A radar inversion method and device for robot perception

Abstract

The application provides a radar inversion method and device for robot perception, acquires radar observation data, constructs a corresponding observation model and calculates a data consistency measurement value; calculates a local geometric consistency measurement value according to the observation data and a constructed weight matrix; calculates a global geometric consistency measurement value according to the observation data and a constructed transmission matrix; optimally optimizes the transmission matrix to obtain a corresponding optimal transmission matrix; constructs a unified optimization model according to the data consistency measurement value, the local geometric consistency measurement value, the global geometric consistency measurement value and the optimal transmission matrix, and calculates an inversion result. Under the unified optimization model framework, the signal observation model, the local geometric consistency in the radar observation data and the global geometric consistency are jointly modeled, stable estimation of target parameters is realized, and the perception stability and structural expression capability of the robot in a complex environment are improved.

Description

Technical Field

[0001] This invention relates to the field of radar signal processing, and in particular to a radar inversion method and apparatus for robot perception. Background Technology

[0002] With the rapid development of mobile robots, autonomous driving systems, and unmanned platforms, the robot's perception capabilities in complex dynamic environments have become a key foundation for system performance. Robots need to make stable and continuous estimates of the spatial position, motion state, and structural features of surrounding targets to support tasks such as path planning, target recognition, and behavioral decision-making. Millimeter-wave radar, due to its advantages such as all-weather operation, insensitivity to lighting conditions, and the ability to directly provide distance and Doppler information, has become one of the important sensors in robot perception systems. However, compared to vision or lidar, millimeter-wave radar echoes have limited spatial resolution. Its perception results typically appear as sparse, discrete, and unstable point clouds. The same object often consists of only a small number of scattering points, and these points fluctuate significantly between different time frames, making it difficult to directly reflect the geometric structure of the real object.

[0003] Existing radar signal processing methods typically rely on a range-Doppler-angle processing flow, combined with super-resolution algorithms such as MUSIC, ESPRIT, or sparse reconstruction to estimate target parameters. The optimization objective of these methods primarily revolves around the consistency between the signal and the model, performing independent parameter estimation for each scattering point, lacking modeling and constraints on the spatial relationships between scattering points. While these methods offer advantages in resolution, they are prone to the following problems in practical applications: the point distribution of the same object is discrete and discontinuous; weak scattering points are difficult to retain stably against a noisy background; points from different objects are prone to aliasing; and multipath echoes generated by specular reflection are difficult to distinguish from the real target. These problems essentially stem from the fact that existing methods neglect the geometric consistency and structural correlation between target points.

[0004] To improve point cloud quality, existing systems typically introduce clustering (such as DBSCAN) and multi-object tracking methods (such as Kalman filtering or data association algorithms) to structure the point cloud during post-processing. However, these methods are performed after front-end parameter estimation, when the input data has already undergone threshold detection or sparse reconstruction, making some weak object information unrecoverable and limiting the overall perception results. Furthermore, in dynamic scenes, due to changes in viewpoint, occlusion effects, and unstable point counts, it is difficult to establish stable correspondences between the same object in different frames, and cross-frame associations are prone to failure, thus affecting trajectory continuity and structural consistency. In addition, clustering and tracking methods often rely on empirical parameters or heuristic rules, making it difficult to maintain robustness in complex environments.

[0005] Therefore, existing millimeter-wave radar sensing methods typically employ a phased processing flow of "signal estimation—point cloud generation—post-processing analysis," treating observation data, which is essentially generated by spatial structure and motion laws, as independent scattering points, lacking a unified modeling and optimization framework. This approach fails to fully utilize the geometric relationships and temporal continuity between scattering points, easily leading to insufficient information utilization and error accumulation.

[0006] Therefore, there is an urgent need for a radar inversion method and device for robot perception to improve the above problems. Summary of the Invention

[0007] The purpose of this invention is to provide a radar inversion method and apparatus for robot perception, which can improve the robot's perception stability and structural representation ability in complex environments, thereby improving the overall performance of the robot perception system.

[0008] In a first aspect, the present invention provides a radar inversion method for robot perception, comprising the steps of: acquiring radar observation data, constructing a corresponding observation model and calculating a data consistency metric; calculating a local geometric consistency metric based on the observation data and the constructed weight matrix; calculating a global geometric consistency metric based on the observation data and the constructed transmission matrix; performing optimal transmission optimization on the transmission matrix to obtain a corresponding optimal transmission matrix; constructing a unified optimization model based on the data consistency metric, the local geometric consistency metric, the global geometric consistency metric, and the optimal transmission matrix, and calculating the inversion result.

[0009] Optionally, the data consistency metric value is:

[0010]

[0011] in, for Radar observation data at any given time; for The first moment The scattering amplitude coefficient of each scattering component; for The first moment The parameter representation of each scattering component; for The corresponding forward model.

[0012] Optionally, calculating the local geometric consistency metric based on the observed data and the constructed weight matrix includes: mapping the observed data from the parameter space to the embedding space, constructing a weight matrix based on neighborhood relationships, and calculating the local geometric consistency metric; and / or the local geometric consistency metric is:

[0013]

[0014] in, for The first moment The scattering component and the first scattering component A weight matrix based on neighborhood relationships between the scattering components, and It is a similarity function; for The first moment Geometric embedding vectors of each scattering component; for The first moment The geometric embedding vector of each scattering component.

[0015] Optionally, calculating the global geometric consistency metric based on the observed data and the constructed transmission matrix includes: constructing a cross-frame geometric cost function based on the observed data and a distance or inconsistency metric function, and calculating the global geometric consistency metric based on the cross-frame geometric cost function and the constructed transmission matrix; and / or the global geometric consistency metric is:

[0016]

[0017] in, for The first moment Each scattering component and The first moment The transmission matrix of each scattering component; for The first moment Each scattering component and The first moment Number of cross-frame geometric cost functions for each scattering component.

[0018] Optionally, performing optimal transmission optimization on the transmission matrix to obtain the corresponding optimal transmission matrix includes: taking minimizing the cross-frame geometric cost function as the objective, introducing an entropy regularization term to construct an optimization objective, and solving for the corresponding optimal transmission matrix by combining edge constraint conditions.

[0019] Optionally, constructing a unified optimization model based on the data consistency metric, local geometric consistency metric, global geometric consistency metric, and optimal transfer matrix, and calculating the inversion result includes: constructing a unified optimization model based on the data consistency metric, local geometric consistency metric, global geometric consistency metric, and optimal transfer matrix; calculating the inversion result by iteratively converging the unified optimization model through an alternating optimization strategy; and / or the unified optimization model is a globally minimizing objective function, and the inversion result includes radar scattering point inversion parameters and structured point cloud.

[0020] Secondly, the present invention provides a radar inversion device for robot perception, the device comprising modules / units for performing any of the possible design methods described in the first aspect above. These modules / units can be implemented in hardware or by hardware executing corresponding software.

[0021] Thirdly, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a program executable on the processor, and when the program is executed by the processor, the electronic device implements a method for performing any of the possible designs described above.

[0022] Fourthly, the present invention provides a readable storage medium storing a program, which, when executed, implements a method of any possible design of any of the above aspects.

[0023] Fifthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0024] The beneficial effects of the method of this invention are as follows: acquiring radar observation data, constructing a corresponding observation model, and calculating a data consistency metric; calculating a local geometric consistency metric based on the observation data and the constructed weight matrix; calculating a global geometric consistency metric based on the observation data and the constructed transmission matrix; performing optimal transmission optimization on the transmission matrix to obtain a corresponding optimal transmission matrix; constructing a unified optimization model based on the data consistency metric, local geometric consistency metric, global geometric consistency metric, and optimal transmission matrix, and calculating the inversion result. Under the framework of the unified optimization model, the signal observation model, the local geometric consistency in the radar observation data, and the global geometric consistency are jointly modeled to achieve stable estimation of target parameters, thereby improving the robot's perception stability and structural representation ability in complex environments. Attached Figure Description

[0025] Figure 1 A schematic flowchart of a radar inversion method for robot perception provided in an embodiment of the present invention;

[0026] Figure 2 A schematic diagram of a radar inversion device for robot perception provided in an embodiment of the present invention;

[0027] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention. Unless otherwise defined, the technical or scientific terms used herein should have the ordinary meaning understood by those skilled in the art. The terms "comprising" and similar expressions used herein mean that the element or object preceding the word covers the element or object listed following the word and its equivalents, but do not exclude other elements or objects.

[0029] The technical solutions of the embodiments of the present invention will be described below with reference to the accompanying drawings. In the description of the embodiments of the present invention, the terminology used in the following embodiments is for the purpose of describing specific embodiments only and is not intended to limit the present invention. The singular expressions “a,” “the,” “the,” and “this” are intended to also include expressions such as “one or more,” unless the context clearly indicates otherwise. It should also be understood that in the following embodiments of the present invention, “at least one” and “one or more” refer to one or more (including two). The term “and / or” is used to describe the relationship between related objects, indicating that three relationships can exist; for example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character “ / ” generally indicates that the preceding and following related objects are in an “or” relationship.

[0030] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the invention. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," and "in still other embodiments" appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including, but not limited to," unless otherwise specifically emphasized. The term "connection" includes both direct and indirect connections, unless otherwise stated. "First" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated.

[0031] In embodiments of the present invention, "exemplarily" or "for example" are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design described as "exemplarily" or "for example" in embodiments of the present invention should not be construed as being more preferred or advantageous than other embodiments or design solutions. Rather, the use of "exemplarily" or "for example" is intended to present the relevant concepts in a specific manner.

[0032] like Figure 1 As shown, this invention provides a radar inversion method for robot perception, including the following steps:

[0033] S101: Acquire radar observation data, construct the corresponding observation model, and calculate the data consistency metric.

[0034] In some embodiments, the data consistency metric is:

[0035]

[0036] in, for Radar observation data at any given time; for The first moment The scattering amplitude coefficient of each scattering component; for The first moment The parameter representation of each scattering component; for The corresponding forward model.

[0037] S102, calculate the local geometric consistency metric value based on the observed data and the constructed weight matrix.

[0038] In some embodiments, calculating the local geometric consistency metric based on the observed data and the constructed weight matrix includes: mapping the observed data from the parameter space to the embedding space, constructing a weight matrix based on neighborhood relationships, and calculating the local geometric consistency metric.

[0039] In other embodiments, the local geometric consistency metric is:

[0040]

[0041] in, for The first moment The scattering component and the first scattering component A weight matrix based on neighborhood relationships between the scattering components, and It is a similarity function; for The first moment Geometric embedding vectors of each scattering component; for The first moment The geometric embedding vector of each scattering component.

[0042] S103, calculate the global geometric consistency metric value based on the observed data and the constructed transfer matrix.

[0043] In some embodiments, calculating the global geometric consistency metric based on the observation data and the constructed transmission matrix includes: constructing a cross-frame geometric cost function based on the observation data and a distance or inconsistency metric function, and calculating the global geometric consistency metric based on the cross-frame geometric cost function and the constructed transmission matrix.

[0044] In other embodiments, the global geometric consistency metric is:

[0045]

[0046] in, for The first moment Each scattering component and The first moment The transmission matrix of each scattering component; for The first moment Each scattering component and The first moment The cross-frame geometric cost function for each scattering component.

[0047] S104, perform optimal transmission optimization on the transmission matrix to obtain the corresponding optimal transmission matrix.

[0048] In some embodiments, optimizing the transmission matrix to obtain the corresponding optimal transmission matrix includes: constructing an optimization objective by introducing an entropy regularization term with the goal of minimizing the cross-frame geometric cost function, and solving for the corresponding optimal transmission matrix by combining edge constraint conditions.

[0049] S105. Construct a unified optimization model based on the data consistency metric, local geometric consistency metric, global geometric consistency metric, and optimal transfer matrix, and calculate the inversion result.

[0050] In some embodiments, constructing a unified optimization model based on the data consistency metric, local geometric consistency metric, global geometric consistency metric, and optimal transfer matrix, and calculating the inversion result includes: constructing a unified optimization model based on the data consistency metric, local geometric consistency metric, global geometric consistency metric, and optimal transfer matrix; and calculating the inversion result by iteratively converging the unified optimization model through an alternating optimization strategy.

[0051] In other embodiments, the unified optimization model is a globally minimizing objective function, and the inversion results include radar scattering point inversion parameters and structured point clouds.

[0052] The advantage of this invention is that, under a unified optimization model framework, the signal observation model, the local geometric consistency in radar observation data, and the global geometric consistency are jointly modeled to achieve stable estimation of target parameters and improve the robot's perception stability and structural expression ability in complex environments.

[0053] To facilitate understanding, this embodiment further elaborates on the specific implementation process of the above method in conjunction with a specific application scenario, which includes the following steps:

[0054] (1) Observation model

[0055] Let the current time (i.e.) (Time) Observation data is The set of parameters to be estimated is:

[0056]

[0057] in, for The first moment The parameters of each scattering component are represented as follows: It is a positive integer. The total number of scattering components to be estimated. The corresponding forward model is denoted as Then the observation satisfies:

[0058]

[0059] in, for The first moment The scattering amplitude coefficient of each scattering component; For noise;

[0060] The data consistency metric is:

[0061]

[0062] (2) Internal geometric consistency constraints (i.e., local geometric consistency)

[0063] Suppose there exists a mapping from the parameter space to the embedding space:

[0064]

[0065] in, for The first moment Geometric embedding vectors of each scattering component (e.g., spatial location or low-dimensional embedding). for The first moment The geometric embedding vector of each scattering component (such as spatial location or low-dimensional embedding).

[0066] Build The first moment The scattering component and the first scattering component The weight matrix based on neighborhood relationships between the scattering components:

[0067]

[0068] in, It is a similarity function.

[0069] The local geometric consistency metric is:

[0070]

[0071] This item is used to constrain the local structural consistency of parameters in the embedding space.

[0072] (3) Cross-frame geometric flow consistency modeling (i.e. global geometric consistency)

[0073] Let the previous time (i.e.) The set of parameters for time (time) is:

[0074]

[0075] Introducing a transfer matrix:

[0076]

[0077] in, Indicates from The first moment Each scattering component is directed towards The first moment The matching relationship or mass flow of each scattering component.

[0078] Define the cross-frame geometric cost function:

[0079]

[0080] in, This is a distance or inconsistency metric function.

[0081] Global geometric consistency metric:

[0082]

[0083] Used to characterize the continuity and structural retention of parameters over time.

[0084] (4) Optimization of transmission matrix

[0085] The transmission matrix is ​​determined through the optimal transmission problem:

[0086]

[0087] Regularization terms (such as entropy regularization) can be combined to improve solution stability:

[0088]

[0089] And satisfy the edge constraint conditions (optional):

[0090]

[0091] (5) Unified optimization model

[0092] Unifying the above, we obtain the complete inversion problem:

[0093]

[0094] Right now:

[0095]

[0096] (6) Optimize the solution strategy

[0097] An alternating optimization strategy is used to solve the problem.

[0098] fixed Solve for the optimal transfer matrix

[0099] fixed Optimize parameters

[0100] The process is iteratively updated until convergence, yielding inversion results including radar scattering point inversion parameters and structured point clouds.

[0101] In summary, by introducing internal geometric consistency constraints (i.e., local geometric consistency metrics) into the inversion process, ) and cross-frame geometric flow consistency constraints (i.e., global geometric consistency metric) This invention achieves structural constraints in the parameter space and continuity constraints in the time dimension within a unified optimization framework, thereby ensuring that the estimation results not only satisfy the consistency of the observed data, but also possess good geometric structure representation and cross-frame stability.

[0102] The key to the embodiments of the present invention lies in:

[0103] (1) Introducing geometric consistency constraints during the inversion process enables a shift from "point-level estimation" to "structure-level estimation." Traditional methods only estimate each scattering component independently based on observational data, lacking modeling of the structural relationships between parameters. This invention introduces internal geometric consistency constraints into the unified optimization model:

[0104]

[0105] This mechanism enables the parameter estimation process to simultaneously satisfy data consistency and local structure consistency, thereby transforming the original "point-level independent estimation" into "structure-aware joint estimation." This mechanism can automatically strengthen the correlation between parameters within the same structure during the optimization process without relying on explicit clustering.

[0106] (2) Based on the cross-frame geometric flow modeling of optimal transmission, realize the temporal consistency constraint of soft association.

[0107] To address the problem that cross-frame association in traditional methods relies on hard matching or heuristic rules, this invention introduces an optimal transmission matrix. Construct cross-frame geometric flow consistency terms:

[0108]

[0109] in, This represents a soft association relationship between frames. Compared to traditional data association methods, this approach does not rely on a fixed number or one-to-one correspondence, can naturally handle changes in the number of points and occlusion, and can describe the matching relationship between targets in the form of continuous weights, thereby achieving robust cross-frame structural consistency modeling.

[0110] (3) Deep coupling of geometric constraints and signal inversion avoids information loss in post-processing.

[0111] Existing methods typically place geometric processing (such as clustering and tracking) after inversion, resulting in structural information from the original signal being excluded from parameter estimation. This invention directly embeds geometric consistency constraints into the inversion objective function:

[0112]

[0113] This mechanism enables joint optimization of the signal model and the geometric model. It allows geometric information to directly influence the parameter solution during the inversion process, thereby suppressing noise-induced discrete estimation, improving the stability of weak targets, and reducing interference from multipath or anomalies.

[0114] (4) A unified spatiotemporal geometric modeling framework to achieve coordinated constraints on spatial structure and temporal continuity.

[0115] This invention simultaneously introduces, within the same optimization framework, the internal geometric consistency constraint of the spatial domain. ) and temporal cross-frame consistency constraints ( This framework forms a unified spatiotemporal geometric modeling mechanism. During the optimization process, it simultaneously constrains the continuity of local structures and the smoothness of cross-frame evolution, thereby avoiding the error accumulation problem caused by traditional staged processing and achieving a stable representation of dynamic targets.

[0116] (5) The implicit structure modeling mechanism based on similarity weights reduces the dependence on priors and parameters.

[0117] This invention constructs weights based on a similarity function. This mechanism achieves implicit modeling of structural relationships without requiring explicit object partitioning or labeling information. It avoids the strong dependence of traditional clustering methods on parameters such as thresholds and neighborhood scales, enabling the system to adaptively form structural constraints in complex environments, thus improving overall robustness and generalization ability.

[0118] The advantages of the embodiments of the present invention are as follows:

[0119] (1) Significantly improves the stability and continuity of point cloud structure

[0120] By introducing internal geometric consistency constraints during the inversion process, this invention can directly constrain the local structural relationships between parameters during the optimization stage, ensuring that multiple scattering components corresponding to the same object remain spatially consistent. Compared to the problems of point cloud discrepancies and structural instability in traditional methods, this invention can form a more continuous and compact point set representation, thereby effectively improving the structural representation capability of the target.

[0121] (2) Enhance the detection capability under weak target and low signal-to-noise ratio conditions.

[0122] In traditional methods, weak scattering components are easily affected by noise and cannot be stably estimated. This invention utilizes geometric constraints to subject weak components to the constraints of the neighborhood structure during the optimization process, thereby preventing them from being submerged by noise. This mechanism is equivalent to introducing structural priors during the inversion process, allowing weak targets to be recovered by attaching to the overall geometric structure, thus improving the robustness of the system under low signal-to-noise ratio conditions.

[0123] (3) Improve cross-frame consistency and enhance the stability of dynamic target tracking.

[0124] By introducing cross-frame geometric flow consistency constraints based on optimal transmission, this invention can establish soft correlations between different time frames during the optimization process. Compared to traditional methods that rely on hard matching or independent tracking, this approach can maintain the continuity and consistency of the target even when the number of points changes, there is occlusion, or the observation is incomplete, thereby significantly improving the target stability and trajectory reliability in dynamic scenes.

[0125] (4) Effectively suppresses multipath and anomaly interference.

[0126] Multipath echoes or noise points typically do not satisfy spatial structure consistency and cross-frame continuity. This invention, by jointly introducing internal geometric constraints and cross-frame consistency constraints, suppresses scattering components that do not conform to the overall structure during the optimization process, thereby achieving automatic filtering of multipath and anomalies. Compared to traditional methods that rely on post-processing to remove anomalies, this invention can complete interference suppression during the inversion stage.

[0127] (5) Avoid information loss and error accumulation caused by phased processing.

[0128] Traditional methods typically employ a phased processing flow of "signal estimation—point cloud generation—clustering—tracking," with each stage operating independently, which can easily lead to insufficient information utilization and error accumulation. This invention constructs a unified optimization model that jointly models signal consistency, spatial structure constraints, and temporal consistency constraints, enabling various types of information to work synergistically within the same framework, thereby improving overall estimation accuracy and reducing system errors.

[0129] (6) Reduce reliance on empirical parameters and heuristic rules

[0130] This invention achieves adaptive characterization of structural relationships and cross-frame associations through a modeling approach based on similarity weights and optimal transmission mechanisms. This avoids the strong dependence of traditional clustering and tracking methods on thresholds, neighborhood scales, and data association rules, and improves the generalization ability and stability of the method in complex environments.

[0131] In summary, this invention achieves structural perception and temporal consistency modeling of radar observation data by introducing spatial geometric constraints and cross-frame consistency constraints during the inversion process. Compared with traditional methods, it can obtain more stable point cloud structures, more robust target detection capabilities, and more reliable cross-frame continuity performance in complex environments, significantly improving the overall performance of the robot perception system.

[0132] like Figure 2As shown, based on the above method, the present invention provides a radar inversion device for robot perception, comprising: an acquisition unit 201, used to acquire radar observation data, construct a corresponding observation model, and calculate a data consistency metric; a first calculation unit 202, used to calculate a local geometric consistency metric based on the observation data and the constructed weight matrix; a second calculation unit 203, used to calculate a global geometric consistency metric based on the observation data and the constructed transmission matrix; an optimization unit 204, used to perform optimal transmission optimization on the transmission matrix to obtain a corresponding optimal transmission matrix; and an inversion unit 205, used to construct a unified optimization model based on the data consistency metric, the local geometric consistency metric, the global geometric consistency metric, and the optimal transmission matrix, and calculate the inversion result.

[0133] It should be understood that all relevant content of each step involved in the above method embodiments can be referenced to the functional description of the corresponding functional module, and will not be repeated here. Furthermore, the use of suffixes such as "module," "component," or "unit" to represent elements is merely for the purpose of illustrative purposes and has no specific meaning in itself. Therefore, "module," "component," or "unit" can be used interchangeably. Terminals can be implemented in various forms. For example, the terminals described in this invention may include mobile terminals such as mobile phones, tablets, laptops, handheld computers, personal digital assistants (PDAs), portable media players (PMPs), navigation devices, wearable devices, smart bracelets, pedometers, etc., as well as fixed terminals such as digital TVs and desktop computers. The following description will use mobile terminals as examples; those skilled in the art will understand that, in addition to elements specifically designed for mobile purposes, the construction according to embodiments of the present invention can also be applied to fixed-type terminals.

[0134] In other embodiments of the present invention, an electronic device 300 is disclosed, such as... Figure 3 As shown, the device may include: one or more processors 301; memory 302; display 303; one or more application programs (not shown); and one or more computer programs 304. These devices can be connected via one or more communication buses 305. The one or more computer programs 304 are stored in the memory 302 and configured to be executed by the one or more processors 301. The one or more computer programs 304 include instructions that can be used to perform actions such as... Figure 1 Each step in the corresponding embodiment.

[0135] Processor 301 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0136] The memory 302 can be an internal storage unit of the electronic device 300, such as a hard disk or RAM of the electronic device 300. The memory 302 can also be an external storage device of the electronic device 300, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or FlashCard equipped on the electronic device 300. Furthermore, the memory 302 can include both internal and external storage units of the electronic device 300. The memory 302 is used to store computer programs and other programs and data required by the electronic device. The memory 302 can also be used to temporarily store data that has been output or will be output.

[0137] The computer program 304 can be divided into one or more modules / units. The one or more modules / units can be a series of computer program instruction segments that can perform a specific function. The instruction segments are used to describe the execution process of the computer program 304 in the electronic device 300.

[0138] In addition to the above-described structure, those skilled in the art will understand that Figure 3 This is merely an example of electronic device 300 and does not constitute a limitation on electronic device 300. Electronic device 300 may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0139] Those skilled in the art will understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the functions described above can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this invention. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0140] Based on the above embodiments, the present invention also discloses a computer-readable storage medium having at least one computer program stored thereon, wherein the computer program, when executed by a processor, implements the methods described in the foregoing embodiments.

[0141] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing a processor. The program can be stored in a computer-readable storage medium, which is a non-transitory medium, such as random access memory, read-only memory, flash memory, hard disk, solid-state drive, magnetic tape, floppy disk, optical disk, and any combination thereof. The storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. This available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital video disc (DVD)), or a semiconductor medium (e.g., solid-state drive (SSD)).

[0142] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0143] Although the embodiments of the present invention have been described in detail above, it will be apparent to those skilled in the art that various modifications and variations can be made to these embodiments. The above descriptions are merely embodiments of the present invention and do not limit the patent scope of the present invention. However, it should be understood that such modifications and variations fall within the scope and spirit of the present invention. Moreover, the present invention described herein may have other embodiments and can be implemented or realized in various ways. All equivalent transformations made based on the description and drawings of the present invention, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A radar inversion method for robot perception, characterized in that, Including the following steps: Acquire radar observation data, construct the corresponding observation model, and calculate the data consistency metric. The local geometric consistency metric is calculated based on the observed data and the constructed weight matrix. The global geometric consistency metric is calculated based on the observed data and the constructed transfer matrix. The transmission matrix is ​​optimized for optimal transmission to obtain the corresponding optimal transmission matrix; A unified optimization model is constructed based on the data consistency metric, local geometric consistency metric, and global geometric consistency metric, and the inversion results are calculated. set up Time observation data is The set of parameters to be estimated is as follows: in, for The first moment The parameters of each scattering component are represented as follows: It is a positive integer. The total number of scattering components to be estimated; The corresponding forward model is denoted as Then the observation satisfies: in, for Radar observation data at any given time; for The first moment The scattering amplitude coefficient of each scattering component; For noise; The data consistency metric is: Suppose there exists a mapping from the parameter space to the embedding space: in, for The first moment Geometric embedding vectors of scattering components; for The first moment Geometric embedding vectors of scattering components; Build The first moment The scattering component and the first scattering component The weight matrix based on neighborhood relationships between the scattering components: in, It is a similarity function; The local geometric consistency metric is: This item is used to constrain the local structural consistency of parameters in the embedding space; set up The set of time parameters is as follows: Introducing the transfer matrix: in, Indicates from The first moment Each scattering component is directed towards The first moment Matching relationship or mass flow of each scattering component; Define the cross-frame geometric cost function: in, For distance or inconsistency metrics; Global geometric consistency metric: Used to characterize the continuity and structural retention of parameters over time; The unified optimization model is as follows: Right now: in, , and These are the weighting coefficients; The inversion results include radar scattering point inversion parameters and structured point clouds.

2. The method according to claim 1, characterized in that, Optimal transmission optimization is performed on the transmission matrix to obtain the corresponding optimal transmission matrix, which includes: With the goal of minimizing the cross-frame geometric cost function, an entropy regularization term is introduced to construct the optimization objective, and the corresponding optimal transmission matrix is ​​obtained by combining the edge constraint conditions.

3. The method according to any one of claims 1-2, characterized in that, A unified optimization model is constructed based on the data consistency metric, local geometric consistency metric, and global geometric consistency metric, and the inversion results are calculated, including: A unified optimization model is constructed based on the data consistency metric, local geometric consistency metric, and global geometric consistency metric. The inversion result is calculated by iteratively converging the unified optimization model through an alternating optimization strategy. The unified optimization model is a globally minimized objective function.

4. A radar inversion device for robot perception, used in the method according to any one of claims 1-3, characterized in that, include: The acquisition unit is used to acquire radar observation data, construct corresponding observation models, and calculate data consistency metrics. The first calculation unit is used to calculate the local geometric consistency metric value based on the observed data and the constructed weight matrix; The second calculation unit is used to calculate the global geometric consistency metric value based on the observation data and the constructed transmission matrix; An optimization unit is used to perform optimal transmission optimization on the transmission matrix to obtain the corresponding optimal transmission matrix. The inversion unit is used to construct a unified optimization model based on the data consistency metric, local geometric consistency metric, and global geometric consistency metric, and to calculate the inversion result.

5. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores a program that can run on the processor, and when the program is executed by the processor, causes the electronic device to perform the method of any one of claims 1-3.

6. A readable storage medium storing a program, characterized in that, When the program is executed, it implements the method of any one of claims 1-3.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-3.

Images