sensor which can speed up the processing speed, reduce the area occupied by the sweeper, and solve the user's privacy problem.
  • FIG. 1 is a schematic diagram of the hardware environment of an optional sweeper data processing method according to an embodiment of the present disclosure
  • FIG. 2 is a flowchart of an optional data processing method for a sweeper according to an embodiment of the present disclosure
  • Fig. 3 is a block diagram of an optional data processing device for a sweeper according to an embodiment of the present disclosure.
  • an embodiment of a data processing method for a sweeper is provided.
  • the above-mentioned sweeping machine data processing method can be applied to the hardware environment constituted by the terminal 101 and the server 103 as shown in FIG. 1.
  • the server 103 is connected to the terminal 101 through the network, and can be set to provide services for the terminal or the client installed on the terminal.
  • the database 105 can be set on the server or independent of the server, and it can be set as the server 103.
  • the aforementioned networks include but are not limited to: wide area networks, metropolitan area networks, or local area networks, and the terminal 101 includes, but is not limited to, smart sweepers.
  • a data processing method for a sweeper in an embodiment of the present disclosure may be executed by the server 103, or may be executed by the server 103 and the terminal 101 together. As shown in FIG. 2, the method may include the following steps:
  • Step S202 Obtain the reflected signal received by the millimeter wave radar sensor installed on the sweeper, where the reflected signal is a signal formed by reflection on the object after the millimeter wave radar sensor sends the transmitted signal.
  • the millimeter wave radar is a radar that works in the millimeter wave band (millimeter wave) detection.
  • millimeter wave refers to the frequency domain of 30 to 300 GHz (wavelength is 1 to 10 mm).
  • the wavelength of millimeter wave is between microwave and centimeter wave, so millimeter wave radar has some advantages of microwave radar and photoelectric radar.
  • the millimeter waveguide seeker is small in size, light in weight, and high in spatial resolution, and can distinguish and identify very small targets. , And can identify multiple targets at the same time, with imaging capabilities.
  • the millimeter-wave radar mentioned in the embodiments of the present disclosure refers to a millimeter-wave radar module, which mainly includes: an antenna, a transceiver module, a signal processing module, etc., which are installed on a sweeper and can detect the distance and distance of obstacles in various environments.
  • the angle and the rate of change of the distance between the electromagnetic wave emitted by the radar and the obstacle (that is, the radial velocity resolution) can reduce false detections, provide high-precision position and travel route data, and maintain the privacy of data collection.
  • the integrated module concentrates all features on a single chip, enabling local processing.
  • the millimeter-wave radar module can provide distance, angle, speed, signal-to-noise ratio at the same time, and can accurately confirm the location of the object, the sweeping direction of the sweeper, and the material of the obstacle in front of it.
  • a processing kernel can also be embedded in the millimeter wave radar sensor to process data in real time, and realize functions such as classifying the material of the object based on the size and motion state of the object in real time.
  • Step S204 Extract signal characteristics in the reflected signal, and the signal characteristics are set to represent fluctuation information of the reflected signal.
  • the signals reflected by objects of different material types have different fluctuation ranges, and the signal characteristics that characterize the fluctuation information of the reflected signals are extracted, and the material of the object can be identified by using the characteristics.
  • the signal characteristics may include signal-to-noise ratio, angle information, distance information, and radial velocity resolution.
  • Step S206 Use the first support vector machine to identify the signal characteristics.
  • the support vector machine is a machine learning method based on statistical learning theory, which shows excellent performance for small sample situations. Dimensionality, local minima and other issues.
  • the good performance of the support vector machine for classification problems is used, the support vector machine is used to identify signal features, and the neural network model and other classification algorithms can also be used, which will not be repeated here.
  • the first support vector machine is used to identify signal features, which can identify the signal-to-noise ratio, and can also identify the combination of signal-to-noise ratio, angle information, distance information, and radial velocity resolution.
  • the signal-to-noise ratio is an indispensable identification item.
  • different weights can be assigned to each feature.
  • the signal-to-noise ratio has the highest weight, and the weight of each feature parameter can be adjusted according to the actual situation.
  • Step S208 Determine the material type of the object according to the recognition result of the signal feature by the first support vector machine.
  • the first support vector machine is obtained after training the second support vector machine using the training data with marking information, and the marking information is set To mark the material type of the training data, the recognition result is set to indicate the association relationship between the object and each material type.
  • the marking information at least identifies the material type of each training data.
  • the recognition result is at least the probability value of the training data belonging to each material type, which is used to indicate the material type to which it belongs.
  • This disclosure uses millimeter-wave radar signals to detect obstacles, and combines support vector machine classification technology to identify and determine the material types of obstacles. Since millimeter-wave radars do not generate image photosensitive data, not only can the material types of obstacles be obtained in recognition, but also It can also protect user privacy, solve the problem of low privacy security in related technologies, and achieve the technical effect of not requiring the use of visual sensors to identify the type of obstacle material and protecting user privacy.
  • the present disclosure proposes a method for extracting the signal-to-noise ratio in the reflected signal for identification by the support vector machine.
  • the technical solution of the present disclosure will be further detailed below in conjunction with the steps shown in FIG. 2.
  • extracting signal characteristics from the reflected signal may include:
  • Step 1 Extract the signal-to-noise ratio in the reflected signal
  • Step 2 Perform time domain conversion and/or frequency domain conversion on the signal-to-noise ratio to obtain signal characteristics.
  • the transmitted signal of the millimeter wave radar will reflect signals with different signal-to-noise ratio values on objects of different materials, and the fluctuation range of the signal-to-noise ratio values of different materials is different, which is an important basis for distinguishing each material .
  • the signal-to-noise ratio value of each material can be converted between time domain and frequency domain as the signal feature.
  • the way to extract the signal-to-noise ratio is:
  • pt is the transmit output power
  • G Rx and G Tx are the gains of the RX and TX antennas
  • c is the speed of light
  • ⁇ is the cross-section of the object illuminated by the radar
  • RCS is the measure of the energy reflected by the object, which determines The detectability of the radar sensor to the object
  • N is the number of pulse frequency modulation
  • T r is the pulse frequency modulation time
  • k is the Boltzmann constant
  • T is the ambient temperature
  • NF is the receiver noise figure
  • SNR det is the algorithm detection The value of the minimum signal-to-noise ratio required by the object.
  • the present disclosure also proposes a method for determining the type of obstacle material by using the recognition result of the signal feature of the support vector machine.
  • the technical solution of the present disclosure will be further detailed in conjunction with the steps shown in FIG. 2 below.
  • determining the material type of the object according to the recognition result of the signal feature by the first support vector machine may include the following steps:
  • Step 1 Obtain the recognition result output by the first support vector machine, and the recognition result includes the predicted value of the object belonging to each material type;
  • Step 2 Use the maximum value of the predicted value as the final recognition result, and use the material type indicated by the maximum value as the final material type of the object.
  • the pre-trained first support vector machine predicts the probability that the obstacle belongs to each material type according to the fluctuation information in the signal feature, and uses the material type with the highest probability as the final material type of the obstacle.
  • the present disclosure also proposes a method for training the first support vector machine used in the embodiments of the present disclosure.
  • the method before determining the material type of the object according to the recognition result of the signal feature of the first support vector machine, the method further includes: initializing various parameters in the second support vector machine through the training data to obtain the third support vector Machine; in the case that the recognition accuracy of the test data by the third support vector machine reaches the target threshold, the third support vector machine is used as the first support vector machine; the recognition accuracy of the test data by the third support vector machine is not reached In the case of the target threshold, continue to use the training data to train the third support vector machine to adjust the values of the parameters in the third support vector machine until the recognition accuracy of the test data by the third support vector machine reaches the target threshold.
  • the reflected signals of multiple millimeter wave radars transmitted on multiple obstacles can be obtained as training samples, and each training sample includes the signal-to-noise ratio, angle information, and distance information of the training sample. And radial velocity resolution, etc., and mark the material type of the obstacle corresponding to the training sample.
  • the training data can come from the actual home scene, collect multiple sets of data, according to the type of material, that is, according to the hard objects (floor, water surface, wood, metal) and soft substances (wires, plastic bags, rags, pet feces) divided, and Labeling according to the category, you can also make adaptive adjustments according to actual needs.
  • the second support vector machine is initialized with the above training data to obtain a third support vector machine, and the third support vector machine is trained until the third support vector machine converges to obtain the first god support vector machine.
  • the above training of the third support vector machine until the third support vector machine converges may include:
  • the embodiments of the present disclosure may also use genetic optimization algorithms to optimize the parameters of the support vector machine, and store the optimized parameters.
  • the algorithm for optimizing the parameters of the support vector machine can also be an optimization algorithm such as ant colony.
  • the present disclosure also proposes a method for controlling the sweeper to further clean after the material type of the obstacle is determined according to the recognition result of the signal feature by the first support vector machine.
  • the method further includes:
  • the sweeper is controlled to act in a target manner matching the preset material type.
  • controlling the sweeper to act in a target manner matching the preset material type may include the following steps:
  • Step 1 Extract distance information, angle information, and radial velocity resolution from the reflected signal.
  • the signal features include distance information, angle information, and radial velocity resolution;
  • Step 2 Determine the position of the object according to the distance information, angle information and radial velocity resolution, and recognize the first shape of the object;
  • Step 3 Establish a virtual wall at the position of the object, and the virtual wall is distributed along the edge of the circumscribed rectangle of the first shape;
  • Step 4 Control the sweeper to clean along the virtual wall according to the traveling speed of the sweeper.
  • the obstacles made of soft materials need to be bypassed to avoid damage to the sweeper or even secondary pollution after contact.
  • the obstacles made of soft materials cannot trigger the sensors around the body of the sweeper, so the virtual Wall, to simulate the sweeping machine colliding with the virtual wall to change the direction of travel.
  • the distance information, angle information and radial velocity resolution in the reflected signal can roughly identify the location and approximate shape of the obstacle, determine the circumscribed rectangle of the shape according to the approximate shape of the obstacle, and use the circumscribed rectangle to generate a virtual wall to ensure Enclose the entire obstacle, or you can further expand the scope of the virtual wall according to actual needs.
  • control the sweeper to "collide" the virtual wall and sweep along the virtual wall.
  • establishing a virtual wall at the location of the object may further include the following steps:
  • Step 1 Obtain the first size of the first shape
  • Step 2 Obtain the product of the first size and the expansion coefficient to obtain the second size
  • Step 3 Build a virtual wall at the position of the object according to the second size.
  • the virtual walls generated by obstacles of different soft materials are different.
  • the virtual wall is generated according to the general shape of the obstacle, but the range of the virtual wall can be larger than the general shape of the recognized obstacle, so the obstacle can be changed.
  • the first shape (approximate shape) of is multiplied by the expansion coefficient to expand and expand the first shape to generate a virtual wall when the obstacle is completely enclosed.
  • the expansion coefficient can be preset to 1.0 to 1.5.
  • controlling the sweeper to act in a target manner matching the preset material type further includes: controlling the sweeper Crash the object so that the sweeper sweeps along the edge of the object.
  • the sweeper is directly controlled to collide with the obstacle, and sensors around the body of the sweeper are triggered to make the sweeper sweep along the edge of the obstacle.
  • Step 1 The sweeper starts to walk
  • Step 2 The millimeter-wave radar collects obstacle information during the sweeping process
  • Step 3 The sweeper controller extracts signal characteristics and analyzes the data
  • Step 4 Assign weights to each signal feature and perform data fusion
  • Step 5 Use the classifier (first support vector machine) to obtain the obstacle material type;
  • Step 6 feedback to the sweeper
  • Step 7 Control the sweeper to clean according to the target method.
  • the present disclosure obtains the reflected signal received by the millimeter wave radar sensor installed on the sweeper, extracts the signal characteristics in the reflected signal, uses the first support vector machine to identify the signal characteristics, and recognizes the signal characteristics according to the first support vector machine
  • the technical solution for determining the material type of the object can detect objects of different materials through millimeter wave radar, identify soft materials such as water surface, plastic bags, rags, and pet feces, and avoid obstacles in time.
  • there is no need to carry a visual sensor which can speed up the processing speed, reduce the area occupied by the sweeper, and solve the user's privacy problem.
  • a data processing device for a sweeper including: a signal acquisition module 301 configured to obtain reflections received by a millimeter wave radar sensor installed on the sweeper The reflected signal is the signal formed by the reflection on the object after the millimeter wave radar sensor sends the transmitted signal; the feature extraction module 303 is set to extract the signal feature in the reflected signal, and the signal feature is set to represent the fluctuation information of the reflected signal; identification The module 305 is set to use the first support vector machine to recognize signal features; the material discrimination module 307 is set to determine the material type of the object according to the recognition result of the signal feature by the first support vector machine.
  • the first support vector machine is It is obtained after training the second support vector machine with the training data with the marking information, the marking information is set as the material type of the marking training data, and the recognition result is set as indicating the association relationship between the object and each material type.
  • the signal acquisition module 301 in this embodiment can be set to perform step S202 in the embodiment of the present disclosure
  • the feature extraction module 303 in this embodiment can be set to perform step S204 in the embodiment of the present disclosure
  • the identification module 305 in this embodiment can be configured to perform step S206 in the embodiment of the present disclosure
  • the material discrimination module 307 in this embodiment can be configured to perform step S208 in the embodiment of the present disclosure.
  • the sweeper data processing device further includes: a signal-to-noise ratio extraction module, which is configured to extract the signal-to-noise ratio in the reflected signal; and a conversion module, which is configured to perform time-domain conversion and/ Or frequency domain conversion to obtain signal characteristics.
  • a signal-to-noise ratio extraction module which is configured to extract the signal-to-noise ratio in the reflected signal
  • a conversion module which is configured to perform time-domain conversion and/ Or frequency domain conversion to obtain signal characteristics.
  • the sweeper data processing device further includes: a recognition result obtaining module configured to obtain the recognition result output by the first support vector machine, the recognition result including the predicted value of the object belonging to each material type; the result determination module , Is set to take the maximum value of the predicted value as the final recognition result, and use the material type indicated by the maximum value as the final material type of the object.
  • a recognition result obtaining module configured to obtain the recognition result output by the first support vector machine, the recognition result including the predicted value of the object belonging to each material type
  • the result determination module Is set to take the maximum value of the predicted value as the final recognition result, and use the material type indicated by the maximum value as the final material type of the object.
  • the sweeper data processing device further includes: a first training module configured to initialize various parameters in the second support vector machine through training data to obtain a third support vector machine; a second training module , Is set to use the third support vector machine as the first support vector machine when the recognition accuracy of the test data by the third support vector machine reaches the target threshold; the third training module is set to be in the third support vector When the recognition accuracy of the test data by the machine does not reach the target threshold, continue to use the training data to train the third support vector machine to adjust the values of the parameters in the third support vector machine until the third support vector machine tests the test data. The recognition accuracy of the data reaches the target threshold.
  • the sweeper data processing device further includes: a control module configured to control the sweeper to act in a target manner matching the preset material type when the material type of the object is a preset material type .
  • the sweeper data processing device further includes: a signal feature extraction module configured to extract distance information, angle information, and radial velocity resolution in the reflected signal, and the signal feature includes distance information, angle information, and Radial velocity resolution;
  • the position and shape determination module is set to determine the position of the object based on the distance information, angle information and the radial velocity resolution, and the first shape of the object is recognized;
  • the virtual wall creation module is set to A virtual wall is established at the position of the object, and the virtual wall is distributed along the edge of the circumscribed rectangle of the first shape;
  • the first action module is set to control the sweeper to clean along the virtual wall according to the traveling speed of the sweeper.
  • the sweeper data processing device further includes: a first size obtaining module configured to obtain the first size of the first shape; and a second size obtaining module configured to obtain the first size and the expansion coefficient The product of, obtains the second size; the virtual wall determination module is set to establish a virtual wall at the position of the object according to the second size.
  • the sweeper data processing device further includes: a second action module configured to control the sweeper to collide with the object, so that the sweeper cleans along the edge of the object.
  • a computer device including a memory and a processor, the memory stores a computer program that can run on the processor, and the processor executes the computer program When implementing the above steps.
  • the memory and processor in the above-mentioned computer equipment communicate through a communication bus and a communication interface.
  • the communication bus may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc.
  • PCI Peripheral Component Interconnect
  • EISA Extended Industry Standard Architecture
  • the communication bus can be divided into address bus, data bus, control bus and so on.
  • the memory may include random access memory (Random Access Memory, RAM for short), and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
  • non-volatile memory such as at least one disk memory.
  • the memory may also be at least one storage device located far away from the foregoing processor.
  • the above-mentioned processor can be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP), etc.; it can also be a digital signal processor (Digital Signal Processing, DSP for short) , Application Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.
  • CPU Central Processing Unit
  • NP Network Processor
  • DSP Digital Signal Processing
  • ASIC Application Specific Integrated Circuit
  • FPGA Field-Programmable Gate Array
  • a computer-readable medium having non-volatile program code executable by a processor.
  • the computer-readable medium is configured to store program code for the processor to execute the following steps:
  • Step S202 Obtain the reflected signal received by the millimeter-wave radar sensor installed on the sweeper, where the reflected signal is a signal formed by reflection on the object after the millimeter-wave radar sensor sends the transmitted signal;
  • Step S204 extract the signal feature in the reflected signal, and the signal feature is set to characterize the fluctuation information of the reflected signal;
  • Step S206 using the first support vector machine to identify signal features
  • Step S208 Determine the material type of the object according to the recognition result of the signal feature by the first support vector machine.
  • the first support vector machine is obtained after training the second support vector machine using the training data with marking information, and the marking information is set To mark the material type of the training data, the recognition result is set to indicate the association relationship between the object and each material type.
  • the embodiments described herein can be implemented by hardware, software, firmware, middleware, microcode, or a combination thereof.
  • the processing unit can be implemented in one or more application specific integrated circuits (ASIC), digital signal processor (Digital Signal Processing, DSP), digital signal processing equipment (DSP Device, DSPD), programmable Logic devices (Programmable Logic Device, PLD), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA), general-purpose processors, controllers, microcontrollers, microprocessors, those that are set to perform the functions described in this disclosure Other electronic units or their combination.
  • ASIC application specific integrated circuits
  • DSP Digital Signal Processing
  • DSP Device digital signal processing equipment
  • PLD programmable Logic Device
  • PLD Field-Programmable Gate Array
  • FPGA Field-Programmable Gate Array
  • the technology described herein can be implemented by a unit that performs the functions described herein.
  • the software codes can be stored in the memory and executed by the processor.
  • the memory can be implemented in the processor or external to the processor.
  • the disclosed device and method may be implemented in other ways.
  • the device embodiments described above are only illustrative.
  • the division of the modules is only a logical function division, and there may be other divisions in actual implementation, for example, multiple modules or components may be combined or It can be integrated into another system, or some features can be ignored or not implemented.
  • the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components 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 units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional units in the various embodiments of the present disclosure may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium.
  • the technical solutions of the embodiments of the present disclosure can essentially or contribute to related technologies or parts of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, It includes several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present disclosure.
  • the aforementioned storage media include: U disk, mobile hard disk, ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

一种扫地机数据处理方法、装置、设备及计算机可读介质。该方法包括：获取安装在扫地机上的毫米波雷达传感器接收到的反射信号（S202）；提取反射信号中的信号特征，信号特征被设置为表征反射信号的波动信息（S204）；利用第一支持向量机对信号特征进行识别（S206）；根据第一支持向量机对信号特征的识别结果确定物体的材质类型（S208）。该方法可以通过毫米波雷达检测不同材质的物体，识别水面、塑料袋、抹布、宠物粪便等软体材质物，及时避障。无需搭载视觉传感器，可以加快处理速度，减少扫地机占用面积，同时解决用户的隐私问题。

Description

扫地机数据处理方法、装置、设备及计算机可读介质
本公开要求于2020年06月22日提交中国专利局、申请号为202010576664.8、发明名称为“扫地机数据处理方法、装置、设备及计算机可读介质”的中国专利申请的优先权，其全部内容通过引用结合在本公开中。
技术领域
本公开涉及扫地机器人技术领域，尤其涉及一种扫地机数据处理方法、装置、设备及计算机可读介质。
背景技术
随着社会的不断发展，人们日渐趋向于生活品质的追求，这促使了扫地机器人进入千家万户。扫地机器人在清扫过程中，软体障碍物(没有固定形状，质量较轻的障碍物，例如拖鞋、袜子、电线、抹布等)会给扫地机器人造成清扫障碍，因为轻体障碍物由于质地轻柔，无法触发机身侧面的碰撞传感器，这种情况下普通扫地机无法处理复杂的家庭环境，造成缠绕袜子、电线、塑料袋等轻体障碍物，容易被困，造成扫地机损伤，甚至遇到宠物粪便会造成室内的二次污染。
目前，相关技术中，常常利用视觉传感器(摄像头)辅助机器人实现针对此类障碍物的避障功能，但软件和处理要求复杂，需要很强的处理器支撑，且增大了扫地机的体积、重量和占用面积，同时会给用户的隐私带来威胁，近年来有关摄像头的隐私泄露事件时有发生，不少人患上了“摄像头焦虑症”，在不使用视觉传感器辅助时又无法很好地识别软体障碍物。
针对上述的问题，目前尚未提出有效的解决方案。
发明内容
本公开提供了一种扫地机数据处理方法、装置、设备及计算机可读介质，以解决上述“不能识别软体障碍物”的技术问题。
第一方面，本公开提供了一种扫地机数据处理方法，包括：获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，反射信号为毫米波雷达传感器发送发射信号后在物体上反射形成的信号；提取反射信号中的信号特征，信号特征被设置为表征反射信号的波动信息；利用第一支持向量机对信号特征进行识别；根据第一支持向量机对信号特征的识别结果确定物体的材质类型，第一支持向量机是采用具有标记信息的训练数据对第二支持向量机进行训练后得到的，标记信息被设置为标记训练数据的材质类型，识别结果被设置为指示物体与各个材质类型的关联关系。
在一些实施方式中，提取反射信号中的信号特征包括：提取反射信号中的信噪比；将信噪比进行时域转换和/或频域转换，得到信号特征。
在一些实施方式中，根据第一支持向量机对信号特征的识别结果确定物体的材质类型包括：获取第一支持向量机输出的识别结果，识别结果包括物体属于各个材质类型的预测值；将预测值中的最大值作为最终的识别结果，并将最大值指示的材质类型作为物体最终的材质类型。
在一些实施方式中，根据第一支持向量机对信号特征的识别结果确定物体的材质类型之前，该方法还包括：通过训练数据对第二支持向量机内各参数进行初始化，得到第三支持向量机；在第三支持向量机对测试数据的识别准确度达到目标阈值的情况下，将第三支持向量机作为第一支持向量机；在第三支持向量机对测试数据的识别准确度未达到目标阈值的情况下，继续使用训练数据对第三支持向量机进行 训练，以调整第三支持向量机内各参数的数值，直至第三支持向量机对测试数据的识别准确度达到目标阈值。
在一些实施方式中，根据第一支持向量机对信号特征的识别结果确定物体的材质类型之后，该方法还包括：在物体的材质类型为预设材质类型的情况下，控制扫地机按照与预设材质类型匹配的目标方式动作。
在一些实施方式中，预设材质类型为软体材质类型时，在物体的材质类型为预设材质类型的情况下，控制扫地机按照与预设材质类型匹配的目标方式动作包括：提取反射信号中的距离信息、角度信息及径向速度分辨率，信号特征包括距离信息、角度信息及径向速度分辨率；根据距离信息、角度信息及径向速度分辨率确定物体的位置，并识别得到物体的第一形状；在物体的位置处建立虚拟墙，虚拟墙沿第一形状的外接矩形边分布；按照扫地机的行进速度，控制扫地机沿虚拟墙进行清扫。
在一些实施方式中，在物体的位置处建立虚拟墙还包括：获取第一形状的第一尺寸；获取第一尺寸与膨胀系数的乘积，得到第二尺寸；按照第二尺寸在物体的位置处建立虚拟墙。
在一些实施方式中，预设材质类型为硬质材质类型时，物体的材质类型为预设材质类型的情况下，控制扫地机按照与预设材质类型匹配的目标方式动作还包括：控制扫地机对物体进行碰撞，以使扫地机沿物体的边缘进行清扫。
第二方面，本公开提供了一种扫地机数据处理装置，包括：信号获取模块，被设置为获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，反射信号为毫米波雷达传感器发送发射信号后在物体上反射形成的信号；特征提取模块，被设置为提取反射信号中的信号特征，信号特征被设置为表征反射信号的波动信息；识别模块，被设置 为利用第一支持向量机对信号特征进行识别；材质判别模块，被设置为根据第一支持向量机对信号特征的识别结果确定物体的材质类型，第一支持向量机是采用具有标记信息的训练数据对第二支持向量机进行训练后得到的，标记信息被设置为标记训练数据的材质类型，识别结果被设置为指示物体与各个材质类型的关联关系。
第三方面，本公开提供了一种计算机设备，包括存储器、处理器，存储器中存储有可在处理器上运行的计算机程序，处理器执行计算机程序时实现上述第一方面任一项方法的步骤。
第四方面，本公开还提供了一种具有处理器可执行的非易失的程序代码的计算机可读介质，程序代码使处理器执行上述第一方面任一方法。
本公开实施例提供的上述技术方案与相关技术相比具有如下优点：
本公开通过获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，提取反射信号中的信号特征，利用第一支持向量机对信号特征进行识别，根据第一支持向量机对信号特征的识别结果确定物体的材质类型的技术方案，可以通过毫米波雷达检测不同材质的物体，识别水面、塑料袋、抹布、宠物粪便等软体材质物，及时避障。同时无需搭载视觉传感器，可以加快处理速度，减少扫地机占用面积，同时解决用户的隐私问题。
附图说明
此处的附图被并入说明书中并构成本说明书的一部分，示出了符合本公开的实施例，并与说明书一起用于解释本公开的原理。
为了更清楚地说明本公开实施例或相关技术中的技术方案，下面将对实施例或相关技术描述中所需要使用的附图作简单地介绍，显而 易见地，对于本领域普通技术人员而言，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1为根据本公开实施例提供的一种可选的扫地机数据处理方法硬件环境示意图；
图2为根据本公开实施例提供的一种可选的扫地机数据处理方法流程图；
图3为根据本公开实施例提供的一种可选的扫地机数据处理装置框图。
具体实施方式
为使本公开实施例的目的、技术方案和优点更加清楚，下面将结合本公开实施例中的附图，对本公开实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本公开的一部分实施例，而不是全部的实施例。基于本公开中的实施例，本领域普通技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例，都属于本公开保护的范围。
在后续的描述中，使用用于表示元件的诸如“模块”、“部件”或“单元”的后缀仅为了有利于本公开的说明，其本身并没有特定的意义。因此，“模块”与“部件”可以混合地使用。
相关技术中，虽然可以利用视觉传感器(摄像头)辅助机器人实现针对软体障碍物进行识别以进一步避障的功能，但软件和处理要求复杂，需要很强的处理器支撑，且增大了扫地机的体积、重量和占用面积，同时会给用户的隐私带来威胁，近年来有关摄像头的隐私泄露事件时有发生，不少人患上了“摄像头焦虑症”，在不使用视觉传感器辅助时又无法很好地识别软体障碍物。
根据本公开实施例的一方面，提供了一种扫地机数据处理方法的实施例。
在本公开实施例的一些实施方式中，上述扫地机数据处理方法可以应用于如图1所示的由终端101和服务器103所构成的硬件环境中。如图1所示，服务器103通过网络与终端101进行连接，可被设置为为终端或终端上安装的客户端提供服务，可在服务器上或独立于服务器设置数据库105，被设置为为服务器103提供数据存储服务，上述网络包括但不限于：广域网、城域网或局域网，终端101包括但不限于智能扫地机等。
本公开实施例中的一种扫地机数据处理方法可以由服务器103来执行，还可以是由服务器103和终端101共同执行，如图2所示，该方法可以包括以下步骤：
步骤S202，获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，反射信号为毫米波雷达传感器发送发射信号后在物体上反射形成的信号。
本公开实施例中，毫米波雷达，是工作在毫米波波段(millimeter wave)探测的雷达。通常毫米波是指30～300GHz频域(波长为1～10mm)的。毫米波的波长介于微波和厘米波之间，因此毫米波雷达兼有微波雷达和光电雷达的一些优点，毫米波导引头体积小、质量轻、空间分辨率高，能分辨识别很小的目标，而且能同时识别多个目标，具有成像能力。
本公开实施例中的提及到的毫米波雷达是指毫米波雷达模块，主要包括：天线、收发模块、信号处理模块等，安装在扫地机上，可以在各种环境下检测障碍物体的距离、角度及雷达发射电磁波与障碍物的距离变化率(即径向速度分辨率)，同时能够减少错误检测、提供高精度的位置和行进路线数据，保持数据采集的隐私性。同时，集成模 块将所有特性集中在单个芯片上，可以实现本地处理。毫米波雷达模块可以同时提供距离、角度、速度、信噪比，可以准确地确认物体位置、扫地机行走方向、前方障碍物材质等信息。本公开实施例中，还可以在毫米波雷达传感器中嵌入处理内核，以实时处理数据，实现实时基于物体大小和运动状态对物体材质进行分类等功能。
步骤S204，提取反射信号中的信号特征，信号特征被设置为表征反射信号的波动信息。
本公开实施例中，不同材质类型的物体所反射的信号具有不同的波动范围，提取表征反射信号的波动信息的信号特征，能够利用该特征识别得到物体的材质。
本公开实施例中，信号特征可以包括信噪比、角度信息、距离信息及径向速度分辨率。
步骤S206，利用第一支持向量机对信号特征进行识别。
本公开实施例中，支持向量机是一种基于统计学习理论的机器学习方法，针对小样本情况表现出了优良的性能，其建立在严格的理论基础之上，较好地解决了非线性高维数，局部极小点等问题。本公开实施例中利用支持向量机对于分类问题的良好性能，采用支持向量机对信号特征进行识别，还可以采用神经网络模型及其他分类算法，在此不再赘述。
本公开实施例中，利用第一支持向量机对信号特征进行识别，可以对信噪比进行识别，还可以对信噪比、角度信息、距离信息及径向速度分辨率的组合进行识别，其中，信噪比为必不可少的识别项，在信号特征组合识别时，可以对每种特征分配不同的权重，信噪比拥有最高的权重，各个特征参数的权重可以根据实际情况进行调整。
步骤S208，根据第一支持向量机对信号特征的识别结果确定物体 的材质类型，第一支持向量机是采用具有标记信息的训练数据对第二支持向量机进行训练后得到的，标记信息被设置为标记训练数据的材质类型，识别结果被设置为指示物体与各个材质类型的关联关系。
本公开实施例中，标记信息至少标识出各个训练数据的材质类型。类似地，识别结果至少为训练数据属于各个材质类型的概率值，用来指示其所属的材质类型。
本公开技术方案中，考虑到考虑到雷达信号灵敏度高、无需直接接触、可穿透性强等特点，特别是芯片级的毫米波雷达的出现，使得消费级雷达信号的可用性大大增强；因此，本公开使用毫米波雷达信号进行障碍物检测，结合支持向量机分类技术，识别并确定障碍物的材质类型，由于毫米波雷达不生成图像感光数据，因此不仅可以在识别得到障碍物材质类型，甚至还可以保护用户隐私，解决相关技术中隐私安全性较低的问题，进而达到无需采用视觉传感器识别障碍物材质类型，且保护用户隐私的技术效果。
本公开提出了一种提取反射信号中的信噪比以供支持向量机进行识别的方法。下面结合图2所示的步骤进一步详述本公开的技术方案。
在一些实施方式中，步骤S204提供的技术方案中，提取反射信号中的信号特征可以包括：
步骤1，提取反射信号中的信噪比；
步骤2，将信噪比进行时域转换和/或频域转换，得到信号特征。
本公开实施例中，毫米波雷达的发射信号在不同材质的物体上会反射出不同信噪比值的信号，不同材质的信噪比值的波动范围不一样，这是区分各个材质的重要依据，可以对各个材质的信噪比值进行时域与频域的转换，作为该信号特征。信噪比的提取方式为：
Figure PCTCN2020140491-appb-000001
其中，pt为发射输出功率，G Rx、G Tx为RX和TX天线增益，c为光速，σ为RCS of the object即雷达照射物体的横截面，RCS是对象反射回来的能量的量度，这决定了雷达传感器对物体的可检测性，N为脉冲调频的数量，T r为脉冲调频的时间，k为玻尔兹曼常数，T为环境温度，NF为接收器噪声系数，SNR det为算法检测物体所需的最小信噪比的值。
本公开还提出了一种利用支持向量机对信号特征的识别结果确定障碍物材质类型的方法，下面结合图2所示的步骤，进一步详述本公开的技术方案。
在一些实施方式中，步骤S208提供的技术方案中，根据第一支持向量机对信号特征的识别结果确定物体的材质类型可以包括以下步骤：
步骤1，获取第一支持向量机输出的识别结果，识别结果包括物体属于各个材质类型的预测值；
步骤2，将预测值中的最大值作为最终的识别结果，并将最大值指示的材质类型作为物体最终的材质类型。
本公开实施例中，预先训练好的第一支持向量机根据信号特征中波动信息预测障碍物属于各个材质类型的概率，并将概率最大的一种材质类型作为障碍物最终的材质类型。
本公开还提出了一种训练本公开实施例中所用的第一支持向量机的方法。
在一些实施方式中，根据第一支持向量机对信号特征的识别结果确定物体的材质类型之前，该方法还包括：通过训练数据对第二支持 向量机内各参数进行初始化，得到第三支持向量机；在第三支持向量机对测试数据的识别准确度达到目标阈值的情况下，将第三支持向量机作为第一支持向量机；在第三支持向量机对测试数据的识别准确度未达到目标阈值的情况下，继续使用训练数据对第三支持向量机进行训练，以调整第三支持向量机内各参数的数值，直至第三支持向量机对测试数据的识别准确度达到目标阈值。
本公开实施例中，可以获取多个毫米波雷达发射信号在多个障碍物上进行反射后的反射信号作为训练样本，每个训练样本中包括该训练样本的信噪比、角度信息、距离信息以及径向速度分辨率等，并标注该训练样本对应的障碍物的材质类型。训练数据可以来自于实际家庭场景下，采集多组数据，按照材质类型，即按照硬质物体(地板、水面、木板、金属)与软体物质(电线、塑料袋、抹布、宠物粪便)划分，并按照类别进行标注，还可以按照实际需要进行适应性调整。利用上述训练数据初始化第二支持向量机，得到第三支持向量机，并训练该第三支持向量机，直至该第三支持向量机收敛，得到第一神支持向量机。
在一些实施方式中，对于障碍物材质类型的识别，上述训练该第三支持向量机，直至该第三支持向量机收敛可以包括：
分别将每一个训练样本输入第三支持向量机，得到障碍物材质类型的训练预测值；
根据多个训练预测值和对应的训练样本中的实际材质类型之间的差异确定损失值；
利用多个损失值修正第三支持向量机，直至第三支持向量机输出结果的精度达到目标阈值。
在一些实施方式中，本公开实施例还可以使用遗传优化算法对支持向量机的参数进行优化，将优化后参数进行存储。优化支持向量机 参数的算法还可以是蚁群等优化算法。
本公开还提出了一种根据第一支持向量机对信号特征的识别结果确定障碍物的材质类型之后，控制扫地机进一步清扫的方法。
在一些实施方式中，根据第一支持向量机对信号特征的识别结果确定物体的材质类型之后，该方法还包括：
在物体的材质类型为预设材质类型的情况下，控制扫地机按照与预设材质类型匹配的目标方式动作。
在一些实施方式中，预设材质类型为软体材质类型时，在物体的材质类型为预设材质类型的情况下，控制扫地机按照与预设材质类型匹配的目标方式动作可以包括以下步骤：
步骤1，提取反射信号中的距离信息、角度信息及径向速度分辨率，信号特征包括距离信息、角度信息及径向速度分辨率；
步骤2，根据距离信息、角度信息及径向速度分辨率确定物体的位置，并识别得到物体的第一形状；
步骤3，在物体的位置处建立虚拟墙，虚拟墙沿第一形状的外接矩形边分布；
步骤4，按照扫地机的行进速度，控制扫地机沿虚拟墙进行清扫。
本公开实施例中，对于软体材质的障碍物，需要绕开，避免接触后损伤扫地机甚至造成二次污染，但是软体材质的障碍物无法触发扫地机机身周围的传感器，于是可以通过建立虚拟墙，来模拟扫地机碰撞虚拟墙从而改变行进方向。反射信号中的距离信息、角度信息及径向速度分辨率可以大致识别到障碍物所在的位置和大致的形状，根据障碍物大致的形状确定该形状的外接矩形，利用外接矩形生成虚拟墙，确保包围整个障碍物，或者还可以根据实际需要进一步扩大虚拟墙的范围。最后控制扫地机“碰撞”虚拟墙，并沿虚拟墙进行清扫。
在一些实施方式中，在物体的位置处建立虚拟墙还可以包括以下步骤：
步骤1，获取第一形状的第一尺寸；
步骤2，获取第一尺寸与膨胀系数的乘积，得到第二尺寸；
步骤3，按照第二尺寸在物体的位置处建立虚拟墙。
本公开实施例中，不同的软体材质的障碍物产生的虚拟墙不同，虚拟墙根据障碍物的大致形状生成，但虚拟墙的范围可以大于识别到的障碍物的大致形状，因此可以将障碍物的第一形状(大致形状)乘以膨胀系数，对第一形状进行膨胀、扩大，在完全包围障碍物的情况下生成虚拟墙，优选的，膨胀系数可以预设为1.0至1.5。
在一些实施方式中，预设材质类型为硬质材质类型时，物体的材质类型为预设材质类型的情况下，控制扫地机按照与预设材质类型匹配的目标方式动作还包括：控制扫地机对物体进行碰撞，以使扫地机沿物体的边缘进行清扫。
本公开实施例中，对于硬质材质的障碍物，直接控制扫地机碰撞障碍物，触发扫地机机身周围的传感器，以使扫地机沿障碍物的边缘进行清扫。
下面以本公开技术方案在实际扫地机清扫时的应用作为简要说明，本公开技术方案的实施过程如下：
步骤1，扫地机开始行走；
步骤2，扫地机行走过程中毫米波雷达采集障碍物信息；
步骤3，扫地机控制器提取信号特征，分析数据；
步骤4，为各个信号特征分配权重，进行数据融合；
步骤5，利用分类器(第一支持向量机)得出障碍物材质类型；
步骤6，反馈给扫地机；
步骤7，控制扫地机按照目标方式进行清扫。
本公开通过获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，提取反射信号中的信号特征，利用第一支持向量机对信号特征进行识别，根据第一支持向量机对信号特征的识别结果确定物体的材质类型的技术方案，可以通过毫米波雷达检测不同材质的物体，识别水面、塑料袋、抹布、宠物粪便等软体材质物，及时避障。同时无需搭载视觉传感器，可以加快处理速度，减少扫地机占用面积，同时解决用户的隐私问题。
根据本公开实施例的又一方面，如图3所示，提供了一种扫地机数据处理装置，包括：信号获取模块301，被设置为获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，反射信号为毫米波雷达传感器发送发射信号后在物体上反射形成的信号；特征提取模块303，被设置为提取反射信号中的信号特征，信号特征被设置为表征反射信号的波动信息；识别模块305，被设置为利用第一支持向量机对信号特征进行识别；材质判别模块307，被设置为根据第一支持向量机对信号特征的识别结果确定物体的材质类型，第一支持向量机是采用具有标记信息的训练数据对第二支持向量机进行训练后得到的，标记信息被设置为标记训练数据的材质类型，识别结果被设置为指示物体与各个材质类型的关联关系。
需要说明的是，该实施例中的信号获取模块301可以被设置为执行本公开实施例中的步骤S202，该实施例中的特征提取模块303可以被设置为执行本公开实施例中的步骤S204，该实施例中的识别模块305可以被设置为执行本公开实施例中的步骤S206，该实施例中的材质判别模块307可以被设置为执行本公开实施例中的步骤S208。
此处需要说明的是，上述模块与对应的步骤所实现的示例和应用 场景相同，但不限于上述实施例所公开的内容。需要说明的是，上述模块作为装置的一部分可以运行在如图1所示的硬件环境中，可以通过软件实现，也可以通过硬件实现。
在一些实施方式中，该扫地机数据处理装置，还包括：信噪比提取模块，被设置为提取反射信号中的信噪比；转换模块，被设置为将信噪比进行时域转换和/或频域转换，得到信号特征。
在一些实施方式中，该扫地机数据处理装置，还包括：识别结果获取模块，被设置为获取第一支持向量机输出的识别结果，识别结果包括物体属于各个材质类型的预测值；结果确定模块，被设置为将预测值中的最大值作为最终的识别结果，并将最大值指示的材质类型作为物体最终的材质类型。
在一些实施方式中，该扫地机数据处理装置，还包括：第一训练模块，被设置为通过训练数据对第二支持向量机内各参数进行初始化，得到第三支持向量机；第二训练模块，被设置为在第三支持向量机对测试数据的识别准确度达到目标阈值的情况下，将第三支持向量机作为第一支持向量机；第三训练模块，被设置为在第三支持向量机对测试数据的识别准确度未达到目标阈值的情况下，继续使用训练数据对第三支持向量机进行训练，以调整第三支持向量机内各参数的数值，直至第三支持向量机对测试数据的识别准确度达到目标阈值。
在一些实施方式中，该扫地机数据处理装置，还包括：控制模块，被设置为在物体的材质类型为预设材质类型的情况下，控制扫地机按照与预设材质类型匹配的目标方式动作。
在一些实施方式中，该扫地机数据处理装置，还包括：信号特征提取模块，被设置为提取反射信号中的距离信息、角度信息及径向速度分辨率，信号特征包括距离信息、角度信息及径向速度分辨率；位置和形状确定模块，被设置为根据距离信息、角度信息及径向速度分 辨率确定物体的位置，并识别得到物体的第一形状；虚拟墙建立模块，被设置为在物体的位置处建立虚拟墙，虚拟墙沿第一形状的外接矩形边分布；第一动作模块，被设置为按照扫地机的行进速度，控制扫地机沿虚拟墙进行清扫。
在一些实施方式中，该扫地机数据处理装置，还包括：第一尺寸获取模块，被设置为获取第一形状的第一尺寸；第二尺寸获取模块，被设置为获取第一尺寸与膨胀系数的乘积，得到第二尺寸；虚拟墙确定模块，被设置为按照第二尺寸在物体的位置处建立虚拟墙。
在一些实施方式中，该扫地机数据处理装置，还包括：第二动作模块，被设置为控制扫地机对物体进行碰撞，以使扫地机沿物体的边缘进行清扫。
根据本公开实施例的又一方面还提供了一种计算机设备，包括存储器、处理器，所述存储器中存储有可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现上述步骤。
上述计算机设备中的存储器、处理器通过通信总线和通信接口进行通信。所述通信总线可以是外设部件互连标准(Peripheral Component Interconnect，简称PCI)总线或扩展工业标准结构(Extended Industry Standard Architecture，简称EISA)总线等。该通信总线可以分为地址总线、数据总线、控制总线等。
存储器可以包括随机存取存储器(Random Access Memory，简称RAM)，也可以包括非易失性存储器(non-volatile memory)，例如至少一个磁盘存储器。可选的，存储器还可以是至少一个位于远离前述处理器的存储装置。
上述的处理器可以是通用处理器，包括中央处理器(Central Processing Unit，简称CPU)、网络处理器(Network Processor，简称NP)等；还可以是数字信号处理器(Digital Signal Processing，简称DSP)、 专用集成电路(Application Specific Integrated Circuit，简称ASIC)、现场可编程门阵列(Field-Programmable Gate Array，简称FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
根据本公开实施例的又一方面还提供了一种具有处理器可执行的非易失的程序代码的计算机可读介质。
在本公开实施例的一些实施方式中，计算机可读介质被设置为存储用于所述处理器执行以下步骤的程序代码：
步骤S202，获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，反射信号为毫米波雷达传感器发送发射信号后在物体上反射形成的信号；
步骤S204，提取反射信号中的信号特征，信号特征被设置为表征反射信号的波动信息；
步骤S206，利用第一支持向量机对信号特征进行识别；
步骤S208，根据第一支持向量机对信号特征的识别结果确定物体的材质类型，第一支持向量机是采用具有标记信息的训练数据对第二支持向量机进行训练后得到的，标记信息被设置为标记训练数据的材质类型，识别结果被设置为指示物体与各个材质类型的关联关系。
在一些实施方式中，本实施例中的具体示例可以参考上述实施例中所描述的示例，本实施例在此不再赘述。
本公开实施例在具体实现时，可以参阅上述各个实施例，具有相应的技术效果。
可以理解的是，本文描述的这些实施例可以用硬件、软件、固件、中间件、微码或其组合来实现。对于硬件实现，处理单元可以实现在一个或多个专用集成电路(Application Specific Integrated Circuits，ASIC)、数字信号处理器(Digital Signal Processing，DSP)、数字信号处 理设备(DSP Device，DSPD)、可编程逻辑设备(Programmable Logic Device，PLD)、现场可编程门阵列(Field-Programmable Gate Array，FPGA)、通用处理器、控制器、微控制器、微处理器、被设置为执行本公开所述功能的其它电子单元或其组合中。
对于软件实现，可通过执行本文所述功能的单元来实现本文所述的技术。软件代码可存储在存储器中并通过处理器执行。存储器可以在处理器中或在处理器外部实现。
本领域普通技术人员可以意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本公开的范围。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的系统、装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在本公开所提供的实施例中，应该理解到，所揭露的装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述模块的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个模块或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位 于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本公开各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。
所述功能如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本公开实施例的技术方案本质上或者说对相关技术做出贡献的部分或者该技术方案的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本公开各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、ROM、RAM、磁碟或者光盘等各种可以存储程序代码的介质。需要说明的是，在本文中，诸如“第一”和“第二”等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
以上所述仅是本公开的具体实施方式，使本领域技术人员能够理解或实现本公开。对这些实施例的多种修改对本领域的技术人员来说将是显而易见的，本文中所定义的一般原理可以在不脱离本公开的精神或范围的情况下，在其它实施例中实现。因此，本公开将不会被限制于本文所示的这些实施例，而是要符合与本文所申请的原理和新颖 特点相一致的最宽的范围。

Claims (11)

  1. 一种扫地机数据处理方法，包括：
    获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，所述反射信号为所述毫米波雷达传感器发送发射信号后在物体上反射形成的信号；
    提取所述反射信号中的信号特征，所述信号特征被设置为表征所述反射信号的波动信息；
    利用第一支持向量机对所述信号特征进行识别；
    根据所述第一支持向量机对所述信号特征的识别结果确定所述物体的材质类型，其中，所述第一支持向量机是采用具有标记信息的训练数据对第二支持向量机进行训练后得到的，所述标记信息被设置为标记所述训练数据的材质类型，所述识别结果被设置为指示所述物体与各个材质类型的关联关系。
  2. 根据权利要求1所述的方法，其中，提取所述反射信号中的信号特征包括：
    提取所述反射信号中的信噪比；
    将所述信噪比进行时域转换和/或频域转换，得到所述信号特征。
  3. 根据权利要求2所述的方法，其中，根据所述第一支持向量机对所述信号特征的识别结果确定所述物体的材质类型包括：
    获取所述第一支持向量机输出的识别结果，其中，所述识别结果包括所述物体属于各个材质类型的预测值；
    将所述预测值中的最大值作为最终的识别结果，并将所述最大值指示的材质类型作为所述物体最终的材质类型。
  4. 根据权利要求1至3中任意一项所述的方法，其中，根据所述 第一支持向量机对所述信号特征的识别结果确定所述物体的材质类型之前，所述方法还包括：
    通过所述训练数据对所述第二支持向量机内各参数进行初始化，得到第三支持向量机；
    在所述第三支持向量机对测试数据的识别准确度达到目标阈值的情况下，将所述第三支持向量机作为所述第一支持向量机；
    在所述第三支持向量机对所述测试数据的识别准确度未达到所述目标阈值的情况下，继续使用所述训练数据对所述第三支持向量机进行训练，以调整所述第三支持向量机内各参数的数值，直至所述第三支持向量机对所述测试数据的识别准确度达到所述目标阈值。
  5. 根据权利要求1至3中任意一项所述的方法，其中，根据所述第一支持向量机对所述信号特征的识别结果确定所述物体的材质类型之后，所述方法还包括：
    在所述物体的所述材质类型为预设材质类型的情况下，控制所述扫地机按照与所述预设材质类型匹配的目标方式动作。
  6. 根据权利要求5所述的方法，其中，所述预设材质类型为软体材质类型时，在所述物体的所述材质类型为预设材质类型的情况下，控制所述扫地机按照与所述预设材质类型匹配的目标方式动作包括：
    提取所述反射信号中的距离信息、角度信息及径向速度分辨率，其中，所述信号特征包括所述距离信息、所述角度信息及所述径向速度分辨率；
    根据所述距离信息、所述角度信息及所述径向速度分辨率确定所述物体的位置，并识别得到所述物体的第一形状；
    在所述物体的位置处建立虚拟墙，其中，所述虚拟墙沿所述第一形状的外接矩形边分布；
    按照所述扫地机的行进速度，控制所述扫地机沿所述虚拟墙进行清扫。
  7. 根据权利要求6所述的方法，其中，在所述物体的位置处建立虚拟墙还包括：
    获取所述第一形状的第一尺寸；
    获取所述第一尺寸与膨胀系数的乘积，得到第二尺寸；
    按照所述第二尺寸在所述物体的位置处建立虚拟墙。
  8. 根据权利要求5所述的方法，其中，所述预设材质类型为硬质材质类型时，所述物体的所述材质类型为预设材质类型的情况下，控制所述扫地机按照与所述预设材质类型匹配的目标方式动作还包括：
    控制所述扫地机对所述物体进行碰撞，以使扫地机沿所述物体的边缘进行清扫。
  9. 一种扫地机数据处理装置，包括：
    信号获取模块，被设置为获取安装在扫地机上的毫米波雷达传感器接收到的反射信号，所述反射信号为所述毫米波雷达传感器发送发射信号后在物体上反射形成的信号；
    特征提取模块，被设置为提取所述反射信号中的信号特征，所述信号特征被设置为表征所述反射信号的波动信息；
    识别模块，被设置为利用第一支持向量机对所述信号特征进行识别；
    材质判别模块，被设置为根据所述第一支持向量机对所述信号特征的识别结果确定所述物体的材质类型，其中，所述第一支持向量机是采用具有标记信息的训练数据对第二支持向量机进行训练后得到的，所述标记信息被设置为标记所述训练数据的材质类型，所述识别 结果被设置为指示所述物体与各个材质类型的关联关系。
  10. 一种计算机设备，包括存储器、处理器，所述存储器中存储有可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现上述权利要求1至8任一项所述的方法的步骤。
  11. 一种具有处理器可执行的非易失的程序代码的计算机可读介质，所述程序代码使所述处理器执行所述权利要求1至8任一所述方法。
PCT/CN2020/140491 2020-06-22 2020-12-29 扫地机数据处理方法、装置、设备及计算机可读介质 WO2021258698A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010576664.8A CN111839360B (zh) 2020-06-22 2020-06-22 扫地机数据处理方法、装置、设备及计算机可读介质
CN202010576664.8 2020-06-22

Publications (1)

Publication Number Publication Date
WO2021258698A1 true WO2021258698A1 (zh) 2021-12-30

Family

ID=72988404

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/140491 WO2021258698A1 (zh) 2020-06-22 2020-12-29 扫地机数据处理方法、装置、设备及计算机可读介质

Country Status (2)

Country Link
CN (1) CN111839360B (zh)
WO (1) WO2021258698A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114831540A (zh) * 2022-04-19 2022-08-02 珠海格力电器股份有限公司 一种机器人及使用其的方法

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111839360B (zh) * 2020-06-22 2021-09-14 珠海格力电器股份有限公司 扫地机数据处理方法、装置、设备及计算机可读介质
CN113693522B (zh) * 2021-02-10 2023-01-17 北京石头创新科技有限公司 清洁机器人脱困方法及装置、介质及电子设备
CN113469000A (zh) * 2021-06-23 2021-10-01 追觅创新科技(苏州)有限公司 区域地图的处理方法及装置、存储介质及电子装置
CN113759377B (zh) * 2021-08-26 2023-09-01 追觅创新科技(苏州)有限公司 自移动设备定位方法、装置、存储介质、设备及系统
CN114223554B (zh) * 2022-01-21 2022-11-25 深圳巴诺机器人有限公司 一种宠物粪便扫地机及其控制方法
CN114972229A (zh) * 2022-05-17 2022-08-30 平安科技（深圳）有限公司 基于材质类型的定损检测方法、装置、电子设备、介质
CN115639820B (zh) * 2022-10-18 2023-08-01 未岚大陆(北京)科技有限公司 虚拟墙的设置方法、自主移动设备和计算机可读存储介质

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070098967A1 (en) * 2005-10-31 2007-05-03 Toyoda Gosei Co., Ltd. Resin product, production method for the same, and deposition method for a metallic coating
CN107450569A (zh) * 2017-09-27 2017-12-08 上海思岚科技有限公司 一种扫地机器人的控制方法及设备
CN108334080A (zh) * 2018-01-18 2018-07-27 大连理工大学 一种针对机器人导航的虚拟墙自动生成方法
CN108960405A (zh) * 2017-05-18 2018-12-07 电装It研究所 识别系统、通用特征值提取单元以及识别系统构成方法
US10353053B2 (en) * 2016-04-22 2019-07-16 Huawei Technologies Co., Ltd. Object detection using radar and machine learning
CN110561459A (zh) * 2018-09-28 2019-12-13 成都家有为力机器人技术有限公司 一种具有双目摄像头的清洁机器人及其工作方法
CN110766067A (zh) * 2019-10-21 2020-02-07 浙江工业大学之江学院 基于机器学习的毫米波雷达材质识别方法
CN111839360A (zh) * 2020-06-22 2020-10-30 珠海格力电器股份有限公司 扫地机数据处理方法、装置、设备及计算机可读介质

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070098967A1 (en) * 2005-10-31 2007-05-03 Toyoda Gosei Co., Ltd. Resin product, production method for the same, and deposition method for a metallic coating
US10353053B2 (en) * 2016-04-22 2019-07-16 Huawei Technologies Co., Ltd. Object detection using radar and machine learning
CN108960405A (zh) * 2017-05-18 2018-12-07 电装It研究所 识别系统、通用特征值提取单元以及识别系统构成方法
CN107450569A (zh) * 2017-09-27 2017-12-08 上海思岚科技有限公司 一种扫地机器人的控制方法及设备
CN108334080A (zh) * 2018-01-18 2018-07-27 大连理工大学 一种针对机器人导航的虚拟墙自动生成方法
CN110561459A (zh) * 2018-09-28 2019-12-13 成都家有为力机器人技术有限公司 一种具有双目摄像头的清洁机器人及其工作方法
CN110766067A (zh) * 2019-10-21 2020-02-07 浙江工业大学之江学院 基于机器学习的毫米波雷达材质识别方法
CN111839360A (zh) * 2020-06-22 2020-10-30 珠海格力电器股份有限公司 扫地机数据处理方法、装置、设备及计算机可读介质

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114831540A (zh) * 2022-04-19 2022-08-02 珠海格力电器股份有限公司 一种机器人及使用其的方法
WO2023202169A1 (zh) * 2022-04-19 2023-10-26 珠海格力电器股份有限公司 一种机器人及使用其的方法

Also Published As

Publication number Publication date
CN111839360B (zh) 2021-09-14
CN111839360A (zh) 2020-10-30

Similar Documents

Publication Publication Date Title
WO2021258698A1 (zh) 扫地机数据处理方法、装置、设备及计算机可读介质
WO2021004299A1 (zh) 人员检测方法、装置、电子设备及存储介质
WO2020103410A1 (zh) 一种检测方法、检测装置、终端以及检测系统
CN101223456B (zh) 利用统计分类器识别移动对象的计算机实现的方法
WO2020258804A1 (zh) 人体活动姿态的监测方法及系统、人体姿态监测器、存储介质、处理器
US20150260838A1 (en) Sparse Array RF Imaging for Surveillance Applications
Zhao et al. Human tracking and identification through a millimeter wave radar
WO2014203041A1 (en) System and method for rfid indoor localization
JP2018163096A (ja) 情報処理方法および情報処理装置
US20230045798A1 (en) Method for predicting structure of indoor space using radio propagation channel analysis through deep learning
CN112348777B (zh) 人体目标的检测方法、装置及终端设备
CN109766851A (zh) 障碍物的确定方法及装置、倒车成像设备
Amin et al. Quality of obstacle distance measurement using ultrasonic sensor and precision of two computer vision-based obstacle detection approaches
US20220026557A1 (en) Spatial sensor system with background scene subtraction
Khan et al. An IR-UWB multi-sensor approach for collision avoidance in indoor environments
KR20220068730A (ko) 실내 보안 응용을 위한 인체 감지 레이다 시스템 및 이를 이용한 인체 감지 방법
CN116106855A (zh) 摔倒检测方法及摔倒检测装置
Cui et al. Human posture capturing with millimetre wave radars
WO2022179207A1 (zh) 视窗遮挡检测方法及装置
US20240053464A1 (en) Radar Detection and Tracking
Che et al. Weighted naive Bayes approach for imbalanced indoor positioning system using UWB
Li et al. Di-gesture: Domain-independent and real-time gesture recognition with millimeter-wave signals
Devnath et al. A systematic study on object recognition using millimeter-wave radar
TWI772208B (zh) 基於毫米波雷達的人數偵測方法
WO2021007760A1 (zh) 身份识别方法、终端设备及计算机存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20942041

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20942041

Country of ref document: EP

Kind code of ref document: A1