Target object detection system

Abstract

Apparatuses and related methods relate to a target measurement system (TMS) configured to measure a moving target. In an illustrative example, if a peak is identified in the first few frequency bins after a first FFT is generated, the TMS can add a clutter signal back to the clutter-removed data to measure a slow-moving target. For example, the TMS can calculate cluster regions based on statistical boundaries and statistical centers of a plurality of clusters associated with a target object and merge one or more of the plurality of first clusters having overlapping cluster regions. For example, the TMS can generate N spectral energy heat maps using N independent detection algorithms. For example, values of each spectral energy heat map can be generated based on raw sensor data independent of values of other spectral energy heat maps. For example, when a target is identified in at least two of the N spectral energy heat maps, the TMS can verify the target detection. Various embodiments can advantageously detect target objects with high accuracy.

Description

[0001] Cross-references to related applications This application claims the benefit of U.S. Provisional Application Serial No. 63 / 548,716 entitled “FALSECLUSTERS IDENTIFICATION AND COMBINATION SYSTEM”, filed February 1, 2024, by David S. Anderson et al.

[0002] This application claims the benefit of U.S. Provisional Application Serial No. 63 / 551,380 entitled “STATIC RE-CLUTTERING FOR FMCW RADAR”, filed by David S. Anderson on February 8, 2024.

[0003] This application also claims the benefit of U.S. Provisional Application Serial No. 63 / 549,302 entitled “MULTISTAGE DYNAMIC ANGLE OF ARRIVAL MEASUREMENT SYSTEM”, filed February 2, 2024 by David S. Anderson et al.

[0004] The entire contents of the foregoing application are incorporated herein by reference. Technical Field

[0005] Various embodiments generally relate to FMCW radar for identifying moving targets within the field of view.

[0006] background Radar technology plays a crucial role in a wide range of applications, providing the ability to detect and track objects by emitting electromagnetic waves and analyzing their reflections. Radar systems are used in various fields, including aviation, defense, weather monitoring, and the automotive industry.

[0007] Frequency-modulated continuous wave (FMCW) radar represents a significant advancement in radar technology. FMCW radar operates by continuously modulating the frequency of the transmitted signal over time. This modulation generates a continuous wave with varying frequencies, enabling FMCW radar systems to simultaneously measure the range and velocity of multiple targets. This dual capability makes FMCW radar ideal for applications such as range measurement, velocity detection, and target tracking.

[0008] The use of the Doppler Fast Fourier Transform (FFT) is a key aspect of radar signal processing. Based on the Doppler effect, which describes the frequency shift of a reflected signal from a moving target, the radial velocity of the target can be determined. By employing the FFT, radar systems can efficiently convert time-domain radar signals into a frequency-domain representation, enabling the extraction of valuable information about target velocity and motion. This process improves the precision and accuracy of radar systems in detecting and analyzing moving objects.

[0009] Overview The apparatus and associated methods relate to a multi-pass object detection system (MODS). In an illustrative example, the MODS may include a signal processor configured to receive signals reflected from one or more target objects. For example, the signal processor may generate a multidimensional point cloud. The signal processor may identify N clusters from the multidimensional point cloud. For example, a clustering processing unit may associate a boundary for each of the N clusters. In one embodiment, the boundary may be determined as a predetermined multiple of the standard deviation of the points of the corresponding cluster. For example, based on the boundary, if overlapping clusters are associated with an overlapping space contained within the boundary, a cluster merging module may generate a detection result by merging the clusters in the N clusters. Various embodiments can reduce the number of erroneously identified objects due to over-clustering of individual objects.

[0010] The apparatus and associated methods relate to a Dynamic Target Measurement System (DTMS) configured for measuring moving targets. In an illustrative example, the DTMS includes a processor configured to automatically measure precise information about a slow-moving object within a static object. Operation includes, for example, receiving measurement signals from orientation and ranging measurement devices. A first component of the measurement signal (e.g., DC offset) can be extracted from the measurement signal. For example, a first FFT of the clutter-removed signal can be generated. For example, peaks within the first few frequency bins of the first FFT can be identified. If any bins are identified within the first few frequency bins, the first component can be added back to the clutter-removed signal, for example, to measure the slow-moving target (SMT). Various embodiments can advantageously detect SMTs with high accuracy.

[0011] The apparatus and associated methods relate to an object detection system (ODS) that uses a multi-stage approach to calculate the angle of arrival (Angle of Arrival) of a detected object (e.g., within a specific range). In an illustrative example, the ODS may generate a first spectral energy representation (SER) of the object detection signal based on a first computational method (FCM). For example, an amplitude threshold may be applied to the first SER to dynamically determine a region of interest (ROI) within the first SER. The ODS may generate, for example, a second SER based on the signal within the ROI using a second computational method (SCM). For example, SCM may require more computational cost than FCM to produce higher resolution results. Various embodiments can advantageously generate object detection measurements with a lower false object detection rate than FCM, and require less computational cost compared to applying SCM entirely to the object detection signal.

[0012] Various embodiments can achieve one or more advantages. For example, some embodiments can advantageously predict potential collisions. For example, some embodiments can improve the processing speed for measuring high-speed targets. For example, some embodiments can advantageously improve the linearity and accuracy of velocity measurements to obtain precise measurement results in a Doppler FFT chamber. Various embodiments can achieve one or more advantages. For example, some embodiments can remove erroneous object detection caused by the inherent error characteristics of single object detection methods. For example, some embodiments can advantageously mitigate object detection errors at predetermined angles. For example, some embodiments can advantageously provide a rapid response for applications requiring rapid decision-making based on object detection measurement results. For example, some embodiments can advantageously reduce the minimum detectable interval for multiple objects.

[0013] Details of various embodiments are set forth in the accompanying drawings and the description below. Other features and advantages will become apparent from the description, the drawings, and the claims. Brief description of the attached diagram Figure 1 An exemplary multipass object detection sensor (MODS) is depicted in an illustrative use case scenario.

[0015] Figure 2A and Figure 2B This is a schematic diagram illustrating an example of exemplary clustering outputs in the various steps of MODS.

[0016] Figure 3 Another example of exemplary clustering outputs from the various steps of MODS is shown.

[0017] Figure 4 This is a flowchart illustrating an exemplary multi-pass object detection method.

[0018] Figure 5 This is a flowchart illustrating an exemplary MODS configuration method.

[0019] Figure 6A and Figure 6B An exemplary static and dynamic matching object detection system (SDMODS) is described.

[0020] Figure 7 An exemplary Dynamic Target Measurement System (DTMS) is described in an illustrative use case scenario.

[0021] Figure 8 An exemplary re-cluttered Doppler FFT generation process using an exemplary slow target identifier (STI) is described.

[0022] Figure 9A and Figure 9B Exemplary error curves for DTDU before and after re-addition of clutter are depicted.

[0023] Figure 10 This is a flowchart illustrating an exemplary moving object detection method.

[0024] Figure 11 An exemplary angle of arrival module (AOAM) is depicted in an illustrative use case scenario.

[0025] Figure 12A and Figure 12B An exemplary heatmap of a single target is depicted, wherein a region of interest is dynamically selected in a first heatmap to generate a second heatmap.

[0026] Figure 12C and Figure 12D An exemplary heatmap depicts multiple targets within a dynamically selected region of interest.

[0027] Figure 13 This is a flowchart illustrating an exemplary dynamic target detection method.

[0028] Figure 14 This is a flowchart illustrating an exemplary AOAM configuration method.

[0029] Figure 15 This is a block diagram depicting an exemplary target object detection system (TODS 1500).

[0030] Similar reference symbols in the various figures indicate similar elements.

[0031] Detailed Description of Illustrative Embodiments Throughout this specification, unless otherwise stated, the symbol “_” (underscore) indicates a subscript. Similarly, unless otherwise stated, the symbol “^” (caret) indicates a superscript. This consistent convention promotes clarity and helps in the accurate interpretation of the mathematical and chemical formulas presented herein.

[0032] Figure 1 An exemplary multipass object detection sensor (MODS 100) employed in an illustrative use case scenario is depicted. For example, MODS 100 may include a system for detecting objects by transmitting and receiving signals (e.g., electromagnetic signals, radio signals, sound waves, optical signals, static / changing waves between 0.000001 Hz and 10000 GHz). For example, MODS 100 may include a time-of-flight (ToF) sensor. For example, MODS 100 may include a radar system. For example, MODS 100 may include a sonar system.

[0033] In this example, MODS 100 includes a transmitter 105 and a receiver 110. In some embodiments, the transmitter 105 may transmit a transmitted signal 115. For example, the transmitter 105 may include one or more transmitting elements. For example, each transmitting element may be configured to generate a signal independently. For example, the receiver 110 may include one or more receiving elements. For example, each receiving element may receive a signal independently.

[0034] For example, the transmitted signal 115 may be reflected at the target object 120. For example, the receiver 110 may receive the reflected signal 125 from the target object 120. Based on the reflected signal 125 and / or the transmitted signal 115, the MODS 100 may determine the presence or absence of an object within its field of view (e.g., target object 120 or its absence). For example, if an object is present, the MODS 100 may determine the position, velocity, and / or other information of the target object 120.

[0035] As shown in the figure, MODS 100 includes a multipass clustering unit (MCU 130) and a signal processing and clustering module (SPACM 135). For example, SPACM 135 can receive signals (e.g., reflected signal 125) from receiver 110. In some embodiments, SPACM 135 can process the received signals. For example, SPACM 135 can identify noise from reflected signal 125. For example, SPACM 135 can also receive signals from transmitter 105. For example, SPACM 135 can receive the raw waveform of transmitted signal 115.

[0036] In some implementations, the SPACM 135 can generate a point cloud based on the reflected signal 125. For example, the point cloud may include a multi-dimensional point cloud. For example, from the point cloud, the SPACM 135 can identify N clusters (C_1, C_2, ..., C_i, ..., C_N), where N is an integer >= 0. For example, the SPACM 135 may include a noise removal algorithm to advantageously remove noisy data before clustering. For example, each i-th cluster in the N clusters includes M_i points (C^1_i, C^2_i, ..., C^j_i, ..., C^M_i).

[0037] In some embodiments, MODS 100 may be a frequency modulated continuous wave (FMCW) radar. For example, MODS 100 may be configured to detect static (e.g., stationary) targets. For example, MODS 100 may be configured to detect moving (e.g., dynamic) targets. In some embodiments, SPACM 135 may use the range and angular position of target object 120 (e.g., using Fast Fourier Transform analysis and / or other signal processing techniques) to identify static objects. In some embodiments, in addition to the range and angular position of target object 120, SPACM 135 may also use target discrimination (e.g., velocity) information determined from reflected signal 125 to identify dynamic objects.

[0038] In some implementations, SPACM 135 can determine the presence of target object 120 based on signals received from transmitter 105 and receiver 110. For example, SPACM 135 can apply clustering algorithms (e.g., K-means clustering, hierarchical clustering, mean-shift clustering, density-based noisy applied spatial clustering (DSCAN) algorithm). For example, SPACM 135 can identify one or more clusters from reflected signal 125. For example, MCU 130 can determine the number of objects within the field of view of transmitter 105 based on the number of clusters identified.

[0039] In some implementations, SPACM 135 may include a distance parameter (e.g., ε in the case of DBSCAN). For example, the distance parameter may be related to the sensitivity of SPACM 135. For instance, SPACM 135 may use the distance parameter to determine whether a given point (e.g., a point retrieved from the signal of receiver 110) is assigned to a cluster. For example, a larger distance parameter may reduce the number of clusters identified from reflected signal 125. In some examples, a smaller distance parameter may increase the number of clusters identified from reflected signal 125.

[0040] For example, MODS 100 can be used to detect multiple objects within the field of view of transmitter 105. In some embodiments, target object 120 may comprise multiple surfaces. For example, multiple surfaces may generate multiple separate reflected signals. In some examples, multiple separate returned signals can be interpreted as different objects within the field of view.

[0041] In this example, MCU 130 includes a dynamic clustering boundary recognizer (DCBI 140) and a clustering merging module (CBM 145). For example, DCBI 140 and CBM 145 can advantageously distinguish between a single complex object (e.g., with multiple surfaces) and multiple objects within the field of view.

[0042] In some implementations, SPACM 135 can use the received signal and a (relative) distance parameter to generate initial clustering results. For example, the distance parameter can be chosen to be small. For example, a small distance parameter may generate false detections of objects. For example, SPACM 135 can generate multiple clusters that may be the same object.

[0043] In some implementations, DCBI 140 can dynamically generate boundaries for each cluster identified in the initial clustering results. For example, DCBI can determine the boundary of the k-th cluster based on each of the M dynamic points (C^1_k, C^2_k, ..., C^j_k, ..., C^M_k) of the k-th cluster (e.g., within the N identified clusters as discussed above).

[0044] In some implementations, DCBI 140 can generate a surface function (e.g., a multinomial function) for each cluster in the initial clustering results. For example, MCU 130 can use the surface function to generate the boundary of each cluster. In some implementations, DCBI 140 can compute a statistical measure of the density between data points within a cluster. For example, the statistical measure can include a multidimensional statistical measure of the data points. For example, DCBI 140 can define the boundary of any cluster as the change in two-dimensional standard deviation from the cluster center multiplied by a predetermined threshold (e.g., 1.5, 2.5, 4.5, 8.5).

[0045] In some implementations, CBM 145 can use surface functions to determine whether to merge any two or more clusters from the initial clustering results. For example, CBM 145 can check for any overlapping boundaries. For example, any clusters with overlapping boundaries can be merged and / or grouped. For example, CBM 145 can advantageously identify and reduce false detections of objects (e.g., caused by over-clustering of a single object).

[0046] As shown in the figure, CBM 145 can generate object detection results for object generation unit 150. For example, object generation unit 150 can be operatively connected to another device. For example, the object detection results can include information about the objects detected after merging N clusters. For example, the object detection results can include the position (e.g., distance) information of each detected object. For example, the object detection results can include the angle information of each detected object (e.g., angle information within the field of view of MODS 100). For example, the object detection results can include the velocity information of each detected object. For example, the object detection results can include other information about each detected object (e.g., historical information, intensity information, color information).

[0047] In some implementations, MODS 100 can be connected to a device for determining the angular position of a target object 120. For example, the connected device can control movement based on object detection results. For example, the connected device can be a robot. For example, the connected device can include an autonomous vehicle. For example, the connected device can include a vehicle guidance system. For example, the connected device can include a loading / unloading platform. For example, the loading / unloading platform can control and / or influence the movement of approaching vehicles. For example, the loading / unloading platform can use MODS 100 to determine the position and / or velocity of surrounding objects and generate instructions to approaching vehicles.

[0048] In various implementations, using predetermined distance parameters, MODS 100 can dynamically adjust the boundary of each detected object based on the received signal. For example, MODS 100 can advantageously identify multiple objects within the field of view. In some implementations, MCU 130 can generate a statistical distribution of each identified cluster from the initial clustering stage (e.g., generated by SPACM 135) to determine whether a cluster can be merged with neighboring clusters.

[0049] In some implementations, the DCBI 140 can generate the statistical distribution of clusters based on data points within clusters. For example, the DCBI 140 can advantageously save computational resources compared to determining the statistical distribution of the entire dataset to generate cluster boundaries. In various examples, the MODS 100 can be advantageously used in real-time applications.

[0050] In various implementations, a multi-pass clustering method (e.g., MODS 100) can generate multiple first clusters using a first, relatively small distance parameter (e.g., generated by SPACM 135). For each of the multiple first clusters, a clustering region is calculated based on the statistical boundaries and statistical centers of each of the multiple first clusters (e.g., via DCBI 140), and clusters with one or more overlapping clustering regions in the multiple first clusters are merged (e.g., performed by CBM 145). For example, if both static and dynamic objects are detected within a matching distance threshold (e.g., 0.25 m), the dynamic object can be selected and the static object discarded. For example, repeated detection of slowly moving objects can be reduced. For example, the matching distance can be generated based on the statistical distribution of the dynamic object cluster of the dynamic object.

[0051] Figure 2A and Figure 2B This is a schematic diagram illustrating an example of the clustering output in the various steps of MODS. For example... Figure 2A As shown, point cloud 200 can be generated, for example, by SPACM 135. For instance, based on a signal received from receiver 110, SPACM 135 can generate point cloud 200 using the DBSCAN algorithm. For example, SPACM 135 can apply the DBSCAN algorithm using relatively small distance parameters. In some implementations, the small distance parameters can be determined based on specific targets in the application of MODS 100.

[0052] For example, relatively small distance parameters can include distance parameters smaller than the expected gap between points received from a single object containing multiple reflective surfaces. For example, the expected gap can be predetermined through an experimental procedure. For example, the experimental procedure can be coordinated based on an application of MODS 100. For example, a dynamic object recognition method can be used to generate point cloud 200.

[0053] As shown in the figure, point cloud 200 includes three identified clusters 205A-205C of dynamic points. Point cloud 200 also includes noise 210. In this example, each of the identified clusters 205A-205C may include centers 215A-215C. In some implementations, centers 215A-215C may be determined based on the (multidimensional) average of the dynamic points within each cluster of clusters 205A-205C. In some examples, centers 215A-215C may be determined as the median point of the dynamic points within each of clusters 205A-205C. In some examples, centers 215A-215C may be determined as the weighted average of the dynamic points within each cluster 205A-205C.

[0054] like Figure 2BAs shown, DCBI 140 can generate boundaries 220A-220C for each cluster in clusters 205A-205C. For example, boundaries 220A-220C can be determined as distances from centers 215A-215C, where each distance is a multiple of the standard deviation of the dynamic points in each cluster 205A-205C.

[0055] As an illustrative example rather than a limitation, assume that cluster 205A comprises N dynamic points (p_1, p_2, ..., p_N), and center 215A is point c. For example, the distance between boundary 220A and center 215A can be given by the following formula: sigma = sqrt(sum((p_i-c)^2) / N) For example, a point x within point cloud 200 is within boundary 220A when the following condition is met: f(x)<|xc|-K sigma=0, where K is a predetermined multiple (e.g., 4.5).

[0056] Based on boundaries 220A-220C, CBM 145 can determine the overlapping boundaries of clusters 205A-205C. For example, a boundary associated with a function f(x) can be considered to have a space defined by f(x) < 0. In this example, the boundaries of clusters 205A-205C all overlap. For example, if the boundaries of 205A-205C are represented by f_A(x) = 0, f_B(x) = 0, and f_C(x) = 0, then there may exist spaces / points x_k where f_A(x_k) < 0, f_B(x_k) < 0, and f_C(x_k) < 0. In some implementations, CBM 145 can merge clusters 205A-205C. For example, MCU 130 can recognize clusters 205A-205C as a single object.

[0057] Figure 3 MODS (e.g.) are shown Figure 1Another example of exemplary clustering output in the various steps of MODS 100. In this example, SPACM 135 can determine two initial clusters 305A-305B in point cloud 300. For example, SPACM 135 can identify the dynamic points and noise 310 of the initial clusters 305A-305B. Next, DCBI 140 can generate boundaries 315A-315B using the dynamic points of the initial clusters 305A-305B respectively. For example, DCBI 140 can determine the centers 320A-320B of the initial clusters 305A-305B and the statistical distribution of the dynamic points. In some examples, DCBI 140 can generate boundaries 315A-315B based on the standard deviation of the initial clusters 305A-305B. In this example, because boundaries 315A and 315B do not overlap, in some embodiments, CBM 145 can determine that clusters 305A-305B are distinct objects.

[0058] Figure 4 This is a flowchart illustrating an exemplary multi-pass object detection method. For example, MCU 130 may execute method 400 to determine the number of objects detected within the field of view of transmitter 105. In this example, method 400 begins when a reflected signal of the field of view is received in step 405. For example, the reflected signal may include a signal reflected from a dynamic object. For example, the reflected signal may be received from a complex surface object.

[0059] Next, in step 410, a point cloud is generated based on the reflection signal. For example, SPACM 135 can generate point cloud 200 using distance data, velocity data, and angle data determined based on the reflection signal 125 from the target object 120. In step 415, initial clusters are determined in the point cloud using predetermined distance parameters. For example, SPACM 135 can use the DBSCAN algorithm with relatively small distance parameters to determine initial clusters 205A-205C.

[0060] After identifying the initial clusters, at decision point 420, it is determined whether more than one initial cluster has been identified. If only one initial cluster is identified, method 400 ends. If more than one initial cluster is identified, then in step 425, a boundary is determined for each initial cluster. For example, DCBI 140 can determine boundary 220A based on the location of the center 215A of cluster 205A and the standard deviation of the locations of dynamic points in cluster 205A.

[0061] At decision point 430, it is determined whether any two or more boundaries overlap. For example, boundaries 220A-220C overlap. If no boundaries overlap, method 400 ends. In step 435, if any overlapping boundaries exist, the initial clusters with overlapping boundaries are merged, and method 400 ends. For example, CBM 145 may merge clusters 205A-205C because they all have overlapping boundaries. For example, CBM 145 may not merge the initial clusters 305A-305B because their boundaries do not overlap.

[0062] Figure 5 This is a flowchart illustrating an exemplary MODS configuration method 500. For example, method 500 may be performed during the calibration step of MODS 100. In some examples, the exemplary MODS configuration method 500 may be performed in a factory during the manufacturing of MODS 100.

[0063] In this example, method 500 begins in step 505 when distance parameters for initial clustering are received. For example, SPACM 135 can be configured to perform clustering operations based on distance parameters (e.g., DBSCAN). Next, boundary multipliers are received in step 510. For example, boundary multipliers can be constant multipliers. In some implementations, boundary multipliers can include a function definition. For example, boundary multipliers can include one or more parameters of a polynomial function. For example, a polynomial function can be applied to one or more statistical distributions of the clusters to compute the boundaries. In step 515, the distance parameters and boundary multipliers are stored in a data repository, and method 500 ends. For example, in operation, SPACM 135 can retrieve the distance parameters and boundary multipliers from the data repository.

[0064] Figure 6A and Figure 6B An exemplary static and dynamic matching object detection system (SDMODS) is described. For example... Figure 6A As shown, the SDMODS 600 includes an FMCW radar 605 and an MCU 130. For example, the FMCW radar 605 can independently identify static (e.g., stationary and / or very slow-moving) objects and moving targets based on reflected signals 125. For example, a slow-moving object could be an object with a speed of less than 0.1 m / s. In some embodiments, the FMCW radar 605 can generate noise-removed signals for the MCU 130. For example, the FMCW radar 605 may include analog-to-digital circuitry. For example, the FMCW radar 605 may include digital signal processing circuitry. For example, the FMCW radar 605 may include analog signal processing circuitry.

[0065] In this example, MCU 130 detects dynamic object 610 and static object 615. For example, dynamic object 610 and static object 615 may include the velocity and / or position (e.g., within the field of view) of each identified object.

[0066] SDMODS 600 includes a Static Object Suppression Engine (STSE 620). In some implementations, STSE 620 can identify duplicate objects detected in a list of identified dynamic objects 610 and a list of identified static objects 615. For example, STSE 620 can compare static objects 615 and dynamic objects 610. For example, STSE 620 can determine overlap between tracked static and dynamic objects. For example, STSE 620 can select some or all objects from the list of identified dynamic objects 610 and the list of identified static objects 615 to generate object detection result 625.

[0067] In some implementations, if an object is identified and is within a predetermined proximity (e.g., within 0.25m, 0.3m, 0.1m) of both the list of identified dynamic objects 610 and the list of identified static objects 615, the STSE 620 can determine that the objects are considered the same. For example, if an object in the list of identified dynamic objects 610 and an object in the list of identified static objects 615 are considered the same, the STSE 620 can select the object in the list of identified dynamic objects 610 and remove it from the list of identified static objects 615. In some implementations, the predetermined proximity can be extended to two-dimensional or three-dimensional space in a multidimensional measurement system. Thus, the STSE 620 can advantageously present a list of all detected objects within the measurement range without duplication for slowly moving objects.

[0068] In various embodiments, the STSE 620 can use statistical information to determine static and dynamic matches in object detection. For example, the SDMODS 600 can collect statistical information determined in clustering algorithms (e.g., using reference data). Figures 1-4 Various methods described are used to dynamically adjust the predetermined proximity for static target suppression.

[0069] As an illustrative example, Figure 6BAn exemplary point cloud 630 with dynamic point clustering is shown. In some embodiments, cluster 635 may be processed by MCU 130 to generate boundary 640. For example, boundary 640 (e.g., shown as a circle in this example) may indicate a multiple of the standard deviation of the corresponding identified cluster (e.g., cluster 635). In some embodiments, MCU 130 may use boundary 640 to compute a probability function. For example, the probability function may provide the probability that a given static point in the exemplary point cloud 630 may belong to an object represented by any of clusters 635. For example, the probability function of the k-th cluster may be determined as a function of the M dynamic points of the k-th cluster.

[0070] For example, STSE 620 can identify duplicate objects from a list of identified dynamic objects 610 and a list of identified static objects 615 based on a probability function. Therefore, for example, MCU 130 can advantageously enhance MODS 100 to identify static objects near tightly clustered dynamic objects by dynamically adjusting the probability function based on the identified dynamic clusters, while reducing false multiple detections in the case of loosely grouped dynamic clusters.

[0071] Figure 7 An exemplary Dynamic Target Measurement System (DTMS) employed in an illustrative use case scenario is depicted. In this example, a control assistance system 700 is embedded in a vehicle 705. For example, the control assistance system 700 can be used to accurately measure the distance to objects within a predetermined proximity range of the vehicle 705. For example, the control assistance system 700 can detect another vehicle (e.g., vehicle 710A in this example). In some examples, the control assistance system 700 can be used to detect nearby vehicles, pedestrians 710B, and / or obstacles 710C around the vehicle 705. As illustrative examples and not limitations, various embodiments can advantageously determine precise distance information of detected targets (e.g., objects) to assist collision avoidance and / or adaptive cruise control systems.

[0072] As shown in the figure, vehicle 705 includes a dynamic object orientation and ranging device (DODAR device 715). For example, DODAR device 715 can identify moving objects (targets) near vehicle 705. For example, in addition to the distance and angular position of the detected object, DODAR device 715 can also simultaneously measure the velocity of the detected object. In some examples, velocity information can be advantageously used to predict potential collisions.

[0073] In this example, the DODAR device 715 includes an FMCW transmitting and receiving element (FTRE 720). For example, the FTRE 720 can be configured to transmit a frequency-modulated signal (e.g., a Doppler chirped signal). For example, the DODAR device 715 can generate velocity information of a moving target based on the Doppler effect measured according to the reflected signal received from the moving target.

[0074] In various examples, the FTRE 720 can be configured to transmit various types of signals. For example, the FTRE 720 can be configured to transmit radio signals. For example, the FTRE 720 can be configured to transmit optical signals. For example, the FTRE 720 can be configured to transmit sound waves. For example, the FTRE 720 can be configured to transmit electromagnetic waves.

[0075] DODAR device 715 includes a DC offset extractor 725 and a Doppler FFT processor 730. For example, the Doppler FFT processor 730 can use a processed (e.g., background-removed) Doppler chirped signal from the DC offset extractor 725 to generate a clutter-removed Doppler FFT 735. For example, the DC offset extractor 725 can receive a signal from an FTRE 720. In some embodiments, the DC offset extractor 725 can process the received signal from the FTRE 720 for use by the Doppler FFT processor 730 to generate a clutter-removed Doppler FFT 735 of the processed signal.

[0076] In various implementations, the DODAR device 715 can be configured to identify moving (dynamic) targets in the presence of many stationary (static) objects within the field of view of the FTRE 720. For example, the DODAR device 715 can use a DC offset extractor 725 to remove DC offsets from each Doppler chirped signal received from the FTRE 720. For example, the DC offset extractor 725 can remove DC offsets from the Doppler chirped signals before performing a generated Doppler FFT. In some examples, this process can be referred to as "static clutter removal." In some implementations, the DC offset extractor 725 can remove DC offsets to advantageously improve processing speed. For example, without DC offsets, the Doppler FFT processor 730 can process Doppler chirped signals even when there are no large numbers of static objects within the field of view. For example, the DC offset extractor 725 can remove DC offsets to prevent the detection threshold (e.g., if a dynamic threshold calculation method such as constant false alarm rate (CFAR)) from increasing due to the presence of static objects. In various examples, a lower detection threshold can advantageously enhance the detection of smaller dynamic objects.

[0077] DODAR device 715 includes object detection unit 740 to process clutter-removed Doppler FFT 735 from Doppler FFT processor 730. In some embodiments, object detection unit 740 may identify one or more peaks in the clutter-removed Doppler FFT 735. Once a peak in the clutter-removed Doppler FFT 735 has been identified, object detection unit 740 may generate a velocity measurement of the peak by applying interpolation techniques to the clutter-removed Doppler FFT 735. In some embodiments, object detection unit 740 may apply one or more interpolation techniques (e.g., polynomial interpolation, least mean square, radial basis function interpolation, spline interpolation) to advantageously improve the linearity and precision of the velocity measurement to obtain accurate measurements in the Doppler FFT chamber. (Refer to...) Figure 8 Some examples of clutter-removed Doppler FFT 735 are described.

[0078] In some implementations, the object detection unit 740 may be configured to detect the distance, orientation, and / or velocity of a moving object. For example, peaks in the clutter-removed Doppler FFT 735 may indicate the orientation of the moving object (e.g., angle of arrival). For example, peaks may indicate the distance of the moving object (e.g., the position of a reference DODAR device 715). For example, peaks may indicate the velocity of the moving object. In some examples, the Doppler FFT processor 730 may generate multiple clutter-removed Doppler FFTs 735 to the object detection unit 740 to determine various measurements of the moving object.

[0079] In some examples, object detection unit 740 may be configured to detect static and / or dynamic objects. For example, object detection unit 740 may include... Figure 6A The SDMODS 600 described herein is used to identify objects. For example, the object detection unit 740 can perform the following as described in the reference. Figures 1 to 4 The described multi-pass object detection method.

[0080] As shown in the figure, the DODAR device 715 includes a memory module 750 and a slow target identifier (STI 755). In this example, both the DC offset extractor 725 and the STI 755 are operatively connected to the memory module 750. For example, the DC offset extractor 725 can extract clutter data 760 from the Doppler chirp signal for storage in the memory module 750.

[0081] In some implementations, the STI 755 can process clutter-removed Doppler FFTs 735 generated by the Doppler FFT processor 730. For example, the STI 755 can process clutter-removed Doppler FFTs 735 in the low-frequency Doppler FFT bins (e.g., the first few Doppler bins). For example, the STI 755 can use peak detection techniques to process the clutter-removed Doppler FFTs 735. In some examples, the STI 755 can use dynamic thresholding techniques to identify potential peaks in the clutter-removed Doppler FFTs 735 (e.g., using a (dynamically determined) constant false alarm rate).

[0082] In some implementations, if the STI 755 identifies any peaks in the first few Doppler cells, the STI 755 can retrieve clutter data 760 from the memory module 750. For example, the Doppler FFT processor 730 can receive the clutter data 760. For example, the Doppler FFT processor 730 can add the clutter data 760 back into the processed Doppler chirped signal. For example, without removing the DC offset, the Doppler FFT processor 730 can use the Doppler chirped signal with the clutter data 760 added to recalculate the clutter-readded Doppler FFT 765.

[0083] For example, object detection unit 740 can identify new peaks from the re-added clutter Doppler FFT 765. In some implementations, object detection unit 740 can use the new peaks to perform interpolation on velocity calculations.

[0084] In some implementations, the object detection unit 740 may generate object orientation and distance measurement results at the communication interface 770. For example, the communication interface 770 may send signals to the control system of the vehicle 705. As an illustrative example and not a limitation, the control system may be embedded in the loading platform of the vehicle 705. For example, the loading platform may measure the slow approach of the vehicle 705 and generate guidance to avoid collisions with the loading platform or surrounding static or moving obstacles.

[0085] Figure 8 An exemplary re-clutter Doppler FFT generation process using an exemplary Slow Target Identifier (STI) is depicted. In this example, the re-clutter Doppler FFT generation process 800 can be performed by an STI 755, as described in reference [reference missing]. Figure 7As described, the STI 755 can receive clutter-removed Doppler FFT data 805. In this example, the STI 755 includes a peak detection module 810 and a CFAR generator 815. For example, the clutter-removed Doppler FFT data 805 can be configured to identify peaks in the clutter-removed Doppler FFT data 805 and the bin number and frequency of those peaks in the clutter-removed Doppler FFT data 805. For example, the peak detection module 810 can be configured to dynamically generate an amplitude threshold 820 in the clutter-removed Doppler FFT data 805. In some implementations, the STI 755 can use the amplitude threshold 820 to determine whether to generate a clutter-re-added Doppler FFT 765.

[0086] As an illustrative example and not a limitation, based on the clutter-removed Doppler FFT data 805, the peak detection module 810 can determine the clutter-removed peak value 825 (e.g., at frequency = 0.981 in bin 1). For example, the peak detection module 810 may include a predetermined bin threshold 830. For example, the peak detection module 810 can determine that a re-cluttered Doppler FFT 765 will be generated when the amplitude threshold 820 is less than or equal to the predetermined bin threshold 830. In some embodiments, the predetermined bin threshold 830 may be a fixed number (e.g., 2, 3, 5). In some embodiments, the predetermined bin threshold 830 may be determined dynamically. For example, the predetermined bin threshold 830 may be determined based on the CFAR generated by the CFAR generator 815.

[0087] In some implementations, the size of the Doppler FFT chamber can be determined based on the number of measurements in the Doppler chirped signal and the timing between the measurements. For example, the object detection unit 740 can apply various interpolation techniques to determine its size based on the re-added DC offset.

[0088] In this example, the peak detection module 810 also includes an amplitude threshold 820. For example, if any bin less than a predetermined bin threshold 830 includes a modulus value (e.g., energy) higher than the amplitude threshold 820, the peak detection module 810 can determine to generate a Doppler FFT 765 with added clutter. For instance, the CFAR generator 815 can generate the amplitude threshold 820 based on the signal and noise received from the FTRE 720.

[0089] As shown in the figure, distortion has been removed. Figure 9B The expected linearity of the re-addition clutter method is shown in the figure. For example, the linearity can be recovered to the original linearity measurement without static clutter removal.

[0090] In this illustrative example, the STI 755 can determine that the Doppler FFT data 840 with re-added clutter can be computed. For example, the STI 755 can determine that the Doppler FFT data 840 with re-added clutter is generated because the location of the de-choked peak 825 is in a bin that is less than a predetermined bin threshold 830. For example, the STI 755 can determine that the Doppler FFT data 840 with re-added clutter is generated because, within the predetermined bin threshold 830, there is a magnitude value in the de-choked Doppler FFT data 805 that exceeds the amplitude threshold 820.

[0091] In this example, the object detection unit 740 can identify a re-added clutter peak 845 that is different from the clutter-removed peak 825 (e.g., at frequency = 0.338 at cell 0). For example, due to errors introduced by removing background noise from the data received from FTRE 720, the re-added clutter peak 845 can be represented as point 850 in the clutter-removed Doppler FFT data 805, with a magnitude smaller than the clutter-removed peak 825. Therefore, for example, STI 755 can advantageously reduce errors and enhance the measurement of the position and velocity of slow-moving objects.

[0092] Figure 9A and Figure 9B Exemplary error curves for DTMS before and after re-addition of clutter are depicted. Figure 9A As shown, the error curve 900 of the Doppler FFT with clutter-removed data (e.g., after the signal has been processed by the DC offset extractor 725) exhibits a relatively high error rate at a low input frequency 905. For example, error curve 900 can be generated to measure interpolation applications (e.g., by a reference). Figure 7 The object detection unit 740 described applies interpolation to the linearity of the Doppler FFT. For example, a low input frequency 905 may correspond to a lower bin number. For example, removing the DC component of the received signal may introduce significant distortion in the clutter-removed Doppler FFT 735. For example, distortion can be generated by removing the signal from the 0-velocity FFT bin. For example, clutter removal may cause distortion in the measured linearity when any slow-moving object is present in the field of view. In some examples, the effect on velocity linearity is mitigated for targets with sufficient velocity because the 0-Doppler FFT bin is not near the interpolation peak representing the moving velocity.

[0093] Figure 9BAn exemplary error curve 910 of a Doppler FFT with re-added clutter data is shown. For example, the clutter data can be re-added by an STI 755 using clutter data 760 stored in a memory module 750. As shown, the linearity of the error curve is exhibited as a sine wave varying with frequency. For example, the sine wave could be the target response of the error after interpolation is applied to the Doppler FFT. In this example, the error oscillates between + / - 0.06. For example, at a low input frequency 905 (e.g., >0.2), the error shown in exemplary error curve 910 is smaller than the error shown in error curve 900. For example, in the re-added clutter Doppler FFT data 840, the distortion indicated by point 850 is removed. For example, as... Figure 9B As shown, the linearity can be restored to the original linearity measurement without static clutter removal.

[0094] Figure 10 This is a flowchart illustrating an exemplary moving object detection method. For example, method 1000 can be derived from, as shown in reference... Figure 7 The described DODAR device 715 is used to perform this. In this example, method 1000 begins when a measurement signal is received from the direction and range measuring device in step 1005. For example, the measurement signal may include Doppler measurement results. For example, the measurement signal may be received from an FMCW radar. For example, the radar may be an FMCW radar. In some examples, Doppler measurement results may be received from other direction and range measuring devices that transmit the signal. In step 1010, a first component of the measurement signal may be generated. For example, the first component may include a DC offset (e.g., static clutter) of the Doppler measurement results. For example, the DC offset may be extracted from the Doppler measurement results. For example, a DC offset extractor 725 may identify background signals from the Doppler measurement results.

[0095] Next, in step 1015, the first component is saved to a data repository. For example, clutter data 760 is saved to memory module 750. After saving the first component, in step 1020, a second component of the measurement signal is generated by removing the first component from the measurement signal. For example, the second component may include a measurement signal with static clutter removed. For example, DC offset extractor 725 can remove static clutter from the signal received by FTRE 720. In step 1025, a first FFT representation (e.g., a Doppler FFT) is generated from the second component of the measurement signal (e.g., a Doppler measurement result with static clutter removed). For example, Doppler FFT processor 730 can use a signal with DC removed to generate a Doppler FFT.

[0096] Next, a bin threshold (B) is determined in step 1030. For example, a predetermined bin threshold 830 can be retrieved from memory module 750. In some embodiments, the bin threshold can be determined by CFAR generator 815 based on Doppler measurements.

[0097] In step 1035, a modulus threshold (M) is determined. For example, the STI 755 may use a CFAR generator 815 to dynamically determine the modulus threshold. For instance, the modulus threshold may be a fixed number stored in the memory module 750.

[0098] At decision point 1040, it is determined whether there are any bins in the generated Doppler FFT that are smaller than B and have a modulus greater than M. If there are no bins smaller than B and with a modulus greater than M, then in step 1045, interpolation is performed on the first FFT representation to identify fast dynamic targets, and method 1000 ends.

[0099] If any bins smaller than B have a magnitude greater than M (e.g., if any peak is identified in bins smaller than B), a first component of the measurement signal is retrieved from the data repository in step 1050. For example, clutter data 760 can be retrieved from memory module 750. Next, an aggregated measurement signal is generated by adding the first and second components of the measurement signal. For example, Doppler FFT processor 730 can recover the Doppler measurement result by adding static clutter back to the clutter-removed Doppler measurement in step 1055. In step 1060, a second FFT representation (e.g., Doppler FFT) is generated from the aggregated measurement signal. For example, the recovered Doppler measurement result can be used to generate a re-cluttered Doppler FFT 765. In step 1065, interpolation is performed on the second FFT representation to identify slow-moving targets, and method 1000 ends. For example, object detection unit 740 can use the re-cluttered Doppler FFT 765 to identify slowly approaching vehicles.

[0100] Figure 11An exemplary angle-of-arrival (AOA) measurement unit employed in an illustrative use case scenario is depicted. In this example, the Target Precision Measurement System (TPMS 1100) includes a wave direction and distance measurement device (WADAR 1105) and a precision measurement unit (PMU 1110). For example, TPMS 1100 may include an object detection system. For example, WADAR 1105 may include a phased array antenna system and a ranging device. For example, TPMS 1100 can use WADAR 1105 to generate an AOA measurement of a detected object. As shown, WADAR 1105 can detect target object 1115. For example, TPMS 1100 may include a time-of-flight (ToF) system. For example, WADAR 1105 may include a frequency-modulated continuous wave (FMCW) radar. For example, the ToF system can be configured to measure the angle and distance of target object 1115 relative to WADAR 1105.

[0101] In some implementations, PMU 1110 can receive measurement signals from WADAR 1105. For example, WADAR 1105 can transmit measurement signals to PMU 1110 via a communication cable. In some examples, WADAR 1105 and PMU 1110 can be wirelessly connected. For example, WADAR 1105 can transmit measurement signals to PMU 1110 via a wireless network. For example, the measurement signal may include whether a target object 1115 is within the field of view (FOV) of WADAR 1105. For example, the measurement signal may include the distance (e.g., ranging) of the target object 1115 from WADAR 1105. For example, the measurement signal may include the angular position of the target object 1115 relative to WADAR 1105. For example, the measurement signal may include the detection of whether the target object 1115 is a dynamic (e.g., moving) object or a static (e.g., stationary) object. For example, if the target object 1115 is determined to be a dynamic object, the measurement signal may include the velocity and angle of arrival of the target object 1115.

[0102] In this example, WADAR 1105 includes a transmitting element 1120 and N receiving elements 1125 (e.g., N>1). For example, the transmitting element 1120 can transmit a signal 1130 to a target object 1115. Based on the reflection of the signal 1130 received at the receiving elements 1125, TPMS 1100 can determine the angular position and / or angle of arrival of the target object 1115. For example, TPMS 1100 can determine the angular position and / or angle of arrival of the target object 1115 based on the phase difference of the reflections of the signal 1130 received by some or all of the receiving elements 1125.

[0103] In some implementations, WADAR 1105 may include a frequency-modulated continuous wave (FMCW) radar. For example, signal 1130 may contain a sinusoidal wave of electromagnetic radiation. For example, WADAR 1105 may measure the arrival time of a wave reflected from target 1115 to determine the distance to target 1115. For example, WADAR 1105 may measure the frequency variation of the reflected wave from target 1115.

[0104] By using more than one receiving element 1125, for example, PMU 1110 can calculate the time difference of arrival of the reflected signal 1130 between each receiving element 1125 to determine the orientation of the target object 1115. For example, PMU 1110 can use the corresponding phase of each received measurement from the receiving element 1125 to calculate the time difference of arrival of the reflected signal 1130 between each receiving element 1125. In various embodiments, WADAR 1105 may include more than one transmitting element 1120. For example, PMU 1110 may be configured to determine the orientation of the target object 1115 based on the reflection of multiple transmitted signals from multiple transmitting elements (e.g., multiple signals 1130 transmitted from more than one transmitting element 1120). In some embodiments, PMU 1110 may be configured to use a combination of multiple transmitting elements and multiple receiving elements to determine the orientation.

[0105] As shown in the figure, PMU 1110 includes an Angle of Arrival (AOAM) module (1135) and a data storage unit (1140). For example, AOAM 1135 may include a storage module. The storage module may include, for example, one or more storage modules (e.g., non-volatile memory). AOAM 1135 includes a Fast Spectral Energy Calculation (FSECM) module (1145) and a Detailed Spectral Energy Calculation (DSECM) module (1150). As shown in the figure, PMU 1110 includes a processor (1142). Processor 1142 may include, for example, one or more processors. Processor 1142 is operatively coupled to AOAM 1135 and data storage unit (1140).

[0106] As an illustrative example, signal 1130 returned from target object 1115 can be received by receiving element 1125. For example, signal 1130 may include a radio frequency (RF) signal. In some embodiments, processor 1142 may first convert the received signal 1155 into a baseband signal. For example, 1142 may digitize the received signal 1155 for object detection, ranging, angle of arrival measurement, or a combination thereof. For example, processor 1142 may include a digital signal processor (DSP) and / or one or more microprocessors.

[0107] In some implementations, FSECM 1145 and DSECM 1150 may be configured to calculate the angle of arrival of the measured object (e.g., target object 1115). For example, FSECM 1145 and DSECM 1150 may include different calculation methods for determining the angle of arrival. For example, the calculation method may include frequency-based algorithms (e.g., Fast Fourier Transform (FFT), Discrete Fourier Transform (DFT)). For example, the calculation method may include digital beamforming methods (e.g., Bartlett beamformer, delay-sum beamformer, adaptive beamformer, Capon beamformer). For example, the calculation method may include subspace-based methods (e.g., Multi-Signal Classification (MUSIC) methods).

[0108] In various implementations, FSECM 1145 and DSECM 1150 may include different performance characteristics (e.g., in terms of computational cost, angle calculation accuracy, error detection robustness, and minimum detectable interval for multiple objects). In some implementations, FSECM 1145 may include coarser performance in terms of accuracy, error detection robustness, and minimum detectable interval for multiple objects compared to DSECM 1150. In some implementations, FSECM 1145 may include lower computational cost requirements than DSECM 1150. For example, DSECM 1150 may have higher computational intensity than FSECM 1145.

[0109] In this example, PMU 1110 stores the received signal 1155 from WADAR 1105 in data repository 1140. Based on the received signal 1155, FSECM 1145 generates a first heatmap 1160 for each distance range. (Reference) Figure 2A Figure 2C illustrates an exemplary heatmap generated by the FSECM 1145.

[0110] For example, DSECM 1150 can use a more efficient but computationally more expensive method to determine the precise location of target object 1115. As shown in the figure, AOAM 1135 also includes a distance selector 1165. For example, distance selector 1165 can be advantageously used to reduce the computational cost of DSECM 1150.

[0111] In some implementations, DSECM 1150 can be configured to process only the region of interest (ROI 1195) of the received signal 1155. For example, ROI 1195 can be selected by a distance selector 1165. In some implementations, the distance selector 1165 can select ROI 1195 based on an amplitude threshold 1170. For example, the distance selector 1165 can apply the amplitude threshold 1170 to a first heatmap 1160 for each distance range to generate ROI 1195. Alternatively, DSECM 1150 can process ROI 1195 of the received signal 1155 to generate a second heatmap 1175.

[0112] As an illustrative example and not a limitation, AOAM 1135 can process the received signal 1155 in multiple stages. In the first stage, FSECM 1145 can generate a low-resolution angular heatmap (e.g., a first heatmap 1160) of the received spectral energy for each distance and angle combination. For example, AOAM 1135 can determine that regions in the first heatmap 1160 above an amplitude threshold 1170 are considered potential object locations. In various examples, the amplitude threshold 1170 can include a predetermined (e.g., fixed) amplitude threshold (PAT 1180), a dynamically determined threshold (DDT 1185), or a combination thereof. For example, DDT 1185 can be determined using a dynamic method (e.g., by determining a constant false alarm rate (CFAR) threshold).

[0113] In some implementations, distance selector 1165 can compare values ​​from first heatmap 1160 (e.g., the entire field of view of first heatmap 1160) with amplitude threshold 1170 to determine ROI 1195. For example, distance selector 1165 can apply amplitude threshold 1170 to dynamically determine the size of ROI 1195 for high-resolution search around peaks identified in first heatmap 1160. For example, distance selector 1165 can dynamically generate DDT 1185 based on signal information of each frequency cell in first heatmap 1160 and signal information of frequency cells adjacent to that frequency cell. In some implementations, distance selector 1165 can apply full 2D CFAR to first heatmap 1160 to generate DDT 1185.

[0114] In the second stage, for example, DSECM 1150 can be configured to determine the precise object location by processing ROI 1195. In this example, AOAM 1135 also includes an error detection mitigation module (FDMM 1190). For example, FDMM 1190 can use a second heatmap 1175 and a first heatmap 1160 to search for objects that do not meet the minimum interval required by the low-resolution algorithm of FSECM 1145.

[0115] In some embodiments, FDMM 1190 may require a positive detection of the target object 1115 only if it is detected in both the first heatmap 1160 and the second heatmap 1175. In some examples, FDMM 1190 may advantageously eliminate false detections in either FSECM 1145 or DSECM 1150 by combining the results from both. Thus, for example, ROI 1195 may advantageously allow for a more aggressive object detection threshold (e.g., a lower value for PAT 1180 or parameters used to generate DDT 1185 with a lower value) without increasing the false object detection rate.

[0116] In various implementations, the AOAM1135 advantageously reduces the number of computationally expensive operations by first identifying potential objects using FSECM 1145 in the first stage. For example, the DSECM 1150 can be configured to perform the more computationally expensive method only at ROI 1195 (which is selected as having identified potential objects, e.g., selected by distance selector 1165) to determine peak values ​​with higher accuracy. In various examples, the reduction in the number of computationally expensive operations can advantageously reduce processing time and allow for faster measurement rates.

[0117] In some examples, the TPMS 1100 can be embedded in mobile systems such as vehicles, conveyors, and crates transported by forklifts. For instance, a mobile system controller can use the TPMS 1100 to advantageously generate a faster response based on the fast but high-resolution measurement rate generated by the PMU 1110.

[0118] In some implementations, the moving target measurement system (e.g., AOAM 1135) can be configured to sequentially generate N spectral energy heatmaps (e.g., a first heatmap 1160 and a second heatmap 1175) using N independent detection algorithms (e.g., using FSECM 1145 and DSECM 1150). For example, each spectral energy heatmap may be generated based on raw sensor data (e.g., received from receiving element 1125) and independent of the values ​​of other spectral energy heatmaps. For example, the value of each spectral energy heatmap may be generated based on raw sensor data and independent of the values ​​of other spectral energy heatmaps. For example, the moving target measurement system can verify the target (e.g., using FDMM 1190) when the target is identified in at least two of the N spectral energy heatmaps.

[0119] In some examples, the computational cost of the (i-1)th detection algorithm may be lower than that of the ith detection algorithm. For example, the ith detection algorithm may be applied to a region of interest (e.g., ROI 1195) identified based on an amplitude threshold (fixed or dynamically determined) to generate a spectral energy heatmap. For example, the raw sensor data may include multiple data streams (e.g., from receiving element 1125). For example, each is received from an independent physical antenna. For example, at least two subsets of the multiple data streams are used to generate the N spectral energy heatmaps, allowing the location of false detections to be adjusted at each computational stage.

[0120] Figure 12A and Figure 12B An exemplary heatmap for a single target is shown, where a region of interest is dynamically selected in a first heatmap to generate a second heatmap. In this example, in Figure 12A The diagram shows a low-resolution heatmap 1200 (e.g., a first heatmap 1160). For example, FSECM 1145 can generate the low-resolution heatmap 1200 using fast computation methods (e.g., FFT, DFT).

[0121] Distance selector 1165 can process the low-resolution heatmap 1200 by generating a dynamic CFAR threshold 1205 (e.g., DDT 1185) applied to the low-resolution heatmap 1200. As shown, distance selector 1165 can identify ROI 1210 based on regions in the low-resolution heatmap 1200 whose amplitudes are higher than the dynamic CFAR threshold 1205.

[0122] In this example, a high-resolution heatmap 1215 is generated. After identifying the ROI 1210, for example, DSECM 1150 can generate the high-resolution heatmap 1215 by processing the ROI 1210 using a more precise method (e.g., Capon, MUSIC). As shown, the high-resolution heatmap 1215 may include narrower peaks 1220, which are narrower than the peaks 1225 in the low-resolution heatmap 1200. For example, narrower peaks 1220 can advantageously enhance the minimum object spacing (e.g., in the high-resolution heatmap 1215, the likelihood of nearby objects containing overlapping peaks is lower). Furthermore, using the ROI 1210, the high-resolution heatmap 1215 may include a reduced number of shear vectors 1230. For example, shear vectors 1230 may include peaks other than peak 1225. For example, reducing shear vectors can advantageously reduce false object detection.

[0123] Figure 12BA decibel-scaled heatmap 1260 of the low-resolution heatmap 1200 is shown. In some embodiments, an ROI 1210 may be selected after 205 is applied to the decibel-scaled heatmap 1260. As shown, in some examples, a reduced number of shear vectors 1230 may be filtered out before being processed by the DSECM 1150. Therefore, for example, the selection of ROI 1210 may advantageously reduce erroneous object detection.

[0124] Figure 12C and Figure 12D An exemplary heatmap depicting multiple targets within a dynamically selected region of interest is shown. Figure 12C As shown, the low-resolution heatmap 1235 can be generated by FSECM 1145. As an illustrative example, the low-resolution heatmap 1235 can be the result of a field of view from two targets with different signal intensities present at + / -12 degrees. In some examples, independent targets placed close to each other (e.g., within 25 degrees, within 20 degrees, or within 15 degrees) can cause broadened peaks in the low-resolution heatmap 1235, as shown by the green curve 1240.

[0125] In this example, the low-resolution heatmap 1235 includes a dynamically selected Region of Interest (ROI) 1245. For example, the dynamically selected ROI 1245 can be determined based on the signal in the received signal 1155. For instance, the dynamically selected ROI 1245 can vary between measurements (e.g., at + / -10 degrees, at + / -12.5 degrees, at + / -15 degrees, at + / -20 degrees). Based on the dynamically selected ROI 1245, the DSECM 1150 can generate a heatmap such as... Figure 12D The high-resolution heatmap 1250 is shown. Here, high-resolution heatmap 1250 shows two peaks 1255A and 1255B. In some examples, with a fixed selection range width, the high-resolution algorithm can identify only objects with higher amplitudes. For example, dynamically selected ROI 1245 can advantageously improve the minimum detectable interval for multiple objects in PMU 1110.

[0126] Figure 13This is a flowchart illustrating an exemplary dynamic target detection method 1300. For example, the dynamic target detection method 1300 may be executed by a PMU 1110 to determine precise measurements (e.g., distance, angular position, velocity, angle of arrival) of one or more target objects 1115. In this example, the dynamic target detection method 1300 begins in step 1305 when a first spectral energy representation of a first set of object detection signals is generated based on a first calculation method. For example, the PMU 1110 may receive a received signal 1155 from a WADAR 1105. For instance, an AOAM 1135 may use the received signal 1155 to generate a first heatmap 1160 using an FSECM 1145.

[0127] In step 1310, an amplitude threshold is retrieved. For example, amplitude threshold 1170 may be retrieved by distance selector 1165. Next, in step 1315, the amplitude threshold is applied to a first spectral energy representation. For example, distance selector 1165 may apply amplitude threshold 1170 to a first heatmap 1160. In some examples, amplitude threshold 1170 may include a fixed, predetermined threshold (e.g., PAT 1180). In some examples, amplitude threshold 1170 may include a dynamically determined threshold (e.g., DDT 1185). In some examples, amplitude threshold 1170 may be a linear combination of a fixed threshold and a dynamic threshold.

[0128] At decision point 1320, it is determined whether any region within the first spectral energy representation includes an amplitude greater than the amplitude threshold. For example, distance selector 1165 may apply a dynamic CFAR threshold 1205 to the low-resolution heatmap 1200. For example, distance selector 1165 may determine that ROI 1210 (e.g., including peak 1225) may include an amplitude greater than the dynamic CFAR threshold 1205. If there is no region within the first spectral energy representation with an amplitude greater than the amplitude threshold, the dynamic target detection method 1300 terminates.

[0129] If any region with an amplitude greater than an amplitude threshold exists within the first spectral energy representation, then in step 1325, the region of interest is determined using the first spectral energy representation. For example, distance selector 1165 can determine ROI 1210 from low-resolution heatmap 1200. In some examples, the region of interest can be a non-overlapping region within the first spectral energy representation. For example, multiple objects may exist within the FOV of WADAR 1105. Therefore, for example, the first heatmap 1160 may appear to have multiple peaks. For example, ROI 1195 may include a range containing these regions with multiple peaks.

[0130] In step 1330, a second spectral energy representation of the second set of object detection signals is generated based on a second calculation method. For example, DSECM 1150 can use a subset of the received signal 1155 to generate an amplitude threshold 1170. For example, the subset of the received signal 1155 can be determined by the ROI 1195 of the received signal 1155. Next, in step 1335, object detection measurement results are generated based on the second spectral energy representation, and the dynamic target detection method 1300 ends. For example, the object detection measurement results can be generated based on a second heatmap 1175. For example, the object detection measurement results can include object detection results indicating the relative position of the object within the field of view of WADAR 1105. For example, the object detection results can include the AoA of each detected object.

[0131] In some implementations, PMU 1110 may vary the number of effective antennas used in steps 1305 and 1330 to reduce detection errors caused by the first and second calculation methods. For example, in some error detection cases, the location of the erroneous object detection may be a function of the number of antennas (e.g., effective antennas) used for processing during the calculation process. For example, if 8 antennas are used for processing, as an illustrative example, error detection may occur at multiples of 180 degrees / 8 antennas = 22.5 degrees. In some examples, the location of these error detections can be adjusted by selecting to process only a subset of data from the available antennas at different stages. For example, if only data from 7 antennas is processed in step 1330, error detection in the second calculation method may occur at multiples of 180 degrees / 7 antennas = 25.7 degrees. Therefore, in some embodiments, PMU 1110 may use a different number of effective antennas in each stage (stage 1, stage 2, ..., stage N, where N≥2) to advantageously reduce error detection at the same location. For example, by changing the effective antenna at different stages, the PMU1110 can advantageously allow for a more aggressive object detection threshold without increasing the probability of false object detection.

[0132] Figure 14This is a flowchart illustrating an exemplary AOAM configuration method 1400. For example, PMU 1110 may be configured using AOAM configuration method 1400 during the manufacturing process. In some examples, a user may use AOAM configuration method 1400 to configure PMU 1110. In this example, the exemplary AOAM configuration method 1400 begins when a first spectral energy calculation method is selected in step 1405. For example, FSECM 1145 may be selected. Next, a second spectral energy calculation method is selected in step 1410. For example, DSECM 1150 may be selected. In some embodiments, the second spectral energy calculation method may require higher computational power than the first spectral energy calculation method. In some embodiments, the second spectral energy calculation method may produce results with higher accuracy than the first spectral energy calculation method.

[0133] In step 1415, a predetermined amplitude threshold is selected. For example, amplitude threshold 1170 can be selected. For example, the predetermined amplitude threshold may include a fixed threshold. For example, the fixed threshold may be determined based on experimental data. For example, the fixed threshold may be determined based on heuristic data. At decision point 1420, it is determined whether to include a dynamic amplitude threshold. If it is determined that a dynamic amplitude threshold is not included, the selected amplitude threshold is set as the predetermined amplitude threshold in step 1425. After setting the selected amplitude threshold, in step 1430, the first spectral energy calculation method and the second spectral energy calculation method, along with the selected amplitude threshold, are saved to a data repository, and AOAM configuration method 1400 ends. For example, FSECM 1145 and DSECM 1150 may be stored in AOAM 1135. For example, amplitude threshold 1170 may be stored in data repository 1140.

[0134] If it is determined at decision point 1420 that a dynamic amplitude threshold should be included, then in step 1435, a calculation method for determining the dynamic amplitude threshold is selected. For example, PMU 1110 may determine DDT 1185 based on the noise level of the received signal 1155. For example, PMU 1110 may use multidimensional CFAR technology to determine DDT 1185. In step 1440, the amplitude threshold is set to a combination of the predetermined amplitude threshold and the dynamic amplitude threshold, and step 1430 is performed. For example, the first spectral energy calculation method and the second spectral energy calculation method, along with the selected amplitude threshold, may be saved to the data repository. In some examples, the selected amplitude threshold may depend only on the dynamic component. For example, in this case, the predetermined amplitude threshold may be set to 0.

[0135] Figure 15This is a block diagram depicting an exemplary target object detection system (TODS 1500). TODS 1500 includes a processor 1505. Processor 1505 may, for example, include one or more processing units. Processor 1505 is operatively coupled to a communication module 1510. Communication module 1510 may, for example, include wired communication. Communication module 1510 may, for example, include wireless communication. For example, communication module 1510 may include a communication interface 770. In the depicted example, communication module 1510 is operatively coupled to a radar 1515. For example, radar 1515 may include a transmitter 105 and a receiver 110. For example, radar 1515 may include an FTRE 720. For example, radar 1515 may include a WADAR 1105.

[0136] Processor 1505 is operatively coupled to memory module 1520. Memory module 1520 may include, for example, one or more memory modules (e.g., random access memory (RAM)). Processor 1505 includes storage module 1525. Storage module 1525 may include, for example, one or more storage modules (e.g., non-volatile memory). In the depicted example, storage module 1525 includes an object clustering and merging engine (OCCE 1530), a slow dynamic target measurement engine (SDTME 1535), and a target AOA determination engine (TADE 1540). Processor 1505 is also operatively coupled to data repository 1545. Data repository 1545 includes statistical boundary rules 1550, signal DC components 1555, fast spectral energy calculation rules 1560, and detailed spectral energy calculation rules 1565.

[0137] For example, OCCE 1530 can apply statistical distributions (such as statistical boundary rule 1550) to signal clustering to determine cluster boundaries. For example, OCCE 1530 can perform reference... Figures 1 to 6B One or more of the methods described.

[0138] For example, SDTME 1535 can be configured to reapply the DC component 1555 of the signal received from radar 1515 when a slow-moving target is identified, in order to obtain a more accurate measurement of the target. For example, SDTME 1535 can perform a reference... Figures 7 to 10 One or more of the methods described.

[0139] For example, TADE 1540 can first apply fast spectral energy calculation rule 1560 to generate a first heatmap. For example, TADE 1540 can use detailed spectral energy calculation rule 1565 to generate a second heatmap at the identified region of interest. For example, TADE 1540 can perform a reference... Figures 11 to 14 One or more of the methods described.

[0140] Although various embodiments have been described with reference to the accompanying drawings, other embodiments are also possible. In various embodiments, the SPACM 135 can dynamically determine the number of clusters in a viewpoint. For example, the SPACM 135 can identify clusters based on whether the intensity of one or more reflected signals is higher than a predetermined threshold.

[0141] In some embodiments, the boundaries of the clusters may include other shapes. For example, the boundaries may be rectangular boxes. In some examples, the boundaries may include ellipses. In some embodiments, the boundaries may be triangular. In some embodiments, the boundaries may include cylinders. For example, when identifying boundaries (e.g., boundaries 220A-220C), the statistical distribution function of each identified cluster can be determined using a single two-dimensional distribution function (e.g., a circle in a two-dimensional point cloud and a sphere in a three-dimensional point cloud). In some examples, the statistical distribution function may include distribution functions (e.g., rectangular or other arbitrary distribution shapes) calculated separately (e.g., independently) in each dimension.

[0142] In some implementations, the PMU 1110 may process the received signal 1155 in three or more stages. For example, after generating the second heatmap 1175, the PMU 1110 may select another ROI within the second heatmap 1175 for processing by other processing methods. For example, a PMU 1110 with more stages may advantageously further reduce the false detection rate.

[0143] Although an exemplary system has been described with reference to the accompanying drawings, other implementations can be deployed in other industrial, scientific, medical, commercial and / or residential applications.

[0144] In some embodiments, the TPMS 1100 can be used on a loading / unloading platform. For example, the loading / unloading platform can detect objects in the environment. For example, based on measurements of the detected objects, the loading / unloading platform can determine whether they are available for loading.

[0145] In some embodiments, a parking lot may use a TPMS 1100 to determine the availability of parking spaces. For example, because a TPMS 1100 can have a small minimum detectable interval for multiple objects, a single TPMS 1100 can be used to monitor multiple parking spaces.

[0146] In some embodiments, the TPMS 1100 can be advantageously used for counting applications (e.g., filling machines). For example, the TPMS 1100 can be used to monitor crates and count the number of items placed in them.

[0147] In some embodiments, the TPMS 1100 can be used in a conveyor belt system. For example, the TPMS 1100 can be used to detect the speed and angle of movement of objects on the conveyor belt system. For example, the TPMS 1100 can be advantageously used to detect jams in the conveyor belt system.

[0148] In various examples, MODS 100, DODAR device 715, and PMU 1110 can be implemented in any combination within the system. For example, DODAR device 715 may include MODS 100 to improve object recognition accuracy by using MCU 130. For example, DODAR device 715 may be connected to PMU 1110 to provide re-cluttered data for generating more accurate measurements of slow-moving objects. In some examples, MODS 100 may be implemented together with PMU 1110 in the same vehicle control system to advantageously improve dynamic target measurement, thereby enhancing, for example, autonomous driving capabilities.

[0149] In various embodiments, some bypass circuitry implementations can be controlled in response to signals from analog or digital components, which can be discrete, integrated, or a combination of each. Some embodiments may include programmable devices, programmable devices, or some combination thereof (e.g., PLA, PLD, ASIC, microcontroller, microprocessor), and may include one or more data repositories (e.g., cells, registers, blocks, pages) that provide single-level or multi-level digital data storage capabilities, and may be volatile, non-volatile, or some combination thereof. Some control functions can be implemented using hardware, software, firmware, or any combination thereof.

[0150] A computer program product may contain a set of instructions that, when executed by a processor device, cause the processor to perform specified functions. These functions may be performed in conjunction with a controlled device that is operatively in communication with the processor. A computer program product that may include software may be stored in a data repository tangibly embedded in a storage medium such as an electronic storage device, magnetic storage device, or rotating storage device, and may be fixed or removable (e.g., hard disk, floppy disk, thumb drive, CD, DVD).

[0151] Although an example of a portable system has been described with reference to the above figures, other implementations can be deployed in other processing applications, such as desktop and networked environments.

[0152] For example, temporary auxiliary energy input can be received from a rechargeable or disposable battery, enabling use in portable or remote applications. Some embodiments may operate using other DC voltage sources, such as (nominal) batteries. Alternating current (AC) input, which can be provided, for example, from a 50 / 60 Hz power port or a portable generator, can be received via a rectifier and appropriate scaling. The supply of AC input (e.g., sine, square, or triangle wave) input may include a line frequency converter to provide voltage boost, voltage buck, and / or isolation.

[0153] While specific characteristics of the architecture have been described, performance can be improved by combining other features. For example, caching techniques (e.g., L1, L2, ...) can be used. Random access memory may be included, for example, to provide high-speed scratch pad memory and / or load stored executable code or parameter information for use during runtime operation. Other hardware and software may be provided to perform the operation, such as network or other communication using one or more protocols, wireless (e.g., infrared) communication, stored operating energy and power supply (e.g., battery), switching and / or linear power supply circuitry, software maintenance (e.g., self-testing, upgrades), etc. One or more communication interfaces may be provided to support data storage and related operations.

[0154] Some systems can be implemented as computer systems that can be used with various embodiments. For example, various embodiments may include digital circuits, analog circuits, computer hardware, firmware, software, or combinations thereof. Apparatus may be implemented in a computer program product tangibly embodied in an information carrier, such as a machine-readable storage device, for execution by a programmable processor; and methods may be executed by a programmable processor executing a program of instructions to perform the functions of various embodiments by manipulating input data and generating output. Various embodiments may advantageously be implemented in one or more computer programs that can be executed in a programmable system including at least one programmable processor coupled to receive data and instructions from a data storage system, at least one input device, and / or at least one output device, and to send data and instructions to the data storage system, at least one input device, and / or at least one output device. A computer program is a set of instructions that can be used directly or indirectly in a computer to perform a specific activity or produce a specific result. Computer programs can be written in any programming language, including compiled or interpreted languages, and can be deployed in any form, including as standalone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

[0155] As an example, suitable processors for executing instruction programs include both general-purpose microprocessors and special-purpose microprocessors, which can include one or more processors in any type of computer. Typically, the processor receives instructions and data from read-only memory or random access memory, or both. The key components of a computer are the processor for executing instructions and one or more memories for storing instructions and data. Typically, a computer will also include one or more mass storage devices for storing data files, or operatively coupled to communicate with such mass storage devices; such devices include disks, such as internal hard disks and removable disks; magneto-optical disks; and optical disks. Storage devices suitable for tangibly containing computer program instructions and data include all forms of non-volatile memory, including, as examples, semiconductor memory devices such as EPROM, EEPROM, and flash memory devices; disks such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM discs. The processor and memory can be supplemented by or incorporated into an ASIC (Application-Specific Integrated Circuit).

[0156] In some implementations, each system can be programmed with the same or similar information and / or initialized with substantially the same information stored in volatile and / or non-volatile memory. For example, a data interface can be configured to perform automatic configuration, automatic download, and / or automatic update functions when coupled to an appropriate host device such as a desktop computer or server.

[0157] In some implementations, one or more user interface features can be customized to perform specific functions. Various embodiments can be implemented in computer systems including graphical user interfaces and / or internet browsers. To provide interaction with the user, some implementations can be implemented on a computer with a display device. The display device can, for example, include an LED (light-emitting diode) display. In some implementations, the display device can, for example, include a CRT (cathode ray tube). In some implementations, the display device can include, for example, an LCD (liquid crystal display). The display device (e.g., a monitor) can be used, for example, to display information to the user. Some implementations can, for example, include a keyboard and / or pointing devices (e.g., a mouse, touchpad, trackball, joystick), through which the user can provide input to the computer.

[0158] In various implementations, the system can communicate using suitable communication methods, equipment, and technologies. For example, the system can communicate with compatible devices (e.g., devices capable of transferring data to and / or from the system) using point-to-point communication, in which messages are transmitted directly from the source to the receiver via a dedicated physical link (e.g., fiber optic link, point-to-point cabling, daisy chain). Components of the system can exchange information via analog or digital data communication of any form or medium, including packet-based messaging over a communication network. Examples of communication networks include, for example, LANs (Local Area Networks), WANs (Wide Area Networks), MANs (Metropolitan Area Networks), wireless networks and / or optical networks, computers and networks forming the Internet, or some combination thereof. Other implementations can transmit messages by broadcasting to all or substantially all devices coupled together by the communication network, for example, by using omnidirectional radio frequency (RF) signals. Other implementations can transmit messages characterized by high directionality, such as RF signals transmitted using directional (i.e., narrow-beam) antennas or infrared signals that may optionally be used with focusing optics. Other implementations are also possible using appropriate interfaces and protocols, such as, by way of example but not intended to be limiting, USB 2.0, Firewire, ATA / IDE, RS-232, RS-422, RS-485, 802.11 a / b / g, Wi-Fi, Ethernet, IrDA, FDDI (Fiber Distributed Data Interface), Token Ring, frequency division, time division, or code division based multiplexing techniques, or some combination thereof. Some implementations may optionally incorporate features such as error detection and correction (ECC) for data integrity, or security measures such as encryption (e.g., WEP) and password protection.

[0159] In various embodiments, the computer system may include Internet of Things (IoT) devices. IoT devices may include objects embedded with electronic devices, software, sensors, actuators, and network connectivity that enable these objects to collect and exchange data. IoT devices can be used with wired or wireless devices by sending data to another device through an interface. IoT devices can collect useful data and then autonomously transfer it between other devices.

[0160] Various examples of circuits including a variety of electronic hardware can be used to implement the module. By way of example, and not limitation, the hardware may include transistors, resistors, capacitors, switches, integrated circuits, other modules, or some combination thereof. In various examples, the module may include analog logic, digital logic, discrete components, traces, and / or memory circuits, or some combination thereof, fabricated on a silicon substrate including various integrated circuits (e.g., FPGAs, ASICs). In some embodiments, the module may relate to the execution of pre-programmed instructions, software executed by a processor, or some combination thereof. For example, various modules may involve both hardware and software.

[0161] In an illustrative aspect, the first object detection system may include signal transmitting and receiving devices, which may include a transmitter and a receiver. For example, the transmitter may be configured to transmit a predetermined waveform toward a target object, and the receiver may be configured to receive reflected signals reflected from the surface of the target object.

[0162] For example, the object detection system may include a signal processor operatively coupled to the signal transmitting and receiving devices. For instance, the signal processor may be configured to receive the reflected signal and generate a multidimensional point cloud, and within the multidimensional point cloud, identify N clusters (C_1, C_2, ..., C_i, ..., C_N) based on a clustering algorithm, where N can be an integer and N≥0, and each i-th cluster (i < N) of the N clusters may include M_i points (P^i_1, P^i_2, ..., P^i_j, ..., P^i_M).

[0163] For example, the object detection system may include a clustering processing unit operatively coupled to the signal processor. For example, the signal processor may include a clustering boundary engine configured to associate a boundary with each of the N clusters. For example, the boundary of the k-th cluster may be determined as a predetermined multiple of the standard deviation of the points (P^k_1, C^k_2, ..., P^k_j, ..., P^k_M) of the k-th cluster. For example, the predetermined multiple may be greater than 4. For example, the signal processor may include a cluster merging module operatively coupled to the clustering boundary engine. For example, the cluster merging module may be configured to generate a detection result, including detected object information, by merging clusters in the N clusters when any two or more clusters in the N clusters are associated with an overlapping space within their respective boundaries. For example, the cluster merging module may reduce false object identification due to over-clustering of individual objects.

[0164] For example, a first object detection system may include one or more of the following features: - For example, signal transmitting and receiving equipment may include FMCW radar.

[0165] - For example, clustering algorithms may include the DBSCAN algorithm. For example, a signal processor performs the DBSCAN algorithm using a small-distance parameter, which is smaller than a predetermined gap between points received from a single object containing multiple reflective surfaces.

[0166] For example, the clustering boundary engine determines the boundary of each of the N clusters as a statistical distribution function of the points in the corresponding cluster. For instance, the statistical distribution function can be calculated independently in each dimension of the multidimensional point cloud.

[0167] - For example, an object detection system may include a static target suppression module. For example, the object detection system may be configured to detect static and dynamic objects independently. For example, object detection may include identifying the location of the object. For example, the static target suppression module may be configured to compare the static object and the dynamic object, and the static object may be removed when it is likely within a proximity threshold of the dynamic object.

[0168] - For example, the proximity threshold can be dynamically determined based on a probability function associated with each of the N clusters. For example, the probability function of the k-th cluster can be determined as a function of the points in the k-th cluster.

[0169] - For example, the proximity threshold may be less than 0.25m.

[0170] In an illustrative aspect, the second object detection system may include a signal processor operatively coupled to a signal transmitting and receiving device. For example, the signal processor may be configured to receive signals from the signal transmitting and receiving device and generate a multidimensional point cloud, within which N clusters (C_1, C_2, ..., C_i, ..., C_N) are identified based on a clustering algorithm, where N may be an integer and N≥0, and each i-th cluster (i < N) of the N clusters may include M_i points (P^i_1, P^i_2, ..., P^i_j, ..., P^i_M).

[0171] For example, the object detection system may include a clustering processing unit operatively coupled to the signal processor. For example, the signal processor may include a clustering boundary engine configured to associate a boundary with each of the N clusters. For example, the boundary of the k-th cluster may be determined as a function of the points (P^k_1, C^k_2, ..., P^k_j, ..., P^k_M) of the k-th cluster, where k < N. For example, the signal processor may include a cluster merging module operatively coupled to the clustering boundary engine. For example, the cluster merging module may be configured to generate a detection result, including detected object information, by merging clusters in the N clusters when any two or more of the N clusters may be associated with overlapping space within their respective boundaries. For example, the cluster merging module may reduce false object identification due to over-clustering of individual objects.

[0172] For example, the second object detection system may include one or more of the following features: - For example, the signal transmitting and receiving equipment may include an FMCW radar.

[0173] - For example, the clustering algorithm may include the DBSCAN algorithm. For example, the signal processor performs the DBSCAN algorithm using a distance parameter that is smaller than a predetermined gap between points received from a single object containing multiple reflective surfaces.

[0174] For example, the clustering boundary engine determines the boundary of each of the N clusters as a statistical distribution function of the points in the corresponding cluster. For example, the statistical distribution function can be calculated independently in each dimension of the multidimensional point cloud.

[0175] - For example, the clustering boundary engine determines the boundary of each of the N clusters as a predetermined multiple of the standard deviation of the points of the corresponding cluster in each dimension of the multidimensional point cloud.

[0176] - For example, the predetermined multiple can be 4.5.

[0177] - For example, the signals received from the signal transmitting and receiving devices may include signals reflected from dynamic targets, and the multidimensional point cloud may include dynamic points.

[0178] - For example, the object detection system may include a static target suppression module. For example, the object detection system may be configured to detect static and dynamic objects independently. For example, object detection may include identifying the location of the detected objects. For example, the static target suppression module may be configured to compare the static object and the dynamic object, and the static object may be removed when the static object is within a proximity threshold of the dynamic object.

[0179] - For example, the proximity threshold can be dynamically determined based on a probability function associated with each of the N clusters. For example, the probability function of the k-th cluster can be determined as a function of the points in the k-th cluster.

[0180] - For example, the proximity threshold may be less than 0.25m.

[0181] In an illustrative aspect, an object detection method may include receiving a signal from a field of view. For example, the object detection method may include identifying N initial clusters within a multidimensional point cloud based on the signal received from the field of view. For example, each i-th cluster (i < N) of the N initial clusters may include M_i points (P^i_1, P^i_2, ..., P^i_j, ..., P^i_M). For example, the object detection method may include determining and associating statistical cluster boundaries for each of the N initial clusters. For example, the statistical cluster boundary of the k-th cluster may be determined as a function of the points (P^k_1, P^k_2, ..., P^k_j, ..., P^k_M) of the k-th cluster, where k < N. For example, the object detection method may include generating a detection result by merging the N initial clusters when any two or more of the N initial clusters may be associated with an overlapping space contained within the respective cluster boundaries, the detection result including detected object information. For example, it can reduce the misidentification of objects caused by over-clustering of individual objects.

[0182] For example, the object detection method may include one or more of the following features: For example, determining the statistical clustering boundary may include determining the statistical distribution function of the points in the corresponding clusters among the N initial clusters. For example, the statistical distribution function may be calculated independently in each dimension of the multidimensional point cloud.

[0183] - For example, the statistical distribution function may include a predetermined multiple of the standard deviation of the points corresponding to the cluster.

[0184] In an illustrative aspect, a dynamic target measurement system may include a first data repository comprising instructions. For example, the dynamic target measurement system may include a processor operatively coupled to the first data repository such that, when the processor executes the instructions, the processor causes operations to be performed to automatically generate multidimensional precision measurements of slow-moving objects within a static object.

[0185] For example, the operation may include receiving a measurement signal from a direction and distance measuring device. For example, the operation may include generating a first component of the measurement signal. For example, the operation may include saving the first component to a second data repository. For example, the operation may include generating a second component of the measurement signal by subtracting the first component from the measurement signal. For example, the operation may include generating a first Fast Fourier Transform (FFT) representation of the second component. For example, the operation may include applying a threshold to the first FFT representation to determine the presence of a slow-moving target.

[0186] For example, the operation may include retrieving the first component from the second data repository when a slow target exists according to the first FFT representation. For example, the operation may include aggregating the first component and the second component to generate an aggregated measurement signal. For example, the operation may include generating a second FFT representation of the aggregated measurement signal. For example, the operation may include applying an interpolation operation to the second FFT representation. For example, the operation may include measuring the velocity of the slow target based on the second FFT representation.

[0187] For example, the dynamic target measurement system may include one or more of the following features: - For example, the direction and range measuring device may include a frequency-modulated continuous wave radar operatively connected to the processor.

[0188] - For example, the measurement signal may include a Doppler chirp signal.

[0189] - For example, the first component of the measurement signal may include a DC offset extracted from the measurement signal.

[0190] - For example, the first FFT representation and the second FFT representation may include a Doppler FFT representation.

[0191] For example, applying a threshold to the first FFT representation to determine whether a slow target exists based on the first FFT representation may include retrieving a bin threshold from a third data repository. For example, the determination may include retrieving a modulus threshold from a fourth data repository. For example, the determination may include identifying peaks in the first FFT representation. For example, the peak may be identified as a modulus of the corresponding FFT bin of the first FFT representation that is greater than the modulus threshold. For example, the determination may include determining the bin number of the corresponding FFT bin for the identified peak. For example, the determination may include comparing the bin number with the bin threshold. For example, the determination may include generating a signal indicating that the slow target has been identified when the bin number may be less than the bin threshold.

[0192] - For example, the modulus threshold can be dynamically generated using the constant false alarm rate (CFAR) method as a function of the measured signal.

[0193] In an illustrative aspect, a computer-implemented method for automatically generating multidimensional precision measurement results of a slow-moving object in a static object, executed by at least one processor, the method comprising: generating a first component of a measurement signal received from a direction and distance measuring device.

[0194] For example, the method may include: storing the first component to a first data repository; generating a second component of the measurement signal by subtracting the first component from the measurement signal. For example, the method may include: generating a first Fast Fourier Transform (FFT) representation of the second component.

[0195] For example, the method may include: determining whether a slow target might exist according to the first FFT representation based on identifying whether there is a peak in the small FFT bin in the first FFT representation.

[0196] For example, the method may include: retrieving the first component from the first data repository when a slow target may exist according to the first FFT representation; aggregating the first component and the second component to generate an aggregated measurement signal; generating a second FFT representation of the aggregated measurement signal; applying interpolation to the second FFT representation; and / or measuring the velocity of the slow target based on the second FFT representation. For example, interpolation errors in the small FFT repository can be reduced.

[0197] For example, the computer implementation method may include one or more of the following features: - For example, the measurement signal may include a Doppler chirp signal.

[0198] - For example, the first component of the measurement signal may include a DC offset extracted from the measurement signal.

[0199] - For example, the first FFT representation and the second FFT representation may include a Doppler FFT representation.

[0200] - For example, determining whether a slow target might exist based on the first FFT representation may include retrieving a bin threshold from a second data repository. For example, the determination may include retrieving a modulus threshold from a third data repository. For example, the determination may include identifying peak values ​​in the first FFT representation. For example, the peak value may be identified as a modulus of the corresponding FFT bin of the first FFT representation that is greater than the modulus threshold. For example, the determination may include determining the bin number of the corresponding FFT bin for the identified peak value. For example, the determination may include comparing the bin number with the bin threshold. For example, the determination may include generating a signal indicating that the slow target has been identified when the bin number is less than the bin threshold.

[0201] - For example, the modulus threshold can be dynamically generated using the constant false alarm rate (CFAR) method as a function of the measured signal.

[0202] In an illustrative aspect, a computer program product includes instructions tangibly embodied on a non-transitory computer-readable medium, wherein, when the instructions are executable on a processor, the processor causes the execution of a detection operation to automatically generate a multidimensional precision measurement of a slow-moving object in a static object, the operation including: receiving a measurement signal from a direction and distance measuring device. For example, the operation may include generating a first component of the measurement signal. For example, the operation may include storing the first component in a first data repository. For example, the operation may include generating a second component of the measurement signal by subtracting the first component from the measurement signal. For example, the operation may include generating a first Fast Fourier Transform (FFT) representation of the second component. For example, the operation may include determining whether a slow target is likely to exist according to the first FFT representation based on identifying the presence of peaks within small FFT bins in the first FFT representation.

[0203] For example, when a slow target may exist based on the first FFT representation, the operation may include retrieving the first component from the first data repository, aggregating the first component and the second component to generate an aggregated measurement signal, generating a second FFT representation of the aggregated measurement signal, applying interpolation to the second FFT representation, and / or measuring the velocity of the slow target based on the second FFT representation. For example, interpolation errors in the smaller FFT repository can be reduced.

[0204] For example, the computer program product may include one or more of the following features: - For example, the measurement signal may include a Doppler chirp signal.

[0205] - For example, the first component of the measurement signal may include a DC offset extracted from the measurement signal.

[0206] - For example, the first FFT representation and the second FFT representation may include Doppler Fast Fourier Transform representations.

[0207] For example, determining whether a slow target might exist according to the first Fast Fourier Transform (FFT) representation may include retrieving a bin threshold from a second data repository. For example, the determination may include retrieving a modulus threshold from a third data repository. For example, the determination may include identifying a peak in the first FFT representation. For example, the peak may be identified as a modulus of the corresponding FFT bin of the first FFT representation that is greater than the modulus threshold. For example, the determination may include determining the bin number of the corresponding FFT bin of the identified peak. For example, the determination may include comparing the bin number with the bin threshold. For example, the determination may include generating a signal indicating that the slow target has been identified when the bin number is less than the bin threshold.

[0208] - For example, the modulus threshold can be dynamically generated using the constant false alarm rate (CFAR) method as a function of the measured signal.

[0209] For example, the interpolation operation may include applying various interpolation techniques to the second FFT representation, thereby improving the linearity and accuracy of the velocity measurement results of the slow target, thus obtaining accurate measurement results in each Doppler FFT chamber.

[0210] In an illustrative aspect, a system may include a first data repository comprising instructions. For example, the system may include a processor operatively coupled to the first data repository, such that when the processor executes the instructions, the processor causes operations to be performed to automatically execute a multi-stage precision measurement of a target, the operations including: generating a first spectral energy representation of a first set of object detection signals based on a first computational method. For example, the first set of object detection signals may be received from a direction and range measuring device corresponding to the field of view of the direction and range measuring device.

[0211] For example, the operation may include: retrieving an amplitude threshold from a second data repository. For example, the operation may include: applying the amplitude threshold to the first spectral energy representation. For example, the operation may include: determining a region of interest within the first spectral energy representation. For example, the region of interest may include a frequency range within the first spectral energy representation where the amplitude is greater than the amplitude threshold.

[0212] For example, the operation may include: generating a second spectral energy representation of a second set of object detection signals based on a second calculation method. For example, the second set of object detection signals may include a subset of the first set of object detection signals corresponding to the region of interest. For example, the second calculation method may have a higher computational cost and include higher resolution than the first calculation method. For example, the operation may include: generating object detection measurement results based on the second spectral energy representation. For example, the object detection measurement results may include multiple objects identified within the first set of object detection signals. For example, the object can be identified when both the first and second spectral energy representations may contain peaks corresponding to the object. For example, the false object detection rate may be reduced to below the first false object detection rate of the first calculation method and the second false object detection rate of the second calculation method.

[0213] For example, the system may include one or more of the following features: - For example, the frequency range of the region of interest can be dynamically determined and may include a continuous range.

[0214] - For example, the amplitude threshold may include a dynamically determined threshold.

[0215] - For example, the dynamically determined threshold can be determined based on the noise level of the detection signals of the first group of objects.

[0216] - For example, the dynamically determined threshold can be determined based on applying a two-dimensional constant false alarm rate algorithm to the overall detection signals of the first group of objects.

[0217] For example, the first set of object detection signals may be received from N antennas, and the second set of object detection signals may include signals received from a subset of the N antennas. For example, erroneous detection locations related to the number of effective antennas generated by the first and second calculation methods can be mitigated.

[0218] In an illustrative aspect, a computer-implemented method for automatically performing multi-stage precision measurements of a target, executed by at least one processor, is provided. For example, the method may include: generating a first spectral energy representation of a first set of object detection signals based on a first computational method. For example, the first set of object detection signals may be received from a direction and ranging device corresponding to the field of view of a direction and ranging device. For example, the method may include: retrieving an amplitude threshold from a second data repository.

[0219] For example, the method may include: applying the amplitude threshold to the first spectral energy representation. For example, the method may include: determining a region of interest within the first spectral energy representation. For example, the region of interest may include a frequency range within the first spectral energy representation where the amplitude is greater than the amplitude threshold. For example, the method may include: generating a second spectral energy representation of a second set of object detection signals based on a second calculation method. For example, the second set of object detection signals may include a subset of the first set of object detection signals corresponding to the region of interest. For example, the second calculation method may have a higher computational load and may include higher resolution than the first calculation method. For example, the method may include: generating object detection measurement results based on the second spectral energy representation.

[0220] For example, the computer implementation method may include one or more of the following features: For example, generating the object detection measurement results may include identifying multiple objects within the first set of object detection signals. For example, the object may be identified when both the first spectral energy representation and the second spectral energy representation may contain peaks corresponding to the object. For example, the false object detection rate may be reduced to below the first false object detection rate of the first calculation method and the second false object detection rate of the second calculation method.

[0221] - For example, the frequency range of the region of interest can be dynamically determined and may include a continuous range.

[0222] - For example, the amplitude threshold may include a dynamically determined threshold.

[0223] - For example, the dynamically determined threshold can be determined based on the noise level of the detection signals of the first group of objects.

[0224] - For example, the dynamically determined threshold can be determined based on applying a two-dimensional constant false alarm rate algorithm to the overall detection signals of the first group of objects.

[0225] For example, the first set of object detection signals may be received from N antennas, and the second set of object detection signals may include signals received from a subset of the N antennas. For example, erroneous detection results related to the effective number of antennas generated by the first and second calculation methods can be mitigated.

[0226] In an illustrative aspect, a computer program product includes an instruction program tangibly embodied on a non-transitory computer-readable medium, wherein when the instructions are executed on a processor, the processor causes measurement and detection operations to automatically perform multi-stage precision measurements of a target. For example, the operations may include: generating a first spectral energy representation of a first set of object detection signals based on a first computational method. For example, the first set of object detection signals may be received from a direction and range measuring device corresponding to the field of view of the direction and range measuring device.

[0227] For example, the operation may include: retrieving an amplitude threshold from a second data repository. For example, the operation may include: applying the amplitude threshold to the first spectral energy representation. For example, the operation may include: determining a region of interest within the first spectral energy representation. For example, the region of interest may include a frequency range within the first spectral energy representation where the amplitude is greater than the amplitude threshold. For example, the operation may include: generating a second spectral energy representation of a second set of object detection signals based on a second calculation method. For example, the second set of object detection signals may include a subset of the first set of object detection signals corresponding to the region of interest. For example, the second calculation method may have a higher computational load and may include a higher resolution than the first calculation method. For example, the operation may include: generating object detection measurement results based on the second spectral energy representation.

[0228] For example, the computer program product may include one or more of the following features: For example, generating the object detection measurement results may include identifying multiple objects within the first set of object detection signals. For example, the object may be identified when both the first spectral energy representation and the second spectral energy representation may contain peaks corresponding to the object. For example, the false object detection rate may be reduced to below the first false object detection rate of the first calculation method and the second false object detection rate of the second calculation method.

[0229] - For example, the frequency range of the region of interest can be dynamically determined and may include a continuous range.

[0230] - For example, the amplitude threshold may include a dynamically determined threshold.

[0231] - For example, the dynamically determined threshold can be determined based on the noise level of the detection signals of the first group of objects.

[0232] - For example, the dynamically determined threshold can be determined based on applying a two-dimensional constant false alarm rate algorithm to the overall detection signals of the first group of objects.

[0233] For example, the first set of object detection signals may be received from N antennas, and the second set of detection signals may include signals received from a subset of the N antennas. For example, erroneous detection signals related to the effective number of antennas generated by the first and second calculation methods can be mitigated.

[0234] Many embodiments have been described. However, it should be understood that various modifications can be made. For example, advantageous results can be achieved if the steps of the disclosed technology are performed in a different order, or if the components of the disclosed system are combined in a different manner, or if these components are supplemented with other components. Therefore, other embodiments are contemplated within the scope of the appended claims.

Claims

1. An object detection system, comprising: a signal transmitting and receiving device comprising a transmitter and a receiver, wherein the transmitter is configured to transmit a predetermined waveform towards a target object, and the receiver is configured to receive a reflection signal reflected from a surface of the target object; a signal processor operably coupled to the signal transmitting and receiving device, wherein the signal processor is configured to receive the reflection signal and generate a multi-dimensional point cloud, and identify N clusters (C_1, C_2, ..., C_i, ..., C_N) within the multi-dimensional point cloud based on a clustering algorithm, wherein N is an integer greater than or equal to 0, and each i-th cluster of the N clusters, i<N, comprises M_i points (P^i_1, P^i_2, ..., P^i_j, ..., P^i_M); and a clustering processing unit operably coupled to the signal processor, the clustering processing unit comprising: a clustering boundary engine configured to associate a boundary with each of the N clusters, wherein the boundary of a k-th cluster is determined as a predetermined multiple of a standard deviation of points (P^k_1, P^k_2, ..., P^k_j, ..., P^k_M) of the k-th cluster, wherein the predetermined multiple is greater than 4; and a clustering merging module operably coupled to the clustering boundary engine, wherein the clustering merging module is configured to, when any two or more clusters among the N clusters are associated with an overlapping space within corresponding boundaries, generate a detection result comprising detected object information by merging the clusters among the N clusters, such that the clustering merging module reduces erroneously identified objects caused by over-clustering of a single object.

2. The object detection system according to claim 1, wherein, The signal transmitting and receiving device comprises an FMCW radar.

3. The object detection system according to claim 1, wherein, The clustering algorithm comprises a DBSCAN algorithm, wherein the signal processor executes the DBSCAN algorithm using a small distance parameter that is smaller than a predetermined gap between points received from the single object comprising a plurality of reflection surfaces.

4. The object detection system according to claim 1, wherein, The clustering boundary engine determines the boundary of each of the N clusters as a statistical distribution function of points of the corresponding cluster among the N clusters, wherein the statistical distribution function is calculated independently in each dimension of the multi-dimensional point cloud.

5. The object detection system according to claim 1, further comprising a static target suppression module, and wherein: the object detection system is configured to detect static objects and dynamic objects independently, wherein detecting an object comprises identifying a position of the object, and the static target suppression module is configured to compare the static objects and the dynamic objects, and remove a static object when the static object is within a proximity threshold of a dynamic object.

6. The object detection system according to claim 5, wherein, the proximity threshold is dynamically determined based on a probability function associated with each of the N clusters, wherein the probability function of the k-th cluster is determined as a function of points of the k-th cluster.

7. The object detection system according to claim 5, wherein, the proximity threshold is less than 0.25m.

8. An object detection system, comprising: a signal processor operably coupled to a signal transmitting and receiving device, wherein the signal processor is configured to receive signals from the signal transmitting and receiving device and generate a multi-dimensional point cloud, and identify N clusters (C_1, C_2, ..., C_i, ..., C_N) within the multi-dimensional point cloud based on a clustering algorithm, wherein N is an integer greater than or equal to 0, and for each i-th cluster of the N clusters, i<N, the i-th cluster comprises M_i points (P^i_1, P^i_2, ..., P^i_j, ..., P^i_M); and a clustering processing unit operably coupled to the signal processor, the clustering processing unit comprising: a cluster boundary engine configured to associate a boundary with each of the N clusters, wherein the boundary of a k-th cluster is determined as a function of points (P^k_1, P^k_2, ..., P^k_j, ..., P^k_M) of the k-th cluster, k<N; and a cluster merging module operably coupled to the cluster boundary engine, wherein the cluster merging module is configured to, when any two or more clusters of the N clusters are associated with an overlapping space within the corresponding boundaries, generate a detection result comprising detected object information by merging clusters of the N clusters, such that the cluster merging module reduces falsely identified objects caused by over-clustering of a single object.

9. The object detection system according to claim 8, wherein, The signal transmitting and receiving device comprises an FMCW radar.

10. The object detection system according to claim 8, wherein, The clustering algorithm comprises a DBSCAN algorithm, wherein the signal processor executes the DBSCAN algorithm using a distance parameter, and the distance parameter is smaller than a predetermined gap between points received from the single object comprising a plurality of reflecting surfaces.

11. The object detection system according to claim 8, wherein, The cluster boundary engine determines the boundary of each of the N clusters as a statistical distribution function of points of the corresponding cluster among the N clusters, wherein the statistical distribution function is calculated independently in each dimension of the multi-dimensional point cloud.

12. The object detection system according to claim 8, wherein, The cluster boundary engine determines the boundary of each of the N clusters as a predetermined multiple of the standard deviation of points of the corresponding cluster on each dimension of the multi-dimensional point cloud.

13. The object detection system according to claim 12, wherein, The predetermined multiple is 4.

5.

14. The object detection system according to claim 8, wherein, The signals received from the signal transmitting and receiving device comprise signals reflected from a dynamic target, and the multi-dimensional point cloud comprises dynamic points.

15. The object detection system according to claim 8, further comprising a static target suppression module, and wherein: the object detection system is configured to detect static objects and dynamic objects independently, wherein detecting an object comprises identifying a position of the detected object, and the static target suppression module is configured to compare the static objects and the dynamic objects, and remove the static objects when the static objects are within a proximity threshold of the dynamic objects.

16. The object detection system according to claim 15, wherein, the proximity threshold is dynamically determined based on a probability function associated with each of the N clusters, wherein the probability function of the k-th cluster is determined as a function of points of the k-th cluster.

17. The object detection system according to claim 15, wherein, The proximity threshold is less than 0.25m.

18. An object detection method, comprising: receiving signals from a field of view; identifying N initial clusters within a multi-dimensional point cloud based on the signals received from the field of view, wherein each i-th initial cluster of the N initial clusters, i<N, comprises M_i points (P^i_1, P^i_2, ……, P^i_j, ……, P^i_M); determining and associating a statistical cluster boundary for each of the N initial clusters, wherein the statistical cluster boundary of the k-th cluster is determined as a function of the points (P^k_1, P^k_2, ……, P^k_j, ……, P^k_M) of the k-th cluster, k<N; and when any two or more clusters among the N initial clusters are associated with overlapping spaces contained within the corresponding cluster boundaries, generating a detection result comprising detected object information by merging the N initial clusters, such that misidentified objects caused by over-clustering of a single object are reduced.

19. The object detection method according to claim 18, wherein, determining the statistical cluster boundary comprises: determining a statistical distribution function of points of a corresponding one of the N initial clusters, wherein the statistical distribution function is calculated independently in each dimension of the multi-dimensional point cloud.

20. The object detection method according to claim 19, wherein, the statistical distribution function comprises a predetermined multiple of the standard deviation of points of the corresponding cluster.

21. A dynamic target measurement system, comprising: a first data storage repository comprising an instruction program; and a processor operably coupled to the first data storage repository, such that when the processor executes the instruction program, the processor causes operations to be performed to automatically generate multi-dimensional precision measurement results of a slowly moving object among static objects, the operations comprising: receiving measurement signals from a direction and ranging measurement device; generating a first component of the measurement signal; saving the first component to a second data storage repository; generating a second component of the measurement signal by subtracting the first component from the measurement signal; generating a first Fast Fourier Transform (FFT) representation of the second component; applying a threshold to the first FFT representation to determine whether a slow target exists; and when a slow target exists according to the first FFT representation, retrieving the first component from the second data storage repository, aggregating the first component and the second component to generate an aggregated measurement signal, generating a second FFT representation of the aggregated measurement signal, applying an interpolation operation to the second FFT representation, and measuring the speed of the slow target based on the second FFT representation.

22. The dynamic target measurement system according to claim 21, wherein, the direction and ranging measurement device comprises a frequency modulated continuous wave radar operably connected to the processor.

23. The dynamic target measurement system according to claim 21, wherein, the measurement signal comprises a Doppler chirp signal.

24. The dynamic target measurement system according to claim 21, wherein, the first component of the measurement signal comprises a DC offset extracted from the measurement signal.

25. The dynamic target measurement system according to claim 21, wherein, the first FFT representation and the second FFT representation comprise Doppler FFT representations.

26. The dynamic target measurement system according to claim 21, wherein, applying a threshold to the first FFT representation to determine whether a slow target exists according to the first FFT representation comprises: retrieving a bin threshold from a third data storage repository; retrieving a magnitude threshold from a fourth data storage repository; Identify peak values ​​in the first FFT representation, wherein the peak value is identified as the modulus of the corresponding FFT bin of the first FFT representation that is greater than the modulus threshold; Determine the warehouse number of the corresponding FFT warehouse for the identified peak value; Compare the warehouse number with the warehouse threshold; and When the warehouse number is less than the warehouse threshold, a signal is generated indicating that the slow target has been identified.

27. The dynamic target measurement system according to claim 26, wherein, The modulus threshold is dynamically generated using the constant false alarm rate (CFAR) method as a function of the measured signal.

28. A computer-implemented method, executed by at least one processor, for automatically generating multidimensional precision measurements of a slow-moving object within a static object, the method comprising: Generate the first component of the measurement signal received from the direction and distance measuring device; Store the first component in the first data repository; The second component of the measurement signal is generated by subtracting the first component from the measurement signal; Generate the first Fast Fourier Transform (FFT) representation of the second component; The presence of a slow target in the first FFT representation is determined based on the identification of the presence of peaks in the small FFT bins of the first FFT representation. as well as, When the first FFT indicates the existence of a slow target... Retrieve the first component from the first data repository. The first component and the second component are aggregated to generate an aggregated measurement signal. Generate a second FFT representation of the aggregated measurement signal. Apply interpolation to the second FFT representation, and, The velocity of the slow target is measured based on the second FFT representation, thereby reducing the interpolation error in the small FFT bin.

29. The computer-implemented method according to claim 28, wherein, The measurement signal includes a Doppler chirp signal.

30. The computer-implemented method according to claim 28, wherein, The first component of the measurement signal includes a DC offset extracted from the measurement signal.

31. The computer-implemented method according to claim 28, wherein, The first FFT representation and the second FFT representation include the Doppler FFT representation.

32. The computer-implemented method according to claim 28, wherein, Determining whether a slow target exists based on the first FFT representation includes: Retrieve warehouse thresholds from the second data repository; Retrieve the modulus threshold from the third data repository; Identify peak values ​​in the first FFT representation, wherein the peak value is identified as the modulus of the corresponding FFT bin of the first FFT representation that is greater than the modulus threshold; Determine the warehouse number of the corresponding FFT warehouse for the identified peak value; Compare the warehouse number with the warehouse threshold; and When the warehouse number is less than the warehouse threshold, a signal is generated indicating that the slow target has been identified.

33. The computer-implemented method according to claim 32, wherein, The modulus threshold is dynamically generated using the constant false alarm rate (CFAR) method as a function of the measured signal.

34. A computer program product comprising an instruction program tangibly embodied on a non-transitory computer-readable medium, wherein, When the instructions are executed on the processor, the processor causes a detection operation to be performed to automatically generate a multidimensional precision measurement result of a slow-moving object in a static object, the operation including: Receive measurement signals from direction and distance measuring equipment; Generate the first component of the measurement signal; Store the first component in the first data repository; The second component of the measurement signal is generated by subtracting the first component from the measurement signal; Generate the first Fast Fourier Transform (FFT) representation of the second component; The presence of a slow target in the first FFT representation is determined based on the presence of a peak in the small FFT bins of the first FFT representation; and... When the first FFT indicates the existence of a slow target... Retrieve the first component from the first data repository. The first component and the second component are aggregated to generate an aggregated measurement signal. Generate a second FFT representation of the aggregated measurement signal. Apply interpolation to the second FFT representation, and, The velocity of the slow target is measured based on the second FFT representation, thereby reducing the interpolation error in the small FFT bin.

35. The computer program product according to claim 34, wherein, The measurement signal includes a Doppler chirp signal.

36. The computer program product according to claim 34, wherein, The first component of the measurement signal includes a DC offset extracted from the measurement signal.

37. The computer program product according to claim 34, wherein, The first FFT representation and the second FFT representation include Doppler Fast Fourier Transform representations.

38. The computer program product according to claim 34, wherein, Determining whether a slow target exists based on the first Fast Fourier Transform representation includes: Retrieve warehouse thresholds from the second data repository; Retrieve the modulus threshold from the third data repository; Identify peak values ​​in the first FFT representation, wherein the peak value is identified as the modulus of the corresponding FFT bin of the first FFT representation that is greater than the modulus threshold; Determine the warehouse number of the corresponding FFT warehouse for the identified peak value; Compare the warehouse number with the warehouse threshold; and When the warehouse number is less than the warehouse threshold, a signal is generated indicating that the slow target has been identified.

39. The computer program product according to claim 38, wherein, The modulus threshold is dynamically generated using the constant false alarm rate (CFAR) method as a function of the measured signal.

40. The computer program product according to claim 34, wherein, The interpolation operation includes applying multiple interpolation techniques to the second FFT representation, thereby improving the linearity and accuracy of the velocity measurements of the slow target used to obtain precise measurements in each Doppler FFT chamber.

41. A system comprising: The first data repository includes instruction programs; and, A processor, operatively coupled to the first data repository, such that when the processor executes the instruction program, the processor causes operations to be performed to automatically perform multi-stage precision measurements of the target, the operations including: A first spectral energy representation of a first set of object detection signals is generated based on a first calculation method, wherein the first set of object detection signals is received from the direction and ranging device corresponding to the field of view of the direction and ranging device; Retrieve the amplitude threshold from the second data repository; The amplitude threshold is applied to the first spectral energy representation; Determine the region of interest within the first spectral energy representation, wherein the region of interest includes the frequency range within the first spectral energy representation whose amplitude is greater than the amplitude threshold; A second spectral energy representation of a second set of object detection signals is generated based on a second computational method, wherein the second set of object detection signals includes a subset of the first set of object detection signals corresponding to the region of interest, and wherein the second computational method has a higher computational cost and higher resolution than the first computational method; and... The object detection measurement result is generated based on the second spectral energy representation, wherein the object detection measurement result includes multiple objects identified in the first set of object detection signals, wherein the object is identified when both the first spectral energy representation and the second spectral energy representation include peaks corresponding to the object, thereby reducing the false object detection rate to a level lower than the first false object detection rate of the first calculation method and the second false object detection rate of the second calculation method.

42. The system according to claim 41, wherein, The frequency range of the region of interest is dynamically determined and includes a continuous range.

43. The system according to claim 41, wherein, The amplitude threshold includes dynamically determined thresholds.

44. The system according to claim 43, wherein, The dynamically determined threshold is based on the noise level of the detection signals of the first group of objects.

45. The system according to claim 43, wherein, The dynamically determined threshold is based on the application of a 2D constant false alarm rate algorithm to the overall detection signals of the first group of objects.

46. ​​The system according to claim 41, wherein, The first set of object detection signals is received from N antennas, and the second set of object detection signals includes signals received from a subset of the N antennas, thereby mitigating erroneous detection locations related to the number of effective antennas generated by the first calculation method and the second calculation method.

47. A computer-implemented method, executed by at least one processor, for automatically performing a multi-stage precision measurement of a target, the method comprising: A first spectral energy representation of a first set of object detection signals is generated based on a first calculation method, wherein the first set of object detection signals is received from the direction and ranging device corresponding to the field of view of the direction and ranging device; Retrieve the amplitude threshold from the second data repository; The amplitude threshold is applied to the first spectral energy representation; Determine the region of interest within the first spectral energy representation, wherein the region of interest includes the frequency range within the first spectral energy representation whose amplitude is greater than the amplitude threshold; A second spectral energy representation of the second set of object detection signals is generated based on a second computational method, wherein the second set of object detection signals includes a subset of the first set of object detection signals corresponding to the region of interest, and wherein the second computational method has a higher computational cost and higher resolution than the first computational method; and The object detection measurement results are generated based on the second spectral energy representation.

48. The computer-implemented method according to claim 47, wherein, Generating the object detection measurement result includes: identifying multiple objects within the first group of object detection signals, wherein when both the first spectral energy representation and the second spectral energy representation include peaks corresponding to the object, the object is identified, thereby reducing the false object detection rate to below the first false object detection rate of the first calculation method and the second false object detection rate of the second calculation method.

49. The computer-implemented method according to claim 47, wherein, The frequency range of the region of interest is dynamically determined and includes a continuous range.

50. The computer-implemented method according to claim 47, wherein, The amplitude threshold includes dynamically determined thresholds.

51. The computer-implemented method according to claim 50, wherein, The dynamically determined threshold is based on the noise level of the detection signals of the first group of objects.

52. The computer-implemented method according to claim 50, wherein, The dynamically determined threshold is based on the application of a 2D constant false alarm rate algorithm to the overall detection signals of the first group of objects.

53. The computer-implemented method according to claim 47, wherein, The first set of object detection signals is received from N antennas, and the second set of object detection signals includes signals received from a subset of the N antennas, thereby mitigating erroneous detection results related to the number of effective antennas generated by the first calculation method and the second calculation method.

54. A computer program product comprising an instruction program tangibly embodied on a non-transitory computer-readable medium, wherein, When the instructions are executed on the processor, the processor causes measurement and detection operations to be performed to automatically perform multi-stage precision measurements of the target, the operations including: A first spectral energy representation of a first set of object detection signals is generated based on a first calculation method, wherein the first set of object detection signals is received from the direction and ranging device corresponding to the field of view of the direction and ranging device; Retrieve the amplitude threshold from the second data repository; The amplitude threshold is applied to the first spectral energy representation; Determine the region of interest within the first spectral energy representation, wherein the region of interest includes the frequency range within the first spectral energy representation whose amplitude is greater than the amplitude threshold; A second spectral energy representation of the second set of object detection signals is generated based on a second computational method, wherein the second set of object detection signals includes a subset of the first set of object detection signals corresponding to the region of interest, and wherein the second computational method has a higher computational cost and higher resolution than the first computational method; and The object detection measurement results are generated based on the second spectral energy representation.

55. The computer program product according to claim 54, wherein, Generating the object detection measurement result includes: identifying multiple objects within the first group of object detection signals, wherein when both the first spectral energy representation and the second spectral energy representation include peaks corresponding to the object, the object is identified, thereby reducing the false object detection rate to below the first false object detection rate of the first calculation method and the second false object detection rate of the second calculation method.

56. The computer program product according to claim 54, wherein, The frequency range of the region of interest is dynamically determined and includes a continuous range.

57. The computer program product according to claim 54, wherein, The amplitude threshold includes dynamically determined thresholds.

58. The computer program product according to claim 57, wherein, The dynamically determined threshold is based on the noise level of the detection signals of the first group of objects.

59. The computer program product according to claim 57, wherein, The dynamically determined threshold is based on the application of a 2D constant false alarm rate algorithm to the overall detection signals of the first group of objects.

60. The computer program product according to claim 54, wherein, The first set of object detection signals is received from N antennas, and the second set of detection signals includes signals received from a subset of the N antennas, thereby mitigating erroneous detection signals related to the effective number of antennas generated by the first calculation method and the second calculation method.

Images