Radar SNR distribution return descriptor for object detection and ground clutter removal
By using a combination method of SNR and density threshold in radar systems, combining clustering or machine learning algorithms to distinguish potential obstacles and ground clutter, the problem that radar systems in the existing technology is difficult to distinguish in off-road environments is improved. Detection accuracy and feasible perception.
Patent Information
- Application Number
- CN202410688027.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-05-30
- Filing Date
- 2024-05-30
- Publication Date
- 2024-12-03
AI Technical Summary
It is difficult for existing radar systems to effectively distinguish potential obstacles and ground clutter in harsh off-road environments, resulting in the signal-to-noise ratio and radar cross-sectional area hard threshold to remove valuable points, affecting the accuracy and generalizability of feasible perception.
By capturing radar sensor data, determining signal-to-noise ratio (SNR) and comparing it to the threshold, combining the density and distance range of points, grouping points using clustering algorithms or machine learning algorithms, identifying high SNR point groups as potential obstacles , The low SNR point group is ground clutter, adjust the noise and density thresholds to improve classification accuracy.
Effectively distinguishing potential obstacles and ground clutter, improving the detection accuracy of radar system in harsh environments, reducing interference to feasible perception, and enhancing the ability to identify ground features.
Smart Images

Figure CN119066558A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to radar SNR distribution return descriptors for object detection and ground clutter removal. Background Art
[0002] The autonomous mobile robot platform can utilize a system that provides terrain and feature analysis of the ground information around the robot. This information can help the robot's navigation system make decisions about where the robot can and cannot go. Therefore, the robot can avoid obstacles, which may be humans, other robots or vehicles, or stationary obstructions. As used herein, "robot" includes driver assistance or autonomous driving vehicles / machines.
[0003] Radar performs better than LiDAR in bad weather, and radar obstacle detection chips for vehicles are commercially available and cheaper than LiDAR systems. However, radar has lower resolution, is more susceptible to noise, and is easily affected by ground clutter. Radar sensors are the desired perception sensors in off-road environments due to their inherent weather and ambiguity robustness. However, off-road environments exacerbate the presence of ground clutter in radar data. As the variation in ground terrain increases, it becomes increasingly difficult to distinguish ground clutter from obstacles. Removing points based on hard thresholds such as signal-to-noise ratio (SNR), radar cross section (RCS), etc. will inevitably remove valuable points and hinder the generalizability that is critical to a viable perception solution.
[0004] It would be desirable to have a radar system that can operate in terrain with a large amount of ground clutter, such as a construction site, but with improved accuracy. Therefore, new systems, methods, and other techniques are needed to address these and other problems.
[0005] Unless otherwise indicated herein, the content described in this section of the specification is not prior art to the present application and is not admitted to be prior art by inclusion in this section. Summary of the invention
[0006] A method and apparatus for distinguishing potential obstacles from ground clutter using a signal-to-noise ratio (SNR) is provided. Radar sensor data from a radar sensor is captured. A signal-to-noise ratio (SNR) of a plurality of points in the radar sensor data is determined. The SNR of each of the plurality of points is compared to a threshold. A first group of points or groups of points having an SNR above the threshold and at least partially surrounded by points or groups of points having an SNR below the threshold is determined. The first group of points is classified as potential obstacles. A second group of points or groups of points having an SNR below the threshold is determined. The second group of points is classified as ground clutter.
[0007] In an embodiment, a density of points around each point is determined. A distance range is determined for each point. The noise threshold is adjusted lower for points with longer distance ranges. The density threshold is adjusted lower for longer distance ranges. The density threshold for a point is required to exceed the density threshold for classifying the point as part of a potential obstacle.
[0008] In one embodiment, a clustering algorithm (such as the DBSCAN clustering algorithm) is used to group the points into clusters. The clusters are then identified as objects (potential obstacles) or noise (e.g., ground clutter) based on density and SNR. In another embodiment, a machine learning algorithm is used to group the points into potential obstacles.
[0009] This summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to be used alone to determine the scope of the claimed subject matter. The subject matter should be understood by reference to appropriate portions of the entire specification of the present invention, any or all of the drawings.
[0010] The above and other features and examples will be described in more detail in the following specification and accompanying drawings.
[0011] The terms and expressions used are used as descriptive and non-limiting terms only, and the use of these terms and expressions is not intended to exclude any equivalents of the features shown and described or a portion thereof. However, it should be recognized that various modifications are possible within the scope of the claimed system and method. Therefore, although the system and method of the present application have been specifically disclosed by way of examples and optional features, those skilled in the art should recognize modifications and variations of the concepts disclosed herein, and such modifications and variations are considered to be within the scope of the system and method defined herein. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] The accompanying drawings, which are included to provide a further understanding of the present invention and are incorporated into and constitute a part of this specification, illustrate embodiments of the present invention and together with the detailed description are used to explain the principles of the present invention. No attempt is made to show the structural details of the present invention in more detail than is necessary for a basic understanding of the present invention and various modes of practicing the present invention.
[0013] Figure 1 An arrangement of a radar sensor according to an embodiment on a vehicle having a soil compactor is shown.
[0014] Figure 2 is a diagram illustrating classification of point clusters into objects or ground clutter, according to an embodiment.
[0015] Figure 3 is a graph illustrating the effect of distance range on the classification of point clusters as objects or ground clutter, according to an embodiment.
[0016] Figure 4 is a diagram illustrating classification of points into clusters based on density using the DBSCAN algorithm, according to an embodiment.
[0017] Figure 5 is a diagram illustrating classification of points as potential obstacles based on both density and SNR, according to an embodiment.
[0018] Figure 6 is a graph illustrating the classification of points as potential obstacles using SNR fed into a machine learning algorithm, according to an embodiment.
[0019] Figure 7 is a block diagram illustrating an optional partitioning of processing between a vehicle processor and the cloud according to an embodiment.
[0020] Figure 8 A simplified computer system for executing the processes of the embodiments is shown according to the embodiments. DETAILED DESCRIPTION
[0021] In an embodiment, the radar sensor is mounted on a mobile machine, such as Figure 1 As shown. Points (e.g., pixels) of radar return signals are classified by their SNR. The SNR distribution is used to separate ground clutter from objects, which is particularly useful in off-road environments. Radar points can be labeled as part of ground clutter or part of an object with a descriptor (or feature if machine learning is used). As used in this article, a descriptor is something (such as a word or characteristic feature) used to describe or identify. Features (in the case of machine learning) are data used as input to a machine learning model to make predictions.
[0022] In an embodiment, clusters are identified by radar point density. It has been found that ground clutter has a more random SNR distribution across a cluster than objects. Object clusters typically have higher SNR points that are concentrated toward the center of the cluster. Therefore, the SNR distribution can be used along with other features to further improve the confidence in identifying a cluster as an object.
[0023] Radar location on vehicle
[0024] Figure 1An example vehicle 102 is shown with a soil compactor attached. The soil compactor includes a roller 104 attached to a frame 106 that moves over uneven terrain 118. A radar sensor 108 is attached to the front of the frame 106. An optional LiDAR sensor 110 can be attached to the roof 112 of the vehicle 102. The frame 106 and the compactor 104 can move laterally relative to the cab 114. The console 116 includes a computer and other sensors, including an odometer and a potentiometer at the connection with the frame 106 to detect its lateral movement. Other examples of vehicles with articulated portions may include a sprayer with a boom arm, a harvester with a harvesting table, a vehicle with a front or rear suspension mechanism, etc. Embodiments of the present invention may be applied to all such vehicles, as well as other vehicles without articulated portions, such as cars, trucks, boats, airplanes, drones, etc.
[0025] The radar sensor 108 is mounted closer to the ground. Existing sensors are tilted upward so that they can detect people and objects on the ground with minimal interference from ground detection (ground clutter). In one embodiment, the radar sensor is positioned so that the emitted radar lobe is parallel to the ground. The embodiments of the invention presented herein reduce / eliminate the need for an upward radar sensor tilt. This is beneficial because an upward tilt will not detect small objects that should be detected and may stop the vehicle. In one embodiment, the radar sensor is a SmartMicro 153 radar sensor.
[0026] In one embodiment, Figure 1 The optional LiDAR sensor 110 is mounted high on the roof 112 so that it has a wide and deep field of view (FOV) that is not obstructed by components of the vehicle 102. In one embodiment, the LiDAR sensor 110 provides a 180 degree or 360 degree view by using a scanning mirror, a rotating mount, or a solid-state mechanism. In one embodiment, the LiDAR sensor is a Cepton P60 LiDAR sensor or an Ouster OS0-128 LiDAR sensor.
[0027] Point Cloud
[0028] Figure 2 is a diagram illustrating classification of point clusters into objects or ground clutter, according to an embodiment. Figure 2 is a simplified example with only a few points. Two clusters of points in front of vehicle 102 are shown, as indicated by edge lines 202 and 204. The SNR of each point is determined. Graph 203 shows Figure 2102. The gray center point 214 of cluster 204 has a high SNR, while the black center point 206 of cluster 202 has a low SNR. The other points in cluster 204 are points 216, 218, and 220. These points have high SNRs, just like center point 214. Cluster 204 does not have any points immediately surrounding it. The combination of clusters exceeding the noise threshold and surrounding points (pixels) that do not exist or have low SNRs indicates a possible object and a potential obstacle for vehicle 102. Therefore, cluster 204 can be characterized as a possible object using appropriate descriptors.
[0029] The other points in cluster 202 are points 208, 210, and 212. These points also have low SNRs, or have a mixture of low and high SNRs in a random pattern. There are also no points immediately surrounding cluster 202. The combination of a cluster that does not exceed the noise threshold and surrounding points (pixels) that are absent or have low SNRs indicates possible noise or ground clutter. Therefore, vehicle 102 can safely ignore cluster 202. These figures are overly simplistic to illustrate the concepts, and there will typically be many more points in and around a cluster.
[0030] Distance Range
[0031] Figure 3 3 is a diagram illustrating the effect of distance range on classification of point clusters as objects or ground clutter according to an embodiment. A first region / arc 302 is within a radius r1 of the front center of the vehicle 102. A second region 304 is between a distance r1 and r2 from the front center of the vehicle 102. A third region 306 is between a distance r2 and r3 from the front center of the vehicle 102.
[0032] The first cluster is composed of points 308, 309, 310 and 311, all of which are shown in black as the center point. The circle around the point shows the cluster point radius around a point, and another point must be located within the cluster point radius to be considered as part of the cluster center. In addition to the points within the cluster point radius, there must also be a cluster threshold number of points within the cluster point radius. Assuming that there are a total of 4 points including the point at the center of the radius, the common cluster threshold number of points is 3 points. The cluster threshold number of points within the cluster point radius is the density threshold. As can be seen, all four black points 308-311 have 3 additional points within their radius, so the density threshold of the cluster center is met. Because there are also 3 points within the cluster point radius of a point 309, the gray point 314 has only one additional point 309 within its cluster point radius, so point 314 is defined as an edge point indicating the edge of the cluster.
[0033] Another criterion according to an embodiment is SNR. If points 308-311 all have high SNR, they will satisfy both the density criterion of an object and the first SNR criterion, and the object may be a potential obstacle for vehicle 102. The second SNR criterion is the cluster usage relative to the center point (e.g. Figure 3 The accumulated points threshold is satisfied by adding additional surrounding points with reduced SNR to the point 314 in the image above. Both the accumulated points threshold and the distance parameter between points considered as neighbors depend on the distance range. An increased distance range produces a combination of a lower accumulated points threshold and an increased distance parameter (the distance between neighboring points can be larger). Quantizing the number of points depends on the sensor model and the sliding window time (the sliding window accumulates points over the past X milliseconds so that the accumulated points on the object can be seen). For example, the SmartMicro 169 produces more points and uses a larger sliding window time than the SmartMicro 153, which will allow more points to be accumulated and therefore require a higher point threshold. For a specific sensor model and configuration, an object at a distance range of 10m with a specific SNR profile may require 10 points, while the same object at a distance range of 30m may require 5 points.
[0034] An alternative implementation is a second SNR criterion that the cluster is at least partially surrounded by points with lower SNRs, or at least by a mixture of points with high and low SNRs. No other points are shown in the immediate vicinity around the cluster of points 308-311, so this criterion is met. The closest points are points 316, 318, and 320, which, if they had low SNRs, would not be considered part of the object due to their SNRs and their density.
[0035] Points 316, 318, and 320 in white are outliers. These points may not meet the density and SNR criteria for an object and are therefore considered noise or ground clutter. Point 320 is in the same distance range 304 as clusters 308-311, so its cluster point radius is the same size. There are no other points within this cluster point radius, so the density requirement of the cluster is not met. It may also not meet the SNR requirement of the object. Similarly, points 316 and 318 have only a single point within their cluster point radius, so the density threshold of the cluster is not met. In the distance range 306, these points are farther apart. Therefore, the cluster point radius can be slightly larger to occupy a longer distance, so the separation between points is significantly wider, and the separation of points appears closer for closer points. However, in this example, even if the cluster point radius is larger, there will still be no 3 other points within the radius. In addition, the SNR of the object will be smaller in a longer distance range, so the SNR threshold is adjusted for this longer distance range.
[0036] Classic classification implementation method
[0037] Figure 4 is a diagram showing the classification of points into clusters based on density using the DBSCAN algorithm according to an embodiment. The DBSCAN algorithm is described by Ester, Martin et al. in "A density-based algorithm for discovering clusters in large spatial databases with noise", KDD, Vol. 96, No. 34, 1996. Figure 4 shows the basic concept. The cluster of point A all has at least 3 other points within a defined radius and thus forms the center of the cluster. Points B and C have no other 3 points in such a radius, but have one point, and that point is part of the center cluster. Therefore, points B and C are considered edge points. Point N does not have any other points within such a radius and is therefore an outlier. Therefore, in radar applications, cluster A is considered a potential object, and points B and C are at the edge of the object, defining its boundaries.
[0038] In one embodiment, the DBSCAN algorithm is implemented as follows:
[0039] 1. The first function determines whether two points are neighbors A simple implementation of this function would use the Euclidean distance as the metric to determine if two points are neighbors. Other variables could be included, such as by using a radar to measure distance to change the distance required to be classified as a neighbor.
[0040] 2. Query Function . For each point, the function loops through all other points, using the first function (distance) above to find neighbors. These points are classified. In the DBSCAN algorithm, they are "center points", "edge points", or "outlier points", as described in the reference article above and in detail below. In summary, a center point has a certain number of other points that form a cluster within a defined close distance. Edge points are at the edge of the cluster, and outliers are clearly outside the cluster. Neighbors are returned as "possible observation points". "Possible observation points" are radar reflections that may be part of an object. An embodiment includes changing the number of neighboring points required to consider a point as a center point based on the SNR of the point. The lower the SNR, the greater the number of surrounding points required to be considered a center point.
[0041] 3. Scaling the cluster Clusters are then identified or grown starting from the identified center point until a point is reached that satisfies the definition of an "edge point".
[0042] 4. Main LoopThe process loops through all points, calling the "expand_cluster" function on all points until all points have been assigned a cluster ID or identified as outliers.
[0043] In one embodiment, the observation threshold is not only the number of points, but also depends on the point distribution. For example, an SNR that follows a relatively smooth decreasing pattern as one moves away from the center of a cluster indicates an object.
[0044] General process
[0045] Figure 5 5 is a diagram showing classification of points as potential obstacles based on both density and SNR according to an embodiment. A radar point cloud 502 is obtained from a radar sensor. Dense point areas 504 are identified. A subset of dense point areas with surrounding low-density areas is identified 506. The SNR of each point is determined 512. The subset from step or module 506 is checked to determine which point group also has a high central SNR and is surrounded by increasingly lower SNR points that additionally meet the overall point threshold. Low-density areas separate the point groups. Clusters of points that meet both density and SNR criteria are identified as potential obstacles 510. The distance range in which each point is located is also determined 514 and used to change the required density in step 504 and the SNR threshold in step 508. At longer distance ranges, lower density is required to be considered an object, and lower SNR is required. These determinations can be performed using various algorithms, such as the DBSCAN algorithm described above, or the machine learning or AI algorithm described below.
[0046] How Machine Learning Is Implemented
[0047] Figure 6 605 is a diagram showing the classification of points as potential obstacles using SNR fed into a machine learning algorithm according to an embodiment. A radar point cloud and / or cluster 602 is provided to a trained machine learning module. A first module 604 is trained to detect features. In this case, the features will be clusters of radar points that are objects and potential obstacles. During training, clusters corresponding to different classifications are fed to the module, which include object types and types corresponding to ground clutter or noise. Each point is provided with its density information, distance range, and SNR (and optional other data). The module learns to identify the features 605 of each classification provided. These features include density and SNR distribution. The identified features 605 are then provided to a classification module 606, where the output is a classification 608, such as an object (potential obstacle), ground clutter, noise, etc. The module can also classify the type of object, such as a person, an animal, a wall, a tree, etc. Then, the module recognizes that a point with a high SNR surrounded by points with a low SNR will correspond to an object.
[0048] Figure 7 is a block diagram illustrating an optional processing partition between a vehicle processor and the cloud according to an embodiment. In one embodiment, all processing is done in the vehicle. In another embodiment, some processing is done in the cloud, such as training a machine learning module using radar point cloud data from multiple vehicles. Vehicle 702 has at least one radar sensor 704. The radar sensor is typically located on a chip that processes radar sensor data using an SNR filter 706, a radar spatial filter 708, and a radar time filter 710, as described above. Alternatively, one or more filter functions may be performed in a local processor. A local processor 712 receives the data and optionally receives ground truth data from a LiDAR sensor 714. Processor 712 uses a memory 716 for programs and data, which may be multiple memories.
[0049] The communication module 718 provides data to a remote server 722 via the Internet 720 and / or other networks. The remote server 722 receives the data via its own communication module 724 and processes the data via a server module such as a machine learning engine 726. The trained machine learning modules can then be downloaded to the vehicle 702 via the communication module 724, which provides them to the local processor 712 to analyze the radar data provided by the radar sensor and filter. In an alternative embodiment, some or all of the remote server processes may be performed by a local processor in the vehicle.
[0050] Figure 8 A simplified computer system 800 is shown in accordance with some embodiments of the present invention. Figure 8 A schematic diagram of one embodiment of a computer system 800 is provided, which can perform some or all steps of the methods provided by various embodiments. It should be noted that Figure 8 It is intended only to provide a general description of the various components, any or all of which may be used as appropriate. Figure 8 It is broadly shown how various system elements may be implemented in a relatively discrete or more integrated manner.
[0051] The computer system 800 is shown to include hardware elements, which may be electrically connected via a bus 805, or may communicate in other ways as appropriate. The hardware elements may include: one or more processors 810, including but not limited to one or more general-purpose processors and / or one or more special-purpose processors, such as digital signal processing chips, graphics acceleration processors, etc.; one or more input devices 815, which may include but are not limited to a mouse, a keyboard, a camera, etc.; and one or more output devices 820, which may include but are not limited to a display device, a printer, etc.
[0052] The computer system 800 may also include and / or communicate with one or more non-transitory storage devices 825, which may include, but are not limited to, local and / or network accessible memory, and / or may include, but are not limited to, disk drives, drive arrays, optical storage devices, solid-state storage devices, such as random access memory ("RAM") and / or read-only memory ("ROM"), which may be programmable, flash-updatable, etc. Such storage devices may be configured to implement any suitable data storage, including, but not limited to, various file systems, database structures, etc.
[0053] The computer system 800 may also include a communication subsystem 830, which may include but is not limited to a modem, a network card (wireless or wired), an infrared communication device, a wireless communication device, and / or a chipset, such as Devices, 802.11 devices, Wi-Fi devices, WiMax devices, cellular communication facilities, etc. The communication subsystem 830 may include one or more input and / or output communication interfaces to allow data to be exchanged with a network (such as the network described below), other computer systems, and / or any other device described herein. Depending on the desired functionality and / or other implementation issues, a portable electronic device or similar device can transmit images and / or other information through the communication subsystem 830. In other embodiments, a portable electronic device, such as a first electronic device, can be included in the computer system 800, for example, as an electronic device of the input device 815. In some embodiments, the computer system 800 will also include a working memory 835, which can include a RAM or ROM device as described above.
[0054] The computer system 800 may also include software elements, which are shown as currently located in the working memory 835, including an operating system 840, device drivers, executable libraries, and / or other code, such as one or more application programs 845, which may include computer programs provided by various embodiments, and / or may be designed to implement methods provided by other embodiments and / or configure systems provided by other embodiments, as described herein. By way of example only, one or more processes described with respect to the methods discussed above may be implemented as code and / or instructions executable by a computer and / or a processor within a computer; then, in one aspect, such code and / or instructions may be used to configure and / or adjust a general-purpose computer or other device to perform one or more operations according to the described methods.
[0055] The set of instructions and / or codes may be stored on a non-transitory computer-readable storage medium, such as the storage device(s) 825 described above. In some cases, the storage medium may be included within a computer system, such as computer system 800. In other embodiments, the storage medium may be separate from the computer system (e.g., removable media, such as an optical disk), and / or provided in an installation package so that the storage medium may be used to program, configure, and / or adjust a general-purpose computer having the instructions / code stored thereon. The instructions may take the form of executable code that may be executed by the computer system 800 and / or may take the form of source code and / or installable code that, when compiled and / or installed on the computer system 800, for example, using any of a variety of commonly available compilers, installers, compression / decompression utilities, etc., takes the form of executable code.
[0056] It will be apparent to those skilled in the art that substantial changes may be made depending on specific requirements. For example, customized hardware may also be used, and / or specific elements may be implemented in hardware or software (including portable software, such as applets, etc.), or both. In addition, connections to other computing devices such as network input / output devices may be employed.
[0057] As described above, in one aspect, some embodiments may use a computer system such as computer system 800 to perform methods according to various embodiments of the technology. According to one set of embodiments, some or all of the processes of these methods are performed by computer system 800 in response to processor 810 executing one or more sequences of one or more instructions, which may be included in operating system 840 and / or other code, such as application program 845 contained in working memory 835. Such instructions may be read into working memory 835 from another computer-readable medium (e.g., one or more storage devices 825). By way of example only, execution of the sequence of instructions contained in working memory 835 may cause processor(s) 810 to perform one or more processes of the methods described herein. Additionally or alternatively, portions of the methods described herein may be performed by dedicated hardware.
[0058] As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any medium that participates in providing data to enable a machine to operate in a particular manner. In an embodiment implemented using computer system 800, various computer-readable media may be involved in providing instructions / codes to (one or more) processors 810 for execution, and / or may be used to store and / or carry such instructions / codes. In many implementations, computer-readable media are physical and / or tangible storage media. Such media may take the form of non-volatile media or volatile media. Non-volatile media include, for example, optical disks and / or magnetic disks, such as (one or more) storage devices 825. Volatile media include, but are not limited to, dynamic memory, such as working memory 835.
[0059] Common forms of physical and / or tangible computer readable media include, for example, floppy disks, diskettes, hard disks, magnetic tapes or any other magnetic media, CD-ROMs, any other optical media, punch cards, paper tapes, any other physical media with patterns of holes, RAM, PROMs, EPROMs, FLASH-EPROMs, any other memory chips or cassettes, or any other medium from which a computer can read instructions and / or code.
[0060] Various forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to processor 810 for execution. By way of example only, the instructions may initially be carried on a magnetic disk and / or optical disk of a remote computer. The remote computer may load the instructions into its dynamic memory and send the instructions as signals over a transmission medium to be received and / or executed by computer system 800.
[0061] The communication subsystem 830 and / or its components will typically receive the signal, and the bus 805 may then carry the signal and / or the data, instructions, etc. carried by the signal to the working memory 835, from which the processor(s) 810 retrieve and execute the instructions. The instructions received by the working memory 835 may optionally be stored on the non-transitory storage device 825 before or after execution by the processor(s) 810.
[0062] The methods, systems, and devices discussed above are examples. Various configurations may omit, replace, or add various processes or components as appropriate. For example, in alternative configurations, the methods may be performed in an order different from that described, and / or various stages may be added, omitted, and / or combined. In addition, features described with respect to certain configurations may be combined in various other configurations. Different aspects and elements of the configurations may be combined in a similar manner. In addition, technology is developing, and therefore, many elements are examples and do not limit the scope of the invention or claims.
[0063] Specific details are given in the specification to provide a comprehensive understanding of the exemplary configurations including implementations. However, the configurations can be practiced without these specific details. For example, in order to avoid obscuring the configurations, well-known circuits, processes, algorithms, structures, and techniques have been shown without unnecessary details. This specification only provides example configurations without limiting the scope, applicability, or configuration of the claims. On the contrary, the previous description of the configuration will provide an enabling description for implementing the described technology to those skilled in the art. Various changes may be made to the functions and arrangements of the elements without departing from the spirit or scope of the present invention.
[0064] In addition, a configuration may be described as a process, which is depicted as a schematic flow chart or block diagram. Although each configuration may describe an operation as a sequential process, many operations may be performed in parallel or simultaneously. In addition, the order of the operations may be rearranged. A process may have additional steps not included in the figure. In addition, examples of these methods may be implemented by hardware, software, firmware, middleware, microcode, hardware description language, or any combination thereof. When implemented in software, firmware, middleware, or microcode, the program code or code segments for performing the necessary tasks may be stored in a non-temporary computer-readable medium such as a storage medium. The processor may perform the described tasks.
[0065] A number of example configurations have been described, but various modifications, alternative constructions, and equivalents may be used without departing from the spirit of the invention. For example, the above elements may be components of a larger system, where other rules may take precedence over or otherwise modify the application of the technology. In addition, multiple steps may be performed before, during, or after considering the above elements. Therefore, the above description does not restrict the scope of the claims.
[0066] As used herein and in the appended claims, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise. Thus, for example, reference to "a user" includes a plurality of such users and reference to "a processor" includes reference to one or more processors and equivalents thereof known to those skilled in the art, and so forth.
[0067] Furthermore, when “comprises,” “comprising,” or “containing” are used in this specification and the appended claims, they are intended to specify the presence of stated features, integers, components, or steps, but they do not preclude the presence or addition of one or more other features, integers, components, steps, actions, or groups.
Claims
1. A method comprising: capturing radar sensor data from a radar sensor; Determining a signal-to-noise ratio (SNR) of a plurality of points in the radar sensor data; comparing the SNR of each of the plurality of points to a noise threshold; Determining a distribution of SNRs of the plurality of points; as well as Based on the SNR distribution characteristics, each point is classified as a potential obstacle or as ground clutter or noise.
2. The method according to claim 1, further comprising: determining a first set of points or point groups having an SNR above the noise threshold and at least partially surrounded by points or point groups having an SNR below the noise threshold; classifying the first set of points as potential obstacles; Determining a second set of points or groups of points having an SNR below the noise threshold; as well as The second set of points is classified as ground clutter or noise.
3. The method according to claim 1, further comprising: Determine the distance range for each point; as well as For points at longer distance ranges, the noise threshold is adjusted lower.
4. The method according to claim 1, further comprising: Determine the density of points around each point; as well as For each point to be classified as part of a potential obstacle, the density of points surrounding the point is required to exceed a density threshold.
5. The method according to claim 4, further comprising: Determine the distance range for each point; as well as For points at longer distance ranges, the density threshold is adjusted lower.
6. The method according to claim 1, further comprising: The plurality of points are grouped into spatial clusters based on density using a DBSCAN clustering algorithm that classifies points as center points, edge points, or outlier points.
7. The method according to claim 6, wherein: The lower the SNR, the greater the number of surrounding points required to be considered as a center point.
8. The method according to claim 1, further comprising: Use a machine learning algorithm to group points into potential obstacles.
9. The method according to claim 1, wherein: A point is classified as a potential obstacle when it is part of a cluster of points that meets a cumulative point threshold and additional surrounding points of the point have a reduced SNR relative to the cluster of points.
10. The method according to claim 1, further comprising: determining a first set of points or point groups having an SNR above the noise threshold and at least partially surrounded by points or point groups having an SNR below the noise threshold; classifying the first set of points as potential obstacles; Determining a second set of points or groups of points having an SNR below the noise threshold; classifying the second set of points as ground clutter or noise; Determine the distance range for each point; as well as For points at longer distance ranges, the noise threshold is adjusted lower.
11. A method comprising: capturing radar sensor data from a radar sensor; Determining a signal-to-noise ratio (SNR) of a plurality of points in the radar sensor data; comparing the SNR of each of the plurality of points to a noise threshold; Determining a distribution of SNRs of the plurality of points; Based on the SNR distribution characteristics, each point is classified as a potential obstacle or as ground clutter or noise; determining a first set of points or point groups having an SNR above the noise threshold and at least partially surrounded by points or point groups having an SNR below the noise threshold; classifying the first set of points as potential obstacles; Determining a second set of points or groups of points having an SNR below the noise threshold; classifying the second set of points as ground clutter or noise; Determine the distance range for each point; For points with a longer distance range, the noise threshold is adjusted to be lower; Determine the density of points around each point; For a point to be classified as part of a potential obstacle, the density of points around the point is required to exceed a density threshold; Determine the distance range within which each point lies; and For points at longer distance ranges, the density threshold is adjusted lower.
12. The method according to claim 11, further comprising: The plurality of points are grouped into spatial clusters based on density using a DBSCAN clustering algorithm that classifies points as center points, edge points, or outlier points.
13. The method according to claim 12, wherein: The lower the SNR, the greater the number of surrounding points required to be considered as a center point.
14. The method according to claim 11, further comprising: Use a machine learning algorithm to group points into potential obstacles.
15. An apparatus comprising: a radar sensor mounted on a mobile machine; a processor, the processor being installed on the mobile machine; a memory installed on the mobile machine and containing a non-transitory computer-readable medium having instructions for: capturing radar sensor data from a radar sensor; Determining a signal-to-noise ratio (SNR) of a plurality of points in the radar sensor data; comparing the SNR of each of the plurality of points to a noise threshold; Determining a distribution of SNRs of the plurality of points; as well as Based on the SNR distribution characteristics, each point is classified as a potential obstacle or as ground clutter or noise.
16. The device according to claim 15, wherein: The non-transitory computer-readable medium also includes instructions for: determining a first set of points or point groups having an SNR above the noise threshold and at least partially surrounded by points or point groups having an SNR below the noise threshold; classifying the first set of points as potential obstacles; Determining a second set of points or groups of points having an SNR below the noise threshold; as well as The second set of points is classified as ground clutter or noise.
17. The device according to claim 16, wherein: The non-transitory computer-readable medium also includes instructions for: Determine the distance range within which each point lies; and For points at longer distance ranges, the noise threshold is adjusted lower.
18. The device according to claim 15, wherein: The non-transitory computer-readable medium also includes instructions for: A point is classified as a potential obstacle when it is part of a cluster of points that meets a cumulative point threshold and additional surrounding points of the point have a reduced SNR relative to the cluster of points.
19. The device according to claim 18, wherein: The accumulation point threshold decreases as the distance range increases.
20. The device according to claim 15, wherein: The non-transitory computer-readable medium also includes instructions for: determining a first set of points or point groups having an SNR above the noise threshold and at least partially surrounded by points or point groups having an SNR below the noise threshold; classifying the first set of points as potential obstacles; Determining a second set of points or groups of points having an SNR below the noise threshold; classifying the second set of points as ground clutter or noise; Determine the distance range for each point; as well as For points at longer distance ranges, the noise threshold is adjusted lower.
