A lidar control system and method thereof

Abstract

The present invention discloses a lidar control system and method, which relates to the technical field of lidar control. The method includes generating a lidar multi-modal noise basis matrix through quantum-classical noise fusion modeling; generating a lidar joint pulse phase encoding based on the noise distribution of the lidar multi-modal noise basis matrix; collecting lidar spatio-temporal marked echo data according to the phase parameters of the lidar joint pulse phase encoding; generating a lidar multi-dimensional target feature tensor with the lidar spatio-temporal marked echo data as the original signal; outputting a lidar enhanced target point cloud based on the time-frequency-space features of the lidar multi-dimensional target feature tensor; and updating the lidar spatio-temporal marked echo data and the lidar joint pulse phase encoding, which solves the problems in aspects such as noise processing, target feature extraction, point cloud quality, and system optimization existing in the existing radar control methods during use.

Description

Technical Field

[0001] The present invention relates to the technical field of lidar control, and specifically to a lidar control system and method thereof. Background Art

[0002] Solid-state lidar has advantages such as high reliability, small size, and fast imaging speed. Therefore, solid-state lidar is suitable for vehicle driving fields such as assisted driving and autonomous driving. Solid-state lidar can detect the surrounding environment based on a transmitting unit and a receiving unit. The transmitting unit includes multiple transmitters, the receiving unit includes multiple receivers, and the transmitters in the transmitting unit correspond one-to-one with the receivers in the receiving unit.

[0003] Existing radar control methods have problems in aspects such as noise processing, target feature extraction, point cloud quality, and system optimization during use, resulting in the radar being difficult to effectively reduce noise interference and improve the detection accuracy, target recognition, and positioning ability of lidar in complex environments. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a lidar control system and method thereof, which solve the problems existing in existing radar control methods in aspects such as noise processing, target feature extraction, point cloud quality, and system optimization during use.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A lidar control method includes the following steps: generating a lidar multimodal noise basis matrix through quantum-classical noise fusion modeling; based on the noise distribution of the lidar multimodal noise basis matrix, using quantum annealing to select low-noise frequency bands, and combining chaotic perturbation to generate a lidar joint pulse phase encoding; driving metasurface beamforming according to the phase parameters of the lidar joint pulse phase encoding, and collecting lidar spatio-temporal marked echo data; using the lidar spatio-temporal marked echo data as the original signal, and combining the noise threshold of the lidar joint pulse phase encoding to perform quantum convolution filtering to generate a lidar multi-dimensional target feature tensor; based on the time-frequency-spatial features of the lidar multi-dimensional target feature tensor, optimizing the confidence threshold through federated reinforcement learning to output a lidar enhanced target point cloud; analyzing the topological structure of the lidar enhanced target point cloud, and driving the metamaterial gene evolution algorithm to update the lidar spatio-temporal marked echo data and the lidar joint pulse phase encoding.

[0006] Further, generating the lidar multimodal noise basis matrix includes the following steps: in the lidar initialization stage, turning off the transmitting module, activating all receiving channels, and synchronously collecting three-dimensional noise data in the time-space-frequency domain:

[0007] Time dimension: Continuously collect for a set duration , divided into several time slices ; Spatial dimension: record the coordinate offset of each receiving unit ; Frequency domain dimension: perform multi-order Fourier transform on the signals of each time slice and extract the band energy ;

[0008] Integrate a superconducting quantum interference device at the receiving end to measure the quantum noise density matrix of the ambient electromagnetic field , and fuse it with the three-dimensional noise data of time-space-frequency domain to construct the lidar multi-modal noise base matrix , with the structure:

[0009] ;

[0010] Among them, is 's weight factor, is 's weight factor, is 's weight factor, is 's quantum noise coupling coefficient;

[0011] Generate extreme environment noise samples based on the trained conditional generative adversarial network , expand the boundary conditions of the lidar multi-modal noise base matrix, and inject them into the lidar multi-modal noise base matrix at a ratio of 5% :

[0012] ;

[0013] At every set interval , resample and update in the lidar multi-modal noise base matrix , and .

[0014] Further, generating the lidar joint pulse phase coding includes the following steps: performing frequency domain slice analysis on the lidar multi-modal noise base matrix, and calculating the comprehensive noise intensity of each frequency band; mapping the frequency band selection problem to the Ising model, solving the optimal frequency band combination through a quantum annealing machine, and selecting the 3 frequency bands with the lowest comprehensive noise intensity as the main-secondary avoidance frequency bands , , representing the three lowest noise frequency bands respectively;

[0015] Generate a triple phase shift keying waveform:

[0016] Main frequency band : Pulse width , phase angle ;

[0017] Auxiliary band : Pulse width , phase angle ;

[0018] Auxiliary band : Pulse width , phase angle ;

[0019] Generates the unperturbed base TPK waveform : , , t is the current time, is pi, is a rectangular window function that defines the pulse's existence interval in the time domain. ; Inject chaotic disturbance into each triple phase keying waveform , generate the enhanced waveform after perturbation: , is the delay time, is the chaos gain coefficient; the enhanced waveform after disturbance is superimposed to generate the laser radar joint pulse phase coding Dynamically adjust the emission energy of each frequency band according to the comprehensive noise intensity of each frequency band in the lidar multimodal noise basis matrix .

[0020] Furthermore, the emission energy The calculation formula is as follows: , is a system constant, , , Main frequency band , auxiliary frequency band and auxiliary bands The comprehensive noise intensity.

[0021] Furthermore, collecting the laser radar spatiotemporal marker echo data includes the following steps: dividing the transmitting array into multiple groups, each group is divided into multiple groups according to the golden section angle. Perform spatial deflection; integrate programmable metasurface units at the transmitting end, according to the golden section angle Dynamically adjust the phase distribution of programmable metasurface units:

[0022] , is the laser wavelength, is pi, is the two-dimensional plane coordinate of the programmable metasurface unit;

[0023] Each group of transmitters emits laser according to the parameters of lidar joint pulse phase coding, and the emission time Adopts pseudo-random delay; uses entangled photon pairs to calibrate the clocks of the transmitting end and the receiving end; records the absolute timestamp of each photon arrival at the receiving end and spatial coordinates ; encapsulates the original data into initial spatio-temporal marked echo data ; , where m is the photon number, is the voltage value corresponding to the photon energy; constructs a spatial correlation mask according to the spatial deflection angle of the emission group:

[0024] ;

[0025] Among them, is the mask value of the receiving unit , and are the abscissa offset and ordinate offset of the receiving unit relative to the reference position respectively; retains the initial spatio-temporal marked echo data with the mask value being 1, denoted as lidar spatio-temporal marked echo data .

[0026] Furthermore, generating the lidar multi-dimensional target feature tensor includes the following steps: performing 5-layer wavelet packet decomposition on the lidar spatio-temporal marked echo data to extract the time-frequency energy of the m-th photon; decomposing the lidar spatio-temporal marked echo data into a core tensor G and factor matrices, and retaining the core tensor G as the denoised data: ; among them, is the time factor matrix, is the space factor matrix, is the total number of receiving units, is the frequency-domain factor matrix; calculates the spatial correlation factor :

[0027] ;

[0028] is the set of adjacent units of the receiving unit where the photon m is located, is the number of adjacent units, is the original signal amplitude of the adjacent units; constructs a quantum convolution kernel, whose Hamiltonian is driven by the quantum noise term of the MNM-Tensor, and calculates the fidelity loss after performing quantum filtering ;

[0029] Constructs a double-threshold filtering function:

[0030] ;

[0031] is the spatial consistency threshold, is the amplitude of the filtered signal, is the main frequency band , the secondary frequency band and the secondary frequency band and the mean noise energy of the secondary frequency band;

[0032] Construct a multi-dimensional target feature tensor of the lidar :

[0033] .

[0034] Furthermore, the output of the lidar enhanced target point cloud includes the following steps: Obtain the comprehensive confidence ; Obtain the confidence threshold stored in the database, filter the target points and generate the lidar enhanced target point cloud :

[0035] ;

[0036] is the three-dimensional coordinate of the target point in the radar coordinate system, is the confidence weight of the target point, is the comprehensive confidence of the target point. When is greater than the set threshold parameter, the target is retained; otherwise, it is determined as noise and excluded.

[0037] Furthermore, obtaining the comprehensive confidence is as follows: Define the confidence metrics: time-frequency stability : The standard deviation of the energy fluctuation of the target signal within the time window , determined based on the target signal energy of each time slice of the lidar multi-dimensional target feature tensor; spatial aggregation : The ratio of the number of target signal energies greater than the set threshold in each time slice of the lidar multi-dimensional target feature tensor to the normalized volume; energy significance : The ratio of the peak energy of the target signal to the background noise energy; Calculate the Betti numbers , , which are the number of connected components and the number of circular structures respectively, quantifying the connectivity and hole characteristics of the target structure;

[0038] Obtain the topological significance index : ;

[0039] ;

[0040] The weight coefficient of is The weight coefficient of is The weight coefficient of is The weight coefficient of is

[0041] Furthermore, analyze the topological structure of the lidar enhanced target point cloud, and drive the metamaterial gene evolution algorithm to update the lidar spatio-temporal marked echo data and the lidar joint pulse phase encoding, including the following steps: perform Delaunay triangulation on the lidar enhanced target point cloud to extract the environmental topological map , where the vertex V is the target point and the edge E is the adjacent relationship; calculate the average degree and the clustering coefficient of the environmental topological map; map the lidar enhanced target point cloud to the Chern-Simons action SCS; determine the environmental topological category according to the action SCS: ; ; For dense scenes: shorten the pulse width of the lidar spatio-temporal marked echo data and increase the update frequency of the lidar joint pulse phase encoding; for sparse scenes: expand the number of frequency bands of the lidar spatio-temporal marked echo data and add new auxiliary frequency bands; encode the programming parameters of the programmable metasurface unit into a binary gene chain, and optimize the beamforming parameters through genetic algorithm crossover and mutation, including the phase distribution of each programmable metasurface unit, the chaotic gain coefficient of each transmitting unit, and the beam focusing offset.

[0042] A lidar control system for the above-mentioned lidar control method, comprising a lidar multimodal noise floor matrix acquisition module, a lidar joint pulse phase encoding acquisition module, a lidar spatio-temporal marked echo data acquisition module, a lidar multi-dimensional target feature tensor acquisition module, a lidar enhanced target point cloud acquisition module, and an update module, wherein: The lidar multimodal noise floor matrix acquisition module is used to generate a lidar multimodal noise floor matrix through quantum-classical noise fusion modeling; The lidar joint pulse phase encoding acquisition module is used to select a low-noise frequency band by quantum annealing based on the noise distribution of the lidar multimodal noise floor matrix, and generate lidar joint pulse phase encoding in combination with chaotic perturbation; The lidar spatio-temporal marked echo data acquisition module is used to drive metasurface beamforming according to the phase parameters of the lidar joint pulse phase encoding, and collect lidar spatio-temporal marked echo data; The lidar multi-dimensional target feature tensor acquisition module is used to perform quantum convolution filtering with the lidar spatio-temporal marked echo data as the original signal in combination with the noise threshold of the lidar joint pulse phase encoding, and generate a lidar multi-dimensional target feature tensor; The lidar enhanced target point cloud acquisition module is used to optimize the confidence threshold through federated reinforcement learning based on the time-frequency - spatial features of the lidar multi-dimensional target feature tensor, and output a lidar enhanced target point cloud; The update module is used to analyze the topological structure of the lidar enhanced target point cloud, and drive the metamaterial gene evolution algorithm to update the lidar spatio-temporal marked echo data and the lidar joint pulse phase encoding.

[0043] The present invention has the following beneficial effects:

[0044] The lidar control method and system construct a multimodal noise floor matrix through quantum-classical noise fusion modeling, dynamically identify low-noise frequency bands and generate anti-interference pulse coding in combination with chaotic perturbation to effectively suppress the noise floor; utilize metasurface beamforming and quantum convolution filtering technologies to separate target features and noise in multiple dimensions of space-time and frequency domain; adaptively optimize the confidence threshold through federated reinforcement learning, significantly improve the topological recognition accuracy of the target point cloud, and at the same time introduce the metamaterial gene evolution algorithm to realize the dynamic optimization of system parameters, enabling the lidar to maintain high-precision and high-robust three-dimensional perception capabilities in extreme noise environments and dynamic scene changes.

[0045] Of course, any product implementing the present invention does not necessarily need to achieve all the above-mentioned advantages simultaneously. Description of the Drawings

[0046] Figure 1 It is a flowchart of the lidar control method of the present invention.

[0047] Figure 2 It is a block diagram of the lidar control system of the present invention. Detailed implementation mode

[0048] Please refer to Figure 1 , an embodiment of the present invention provides a technical solution: a lidar control method, including the following steps: generating a lidar multimodal noise basis matrix through quantum-classical noise fusion modeling; in the initialization stage of the lidar, turning off the transmitting module, activating all receiving channels, and synchronously collecting three-dimensional noise data in the time-space-frequency domain: turning off the laser emission in the initialization stage, and synchronously collecting three-dimensional noise data in the time-space-frequency domain through all receiving channels. Eliminate the interference of the transmitted signal on the measurement of the noise basis to ensure the purity of the noise data.

[0049] Time dimension: Continuously collect for a set duration = 50ms, divided into several (100) time slices ; Continuously collect 50ms divided into 100 time slices (0.5ms / slice) to capture the dynamic characteristics of the noise (such as periodic electromagnetic interference). Spatial dimension: Record the coordinate offset of each receiving unit ( , and are the horizontal coordinate offset and vertical coordinate offset of the receiving unit relative to the reference position respectively); Frequency domain dimension: Perform multi-(128)-order Fourier transform on the signal of each time slice to extract the frequency band energy ( ); The frequency band energy distribution with a resolution of 0.078kHz is refined to identify narrowband noise and broadband interference.

[0050] This multi-dimensional noise data acquisition method can comprehensively capture the noise characteristics of the lidar at different time and space positions, as well as the noise energy distribution in different frequency bands, providing a rich data basis for the subsequent generation of a more accurate noise basis matrix.

[0051] Integrate a superconducting quantum interference device at the receiving end to measure the quantum noise density matrix of the environmental electromagnetic field , and fuse it with the three-dimensional noise data in the time-space-frequency domain to construct a lidar multimodal noise basis matrix , the structure is:

[0052] ;

[0053] Among them, is the weight factor of , is the weight factor of , is the weight factor of , is the weight factor of The quantum noise coupling coefficient; capturing quantum fluctuation noise (the blind spot of traditional sensors) through a superconducting quantum interference device, and combining the spatial-frequency domain distribution of classical noise to construct a holographic noise feature matrix. By fusing quantum noise, the characteristics of environmental noise can be characterized more precisely, improving the accuracy and robustness of the noise model, and helping to better adapt to the actual environment in subsequent noise suppression and signal processing processes.

[0054] Generating extreme environmental noise samples based on a trained conditional generative adversarial network , expanding the boundary conditions of the lidar multimodal noise basis matrix and injecting it into the lidar multimodal noise basis matrix at a ratio of 5% , the 5% injection ratio not only maintains the original data distribution characteristics but also improves the generalization ability of the model to extreme scenarios:

[0055] ; Using the generative adversarial network can generate noise samples close to the real extreme environment, expanding the boundary conditions of the noise basis matrix, and making the lidar system more adaptable and robust in the face of complex and changing environments.

[0056] At every set interval , resampling and updating the lidar multimodal noise basis matrix , , and , , and According to the least squares fitting, , is the quantum learning rate, is the fidelity loss. The noise environment is dynamically changing. By regularly updating the noise basis matrix, the latest characteristics of environmental noise can be reflected in a timely manner, ensuring that the lidar system can perform effective signal processing and target detection based on an accurate noise model at different time periods.

[0057] Based on the noise distribution of the lidar multimodal noise basis matrix, using quantum annealing to select low-noise frequency bands, and combining chaotic perturbation to generate lidar joint pulse phase encoding;

[0058] Performing frequency-domain slice analysis on the lidar multimodal noise basis matrix and calculating the comprehensive noise intensity of each frequency band;

[0059] Mapping the frequency band selection problem to the Ising model, solving the optimal frequency band combination through a quantum annealing machine, and selecting the 3 frequency bands with the lowest comprehensive noise intensity as the main-secondary avoidance frequency bands , , representing the three lowest noise frequency bands respectively;

[0060] Generate a triple phase keying waveform:

[0061] Main frequency band : Pulse width , Phase angle ;

[0062] Auxiliary frequency band : Pulse width , Phase angle ;

[0063] Auxiliary frequency band : Pulse width , Phase angle ;

[0064] Generate an undisturbed basic TPK waveform :

[0065] , , where t is the current time, is the pi, is the rectangular window function, defining the existence interval of the pulse in the time domain ; Through detailed frequency domain analysis, the noise situation of each frequency band can be accurately understood, providing a basis for subsequent selection of low-noise frequency bands, thereby reducing the interference of noise on lidar signals. The Ising model is a classic statistical physics model that can well describe the interactions and energy states in complex systems. Transforming the frequency band selection problem into the Ising model can use the quantum annealing algorithm to quickly and effectively find the global optimal solution, ensuring that the selected frequency band combination has the lowest noise intensity and improving the quality and reliability of lidar signals.

[0066] Inject chaotic perturbations into each triple phase keying waveform , generating an enhanced waveform after perturbation: , is the delay time, is the chaotic gain coefficient;

[0067] Superimpose the enhanced waveforms after perturbation to generate the lidar joint pulse phase coding ; The triple phase keying waveform can provide rich phase information and enhance the anti-interference ability of the signal. The injection of chaotic perturbations further increases the complexity and randomness of the signal, making the signal more difficult to be interfered with and cracked during transmission, improving the security and stability of the lidar system.

[0068] Dynamically adjust the transmission energy of each frequency band according to the comprehensive noise intensity of each frequency band in the lidar multimodal noise basis matrix .

[0069] Transmission energy The calculation formula is as follows:

[0070] , is a system constant, , , are the main frequency band , the secondary frequency band and the secondary frequency band respectively. The comprehensive noise intensity. By dynamically adjusting the transmission energy, while ensuring the signal transmission quality, the energy usage efficiency can be optimized, unnecessary energy waste can be avoided, the service life of the lidar system can be extended, and its adaptability in different environments can be improved. This dynamic energy adjustment method based on noise intensity can ensure that the transmission energy is increased in high-noise frequency bands to overcome the noise impact and guarantee the signal transmission quality; while the transmission energy is reduced in low-noise frequency bands, thereby saving energy, extending the system working time, and improving the energy utilization efficiency.

[0071] Drive the metasurface beamforming according to the phase parameters of the lidar joint pulse phase coding, and collect the lidar spatio-temporal marked echo data; divide the transmitting array into multiple groups (4 groups), and each group is spatially deflected according to the golden ratio angle ; integrate programmable metasurface units at the transmitting end, and dynamically adjust the phase distribution of the programmable metasurface units according to the golden ratio angle :

[0072] , is the laser wavelength, is the pi, is the two-dimensional plane coordinate of the programmable metasurface unit; the golden ratio angle has good spatial coverage characteristics and uniformity. In this way, it can be ensured that the distribution of the laser beam in space is more reasonable, the detection blind area is reduced, and the omnidirectional detection ability for targets is improved. The programmable metasurface unit can flexibly control the phase of the laser beam, thereby realizing the precise regulation of the beam shape and direction. This dynamic adjustment ability can optimize the propagation characteristics of the beam according to different detection scenarios and target positions, and improve the detection efficiency and resolution of the lidar.

[0073] Each group of transmitters emits laser according to the parameters of the lidar joint pulse phase coding (including the determined main-secondary avoidance frequency bands , the pulse widths and phase angles of the three frequency bands, and the transmission energies corresponding to the three frequency bands), and the transmission time Using pseudo-random delay; pseudo-random delay emission can avoid signal interference between multiple groups of transmitters, improve the signal discrimination and anti-interference ability. The use of entangled photon pairs can achieve high-precision clock calibration, ensure the time synchronization between the transmitter and the receiver, which is crucial for accurately measuring the flight time of photons, and thus improve the accuracy of target distance measurement.

[0074] Calibrating the transmitter clock and the receiver clock using entangled photon pairs;

[0075] Recording the absolute timestamp of each photon arrival at the receiver and the spatial coordinates ; Recording the detailed arrival information of each photon can retain rich spatio-temporal data, providing accurate data support for subsequent target feature extraction and positioning. Data encapsulation is beneficial to the unified management and processing of data, improving the efficiency of data processing.

[0076] Encapsulating the original data into initial spatio-temporal marked echo data ; , where m is the photon number, is the voltage value corresponding to the photon energy;

[0077] Constructing a spatial correlation mask according to the spatial deflection angle of the emission group:

[0078] ;

[0079] Among them, is the mask value of the receiving unit , and are the abscissa offset and ordinate offset of the receiving unit relative to the reference position respectively;

[0080] Retaining the initial spatio-temporal marked echo data with the mask value being 1, denoted as lidar spatio-temporal marked echo data . By constructing a spatial correlation mask, data with high correlation with the target can be effectively screened out, irrelevant or noisy data can be removed, the purity and effectiveness of the data can be improved, and thus the accuracy of target detection and recognition can be enhanced.

[0081] Taking the lidar spatio-temporal marked echo data as the original signal, combining with the noise threshold of lidar joint pulse phase encoding for quantum convolution filtering to generate a lidar multi-dimensional target feature tensor;

[0082] Performing 5-layer wavelet packet decomposition on the lidar spatio-temporal marked echo data to extract the time-frequency energy of the m-th photon ; Wavelet packet decomposition can provide information in both the time and frequency dimensions. Through multi-level decomposition, the energy distribution of the signal at different times and frequencies can be captured, which helps to more comprehensively understand the characteristics of the target.

[0083] Decompose the lidar spatio-temporal tagged echo data into the core tensor G and the factor matrices, and retain the core tensor G as the denoised data:

[0084] ;

[0085] Among them, , is the time factor matrix, , is the space factor matrix, is the total number of receiving units, , is the frequency domain factor matrix; Tensor decomposition can decompose complex data structures into simpler components. By retaining the core tensor, noise can be effectively removed while the main features and information in the data are retained, improving the purity and usability of the data.

[0086] Calculate the spatial correlation factor :

[0087] ;

[0088] is the set of adjacent units of the receiving unit where photon m is located , is the number of adjacent units, is the original signal amplitude of the adjacent units; The spatial correlation factor can reflect the correlation and continuity of the signal in space, which helps to identify and connect signals belonging to the same target, improving the integrity and accuracy of the target.

[0089] Construct a quantum convolution kernel, whose Hamiltonian is driven by the quantum noise term of the MNM-Tensor. After performing quantum filtering, calculate the fidelity loss , for update; The quantum convolution kernel utilizes the advantages of quantum computing and can more efficiently process and filter high-dimensional data, improving the speed and effect of filtering. At the same time, the calculation of the fidelity loss helps to evaluate the quality of the filtered data and guide subsequent optimization and adjustment.

[0090] Construct a dual-threshold filtering function:

[0091] ;

[0092] is the spatial consistency threshold, is the amplitude of the filtered signal, is the main frequency band , the secondary frequency band And the secondary frequency band The mean noise energy; The dual-threshold filtering function can flexibly perform filtering operations according to different signal intensities and noise levels, which can not only remove low-intensity noise signals but also retain high-intensity target signals, improving the filtering accuracy and adaptability.

[0093] Construct a multi-dimensional target feature tensor of lidar :

[0094] The multi-dimensional target feature tensor can integrate information in multiple dimensions such as time, frequency, and space, comprehensively describe the characteristics of the target, provide a richer and more accurate data basis for subsequent target recognition and classification, and improve the target detection and recognition capabilities of the lidar system.

[0095] Based on the time-frequency - space features of the multi-dimensional target feature tensor of lidar, optimize the confidence threshold through federated reinforcement learning, and output the lidar enhanced target point cloud;

[0096] Outputting the lidar enhanced target point cloud includes the following steps:

[0097] Obtain the comprehensive confidence ;

[0098] Define the confidence index:

[0099] Time-frequency stability : The standard deviation of the energy fluctuation of the target signal within the time window , determined based on the target signal energy of each time slice of the multi-dimensional target feature tensor of lidar; The time-frequency stability can reflect the consistency of the target signal in time. A lower standard deviation of energy fluctuation indicates that the target signal is more stable in time, thereby improving its reliability.

[0100] Spatial aggregation degree : The ratio of the number of target signal energies greater than the set threshold in each time slice of the multi-dimensional target feature tensor of lidar to the standardized volume; The spatial aggregation degree can reflect the degree of concentration of the target signal in space. A higher spatial aggregation degree means that the target signal is more concentrated in space, thereby improving the identifiability and accuracy of the target.

[0101] Energy significance : The ratio of the peak energy of the target signal to the background noise energy; The energy significance can reflect the intensity of the target signal relative to the background noise. A higher ratio of peak energy to background noise energy indicates that the target signal is more prominent, thereby improving the detection accuracy of the target.

[0102] Calculate the Betti number of the multi-dimensional target feature tensor of lidar , , which are the number of connected components and the number of ring structures respectively, quantifying the connectivity and hole characteristics of the target structure; the Betti numbers can provide detailed information about the topological characteristics of the target structure. The number of connected components reflects the connectivity of the target, and the number of ring structures reflects the hole characteristics of the target. This information helps to more comprehensively understand the shape and structure of the target.

[0103] Obtain the topological saliency index : ; The topological saliency index synthesizes information from multiple aspects and can more comprehensively evaluate the saliency and reliability of the target. By adjusting the weighting coefficients, the importance of certain features can be flexibly emphasized according to the requirements of different application scenarios.

[0104] ;

[0105] is 's weight coefficient, is 's weight coefficient, is 's weight coefficient, [[ID=DB27]]is 's weight coefficient. The calculation of the comprehensive confidence can comprehensively evaluate the reliability of the target points. Combining features and topological information from multiple dimensions makes the judgment of the target points more accurate and comprehensive, reducing the possibility of misjudgment and missed judgment.

[0106] Obtain the confidence threshold stored in the database, filter the target points and generate the lidar enhanced target point cloud :

[0107] ;

[0108] is the three-dimensional coordinate of the target point in the radar coordinate system, is the confidence weight of the target point, is the comprehensive confidence of the target point. When is greater than the set threshold parameter, the target is retained; otherwise, it is determined as noise and removed. By setting the confidence threshold, low-confidence noise points can be effectively removed, and high-confidence target points can be retained, improving the quality and reliability of the point cloud data and providing more accurate data support for subsequent target recognition and analysis.

[0109] Analyze the topological structure of the lidar enhanced target point cloud, and drive the metamaterial gene evolution algorithm to update the lidar spatio-temporal marked echo data and lidar joint pulse phase coding.

[0110] Perform Delaunay triangulation on the lidar enhanced target point cloud to extract the environmental topological map , the vertex V is the target point, and the edge E is the adjacent relationship;

[0111] Calculate the average degree of the environmental topology graph and the clustering coefficient ;

[0112] Map the lidar enhanced target point cloud to the Chern-Simons action SCS;

[0113] Determine the environmental topology category according to the action SCS:

[0114] ;

[0115] ;

[0116] Through topological structure analysis, the spatial distribution and environmental characteristics of the target point cloud can be deeply understood, providing a basis for subsequent scene adaptability adjustment. Delaunay triangulation can effectively capture the spatial relationship between target points, while the Chern-Simons action can quantify the topological characteristics of the environment, helping the system identify different scene types, such as dense scenes or sparse scenes.

[0117] For dense scenes ( > 5): Shorten the pulse width of the lidar spatio-temporal marked echo data , and increase the update frequency of the lidar joint pulse phase encoding ; For sparse scenes ( < 2): Expand the number of frequency bands of the lidar spatio-temporal marked echo data, and add as an auxiliary frequency band; Scene adaptability update can enable the lidar system to dynamically adjust working parameters according to different environmental conditions and optimize the detection performance. In dense scenes, shortening the pulse width can improve the resolution, and increasing the update frequency can enhance the real-time performance; in sparse scenes, expanding the number of frequency bands can increase the detection range, and adding an auxiliary frequency band can improve the signal diversity.

[0118] Encode the programming parameters of the programmable metasurface unit into a binary gene chain, and optimize the beamforming parameters through genetic algorithm crossover and mutation, including the phase distribution of each programmable metasurface unit, the chaotic gain coefficient of each transmitting unit, and the beam focusing offset. Specifically:

[0119] Based on , and Construct a binary gene chain , and are the phase distribution and focusing offset of the th metasurface unit, is the Chaotic gain of a transmitter group;

[0120] Initialization: Randomly generate 100 groups of gene chains;

[0121] Calculate the fitness of each group , sort and retain the top 20%:

[0122] , is the output signal-to-noise ratio (the energy ratio of the filtered signal to the noise), is the total transmission power, is the number of correctly detected targets, is the total number of photons, e is the natural constant, is the number of ring structures.

[0123] Randomly select parental gene chains, generate offspring by single-point crossover; randomly flip bits of the offspring gene chains; directly retain the top 5% of the optimal individuals in each generation to avoid degradation; Metasurface phase reconfiguration: Write the optimized into the FPGA, with a refresh rate of 10 ms / cell; Chaotic gain synchronization: Update through the SPI interface to ensure bit perturbation synchronization; Adjust the beamforming parameters according to to improve the resolution.

[0124] The metamaterial gene evolution algorithm can gradually optimize the beamforming parameters by simulating the process of biological evolution, improving the directivity and focusing effect of the beam. This optimization method can find the optimal beamforming scheme in a complex electromagnetic environment, enhancing the detection accuracy and anti-interference ability of the lidar system.

[0125] A lidar control system for the above-mentioned lidar control method, such as Figure 2As shown, it includes a lidar multi-modal noise floor matrix acquisition module, a lidar joint pulse phase encoding acquisition module, a lidar spatio-temporal marked echo data acquisition module, a lidar multi-dimensional target feature tensor acquisition module, a lidar enhanced target point cloud acquisition module, and an update module. Among them: The lidar multi-modal noise floor matrix acquisition module is used to generate a lidar multi-modal noise floor matrix through quantum-classical noise fusion modeling; The lidar joint pulse phase encoding acquisition module is used to select a low-noise frequency band using quantum annealing based on the noise distribution of the lidar multi-modal noise floor matrix, and generate lidar joint pulse phase encoding in combination with chaotic perturbation; The lidar spatio-temporal marked echo data acquisition module is used to drive metasurface beamforming according to the phase parameters of the lidar joint pulse phase encoding and collect lidar spatio-temporal marked echo data; The lidar multi-dimensional target feature tensor acquisition module is used to perform quantum convolution filtering with the lidar spatio-temporal marked echo data as the original signal in combination with the noise threshold of the lidar joint pulse phase encoding to generate a lidar multi-dimensional target feature tensor; The lidar enhanced target point cloud acquisition module is used to optimize the confidence threshold through federated reinforcement learning based on the time-frequency - space features of the lidar multi-dimensional target feature tensor and output the lidar enhanced target point cloud; The update module is used to analyze the topological structure of the lidar enhanced target point cloud and drive the metamaterial gene evolution algorithm to update the lidar spatio-temporal marked echo data and the lidar joint pulse phase encoding.

[0126] An electronic device includes: a processor; and a memory in which computer program instructions are stored, and when the computer program instructions are run by the processor, the processor is caused to execute the lidar control method as described above.

[0127] A computer-readable storage medium is used to store a program, and when the program is executed by a processor, the lidar control method as described above is implemented.

[0128] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, 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. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0129] The present invention is described with reference to flowchart illustrations and / or block diagrams of systems, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each flow and / or block in the flowchart illustrations and / or block diagrams, and combinations of flows and / or blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions may 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 executed by the processor of the computer or other programmable data processing apparatus create means for implementing the functions specified in the flow Figure 1 one flow or more flows and / or blocks Figure 1 one block or more blocks.

[0130] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the functions specified in the flow Figure 1 one flow or more flows and / or blocks Figure 1 one block or more blocks.

[0131] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in the flow Figure 1 one flow or more flows and / or blocks Figure 1 one block or more blocks.

[0132] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present invention.

[0133] It is obvious that those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A lidar control method, characterized in that, It includes the following steps: Generate the lidar multimodal noise basis matrix through quantum-classical noise fusion modeling; Based on the noise distribution of the lidar multimodal noise basis matrix, use quantum annealing to select low-noise frequency bands, and combine chaotic perturbation to generate the lidar joint pulse phase encoding; Drive metasurface beamforming according to the phase parameters of the lidar joint pulse phase encoding, and collect lidar spatio-temporal marked echo data; Taking the lidar spatio-temporal marked echo data as the original signal, perform quantum convolution filtering in combination with the noise threshold of the lidar joint pulse phase encoding to generate the lidar multi-dimensional target feature tensor; Based on the time-frequency - spatial features of the lidar multi-dimensional target feature tensor, optimize the confidence threshold through federated reinforcement learning, and output the lidar enhanced target point cloud; Analyze the topological structure of the lidar enhanced target point cloud, and drive the metamaterial gene evolution algorithm to update the lidar spatio-temporal marked echo data and the lidar joint pulse phase encoding.

2. The lidar control method according to claim 1, wherein, Generating the lidar multimodal noise basis matrix includes the following steps: In the lidar initialization stage, turn off the transmitting module, activate all receiving channels, and synchronously collect three-dimensional noise data in the time - space - frequency domain: Time dimension: Continuously collect for a set duration , and divide it into several time slices ; Spatial dimension: Record the coordinate offset of each receiving unit ; Frequency domain dimension: For each time slice perform multi-order Fourier transform on the signal and extract the frequency band energy ; Integrate a superconducting quantum interference device at the receiving end to measure the quantum noise density matrix of the environmental electromagnetic field , and fuse it with the three-dimensional noise data in the time-space-frequency domain to construct a lidar multimodal noise floor matrix , with the structure as follows: ； Among them, is the weighting factor of is the weighting factor of is the weighting factor of is the quantum noise coupling coefficient of Generate extreme environmental noise samples based on the trained conditional generative adversarial network , expand the boundary conditions of the lidar multi-modal noise base matrix, and inject it into the lidar multi-modal noise base matrix at a ratio of 5% : ； At each set interval , resample and update in the lidar multimodal noise floor matrix , , and .

3. The lidar control method according to claim 2, characterized in that, Generating the lidar joint pulse phase encoding includes the following steps: Perform frequency-domain slice analysis on the lidar multimodal noise basis matrix, and calculate the comprehensive noise intensity of each frequency band; Map the frequency band selection problem to the Ising model, solve for the optimal frequency band combination using a quantum annealing machine, and select the three frequency bands with the lowest combined noise intensity as the primary-secondary avoidance frequency bands , , representing the three lowest noise frequency bands respectively; Generate a triple phase shift keying waveform: Main frequency band : Pulse width , Phase angle ; Secondary frequency band : Pulse width , Phase angle ; Secondary frequency band : Pulse width , Phase angle ; Generate an undisturbed base TPK waveform : , , where t is the current time, is the pi, is the rectangular window function, defining the existence interval of the pulse in the time domain ; Inject chaotic perturbations into each triple-phase keying waveform , generating an enhanced waveform after perturbation: , where is the delay time, and is the chaotic gain coefficient; Superimpose the enhanced waveforms after perturbation to generate lidar joint pulse phase coding ; Dynamically adjust the transmission energy of each frequency band according to the comprehensive noise intensity of each frequency band in the lidar multimodal noise floor matrix .

4. A lidar control method according to claim 3, wherein Transmission energy The calculation formula is as follows: , is a system constant, , , are respectively the main frequency band , the secondary frequency band and the secondary frequency band of the combined noise intensity.

5. A method for controlling a lidar according to claim 1, characterized in that, Collecting the lidar spatio-temporal marked echo data includes the following steps: Divide the transmitting array into multiple groups, and each group is spatially deflected according to the golden section angle ; Integrate programmable metasurface units at the transmitting end and dynamically adjust the phase distribution of the programmable metasurface units according to the golden ratio angle : , is the laser wavelength, is the pi, are the two-dimensional planar coordinates of the programmable metasurface unit; Each group of transmitters emits laser according to the parameters of lidar joint pulse phase coding, and the emission time uses pseudo-random delay; Use entangled photon pairs to calibrate the transmitter clock and the receiver clock; Record the absolute timestamp of each photon arrival at the receiving end and the spatial coordinates ; Encapsulate the original data into the initial spatio-temporal marked echo data ; , where m is the photon number, is the voltage value corresponding to the photon energy; According to the spatial deflection angle of the transmitting group, construct a spatial correlation mask: ； Among them, is the mask value of the receiving unit , and are respectively the abscissa offset and ordinate offset of the receiving unit relative to the reference position; Reserved mask value The initial spatio-temporal marked echo data with a value of 1, denoted as lidar spatio-temporal marked echo data .

6. A lidar control method according to claim 5, characterized in that, Generating the lidar multi-dimensional target feature tensor includes the following steps: For the spatio-temporal marked echo data of lidar Perform 5-layer wavelet packet decomposition to extract the time-frequency energy of the m-th photon ; Decompose the lidar spatio-temporal marked echo data into the core tensor G and the factor matrix, and retain the core tensor G as the denoised data: ； Among them, is the time factor matrix, is the space factor matrix, is the total number of receiving units, is the frequency domain factor matrix; Calculate the spatial correlation factor : ； is the set of adjacent units of the receiving unit where photon m is located and is the number of adjacent units and is the original signal amplitude of the adjacent unit; Construct a quantum convolutional kernel whose Hamiltonian is driven by the quantum noise terms of the MNM-Tensor, and calculate the fidelity loss after performing quantum filtering ; Construct a dual-threshold filtering function: ； is the spatial consistency threshold, is the amplitude of the filtered signal, is the main frequency band , the secondary frequency band and the secondary frequency band is the mean noise energy; Constructing a multi-dimensional target feature tensor for lidar : 。 7. A lidar control method according to claim 1, characterized in that, Outputting the lidar enhanced target point cloud includes the following steps: Obtain the comprehensive confidence level ; Obtain the confidence threshold stored in the database, screen the target points, and generate the lidar enhanced target point cloud : ； is the three-dimensional coordinates of the target point in the radar coordinate system, is the confidence weight of the target point, is the comprehensive confidence of the target point. When is greater than the set threshold parameter, the target is retained; otherwise, it is determined as noise and eliminated.

8. A lidar control method according to claim 7, wherein Obtain the comprehensive confidence, and the process is as follows: Define the confidence index: Time-frequency stability : The standard deviation of the energy fluctuation of the target signal within the time window is determined based on the target signal energy of each time slice of the lidar multi-dimensional target feature tensor; Spatial aggregation degree : The ratio of the number of target signal energies greater than the set threshold in each time slice of the multi-dimensional target feature tensor of the lidar to the standardized volume; Energy significance : The ratio of the peak energy of the target signal to the energy of the background noise; Calculating the Betti numbers of the multi-dimensional target feature tensor of lidar , , namely the number of connected components and the number of circular structures respectively, to quantify the connectivity and hole characteristics of the target structure; Obtain topological significance index : ; ； is the weight coefficient of is the weight coefficient of is the weight coefficient of is the weight coefficient of.

9. A lidar control method according to claim 1, characterized in that, Analyzing the topological structure of the lidar enhanced target point cloud, and driving the metamaterial gene evolution algorithm to update the lidar spatio-temporal marked echo data and the lidar joint pulse phase encoding includes the following steps: Perform Delaunay triangulation on the lidar-enhanced target point cloud to extract the environmental topology map , where the vertex V is the target point and the edge E is the adjacent relationship; Calculate the average degree of the computational environment topology graph and the clustering coefficient ; Map the lidar enhanced target point cloud to the Chern-Simons action SCS; Determine the environmental topological category according to the action SCS; For dense scenes: shorten the pulse width of the lidar spatio-temporal marked echo data, and increase the update frequency of the lidar joint pulse phase encoding; For sparse scenes: expand the number of frequency bands of the lidar spatio-temporal marked echo data, and add new auxiliary frequency bands; Encode the programming parameters of the programmable metasurface unit into a binary gene chain, and optimize the beamforming parameters through genetic algorithm crossover and mutation, including the phase distribution of each programmable metasurface unit, the chaotic gain coefficient of each transmitting unit, and the beam focusing offset.

10. A lidar control system for the lidar control method according to any one of claims 1-9, characterized in that, It includes a lidar multimodal noise basis matrix acquisition module, a lidar joint pulse phase encoding acquisition module, a lidar spatio-temporal marked echo data acquisition module, a lidar multi-dimensional target feature tensor acquisition module, a lidar enhanced target point cloud acquisition module, and an update module, where: The lidar multi-modal noise basis matrix acquisition module is used to generate the lidar multi-modal noise basis matrix through quantum-classical noise fusion modeling; The lidar joint pulse phase encoding acquisition module is used to select low-noise frequency bands using quantum annealing based on the noise distribution of the lidar multi-modal noise basis matrix, and combine chaotic perturbation to generate the lidar joint pulse phase encoding; The lidar spatio-temporal marked echo data acquisition module is used to drive metasurface beamforming according to the phase parameters of the lidar joint pulse phase encoding and collect the lidar spatio-temporal marked echo data; The lidar multi-dimensional target feature tensor acquisition module is used to perform quantum convolution filtering on the lidar spatio-temporal marked echo data as the original signal in combination with the noise threshold of the lidar joint pulse phase encoding to generate the lidar multi-dimensional target feature tensor; The lidar enhanced target point cloud acquisition module is used to optimize the confidence threshold through federated reinforcement learning based on the time-frequency - spatial features of the lidar multi-dimensional target feature tensor and output the lidar enhanced target point cloud; The update module is used to analyze the topological structure of the lidar enhanced target point cloud and drive the metamaterial gene evolution algorithm to update the lidar spatio-temporal marked echo data and the lidar joint pulse phase encoding.

Images