Human target tracking system and method

Abstract

A human target tracking system and method are disclosed. In one embodiment, a method for tracking a human target includes: receiving a radar signal using a radar sensor; generating a range-Doppler map based on the received radar signal; detecting the human target based on the range-Doppler map, wherein detecting the human target includes determining a distance and a bounding box size defining a bounding box of the detected human target, the bounding box at least partially surrounding the detected human target; and adding a new detection point including the determined distance and the bounding box size to the track when the determined distance is within an expected area associated with the track, wherein the expected area is determined based on the bounding box size of the detection point of the track.

Description

Technical Field

[0001] The present disclosure relates generally to electronic systems and methods, and in particular embodiments, to human target tracking systems and methods. Background Art

[0002] Applications in the millimeter-wave frequency region have attracted significant interest over the past few years due to the rapid development of low-cost semiconductor technologies such as silicon-germanium (SiGe) and fine-geometry complementary metal-oxide-semiconductor (CMOS) processes. The availability of high-speed bipolar and metal-oxide-semiconductor (MOS) transistors has led to a growing demand for integrated circuits for millimeter-wave applications at 24 GHz, 60 GHz, 77 GHz, 80 GHz, and above 100 GHz. These applications include, for example, static and moving target detection and tracking.

[0003] In some radar systems, the distance between the radar and the target is determined by transmitting a frequency modulated signal, receiving the reflection of the frequency modulated signal (also called an echo), and determining the distance based on the time delay and / or frequency difference between the transmission and reception of the frequency modulated signal. Therefore, some radar systems include a transmitting antenna that transmits a radio frequency (RF) signal, a receiving antenna that receives the RF, and associated RF circuits for generating the transmit signal and receiving the RF signal. In some cases, multiple antennas can be used to achieve directional beams using phased array technology. Multiple-input multiple-output (MIMO) configurations with multiple chipsets can also be used to perform coherent and non-coherent signal processing.

[0004] In some settings, static objects coexist with moving objects. For example, in indoor environments, static objects, such as furniture and walls, coexist with moving objects, such as humans. Indoor settings can also include objects that exhibit periodic motion, such as fans. Doppler analysis has been used to distinguish between moving and static objects. Summary of the Invention

[0005] According to an embodiment, a method for tracking a human target includes: receiving a radar signal using a radar sensor; generating a range-Doppler map based on the received radar signal; detecting the human target based on the range-Doppler map, wherein detecting the human target includes determining a distance and a bounding box size of a bounding box defining the detected human target, the bounding box at least partially surrounding the detected human target; and when the determined distance is within an expected area associated with a track, adding a new detection point including the determined distance and the bounding box size to the track, wherein the expected area is determined based on the bounding box size of the detection point of the track.

[0006] According to an embodiment, a millimeter-wave radar system includes: a radar sensor; and a processor. The processor is configured to: transmit a radar signal using the radar sensor; receive a reflected radar signal using the radar sensor; generate a range-Doppler map based on the received reflected radar signal; and detect a human target based on the range-Doppler map. Detecting the human target includes determining a distance and a bounding box size of a bounding box of the detected human target, the bounding box at least partially surrounding the detected human target. When the determined distance is within an expected area associated with a track, adding a new detection point including the determined distance and the bounding box size to the track, wherein the expected area is determined based on the bounding box size of the detection point in the track.

[0007] According to an embodiment, a method for simultaneously tracking multiple human targets includes: receiving a radar signal using a radar sensor; generating a range-Doppler map based on the received radar signal; detecting human targets based on the range-Doppler map, wherein detecting human targets includes: determining a distance; generating a first expected area based on a first track associated with a first human target; generating a second expected area based on a second track associated with a second human target; when the determined distance is within the first expected area and the second expected area, adding a new detection point including the determined distance to the first track or the second track using a nearest neighbor algorithm; and when the determined distance is within the first expected area and outside the second expected area, or when the determined distance is within the second expected area and outside the first expected area, adding a new detection point including the determined distance to the first track or the second track using a probabilistic data association filter (PDAF). BRIEF DESCRIPTION OF THE DRAWINGS

[0008] For a more complete understanding of the present invention and its advantages, reference is now made to the following description taken in conjunction with the accompanying drawings, in which:

[0009] Figure 1 A radar system according to an embodiment of the present invention is shown;

[0010] Figure 2 shows a range-Doppler diagram of a moving human body captured by a millimeter wave radar system according to an embodiment of the present invention;

[0011] Figure 3A and Figure 3B shows fast time versus slow time plots of range-Doppler responses of a human target after preprocessing and after target detection and clustering according to an embodiment of the present invention, respectively;

[0012] Figure 4A and Figure 4B respectively show the corresponding bounding boxes according to an embodiment of the present invention. Figure 3A and Figure 3B Fast time versus slow time plot of the range-Doppler response of a human target;

[0013] Figure 5 shows a trajectory of a position of a human target according to an embodiment of the present invention;

[0014] Figure 6 shows the signal strength of a standing human target according to an embodiment of the present invention;

[0015] Figure 7A shows a trajectory of a position of a human target according to an embodiment of the present invention;

[0016] Figure 7B A flow chart of an exemplary method for determining whether to maintain the existence of a trajectory of a human target according to an embodiment of the present invention is shown;

[0017] Figure 8A shows a trajectory of a position of a human target according to an embodiment of the present invention;

[0018] Figure 8B A flow chart of an embodiment method for determining whether to maintain tracking presence of a human target according to an embodiment of the present invention is shown;

[0019] Figure 9A shows a range spectrum of a human target moving in a zigzag manner in an indoor environment surrounded by walls according to an embodiment of the present invention;

[0020] Figure 9B shows three bounding boxes for three different trajectories, respectively, according to an embodiment of the present invention;

[0021] Figure 10 A flow chart of an exemplary method for removing ghost tracks according to an embodiment of the present invention is shown;

[0022] Figure 11A and Figure 11B Two trajectories of simultaneously tracking two human targets according to an embodiment of the present invention are shown;

[0023] Figure 12 Two trajectories of simultaneously tracking two human targets according to an embodiment of the present invention are shown;

[0024] Figure 13 A flow chart of an exemplary method for simultaneously tracking multiple human targets according to an embodiment of the present invention is shown.

[0025] Figure 14 and Figure 15 A flow chart of an exemplary method for detecting and tracking a human target according to an embodiment of the present invention is shown;

[0026] Figure 161 shows the detection result of two people walking across each other in an indoor environment according to an embodiment of the present invention;

[0027] Figure 17 The embodiment of the present invention is shown Figure 16 The tracking results of two people;

[0028] Figure 18 shows detection results of a person walking in a zigzag manner in an indoor environment according to an embodiment of the present invention; and

[0029] Figure 19 The embodiment of the present invention is shown Figure 18 The tracking results of people.

[0030] Corresponding numerals and symbols in the different figures generally refer to corresponding parts unless otherwise indicated.The drawings are drawn to clearly illustrate the relevant aspects of the preferred embodiments and are not necessarily drawn to scale. DETAILED DESCRIPTION

[0031] The making and using of the disclosed embodiments are discussed in detail below. However, it should be understood that the present invention provides many applicable inventive concepts that can be embodied in a wide variety of specific contexts. The specific embodiments discussed are merely illustrative of specific ways to make and use the invention and do not limit the scope of the invention.

[0032] The following description shows various specific details to provide an in-depth understanding of several example embodiments according to the description. An embodiment may be obtained without one or more of the specific details, or may be obtained using other methods, components, materials, etc. In other cases, known structures, materials, or operations are not shown or described in detail so as not to obscure different aspects of the embodiment. References to "embodiments" in this description indicate that a particular configuration, structure, or feature described with respect to the embodiment is included in at least one embodiment. Therefore, phrases such as "in one embodiment" that may appear at different points in this description do not necessarily refer to the same embodiment. In addition, in one or more embodiments, specific configurations, structures, or features may be combined in any appropriate manner.

[0033] Embodiments of the present invention will be described in the specific context of systems and methods for detecting and tracking human targets in indoor environments. Embodiments of the present invention can also be used to detect and track other moving objects, such as animals and robots. Some embodiments can be used outdoors.

[0034] In an embodiment of the present invention, a millimeter wave radar is used to detect and track human targets in an indoor environment. After the initial detection of the target and the corresponding clustering, a bounding box is determined to surround the human target, such as the torso, hands, and feet of the detected human target. The size of the bounding box is tracked together with the angle of arrival of each human target, the position and velocity of the center of mass of each human target, to estimate the control region associated with the tracking. Only targets detected within the tracked control region are used to update the tracking. In some embodiments, multiple human targets are tracked simultaneously using their respective trajectories.

[0035] Radar (e.g., millimeter wave radar) can be used to detect and track human bodies. For example, Figure 1 A radar system 100 according to an embodiment of the present invention is shown. The radar system 100 includes a millimeter wave radar 102 and a processor 104. In some embodiments, the millimeter wave radar 102 includes the processor 104.

[0036] During normal operation, the millimeter wave radar 102 transmits a plurality of radiation pulses 106, such as chirps, toward a scene 108. In some embodiments, the chirps are linear chirps (ie, the instantaneous frequency of the chirps varies linearly with time).

[0037] The emitted radiation pulse 106 is reflected by an object in the scene 108. The reflected radiation pulse ( Figure 1 The echo signal (not shown) is also called an echo signal, which is detected by the millimeter wave radar 102 and processed by the processor 104 to, for example, detect and track a human body.

[0038] Objects in scene 108 may include stationary persons, such as lying person 110 (exhibiting low and infrequent motion), standing person 112, and moving persons, such as running or walking persons 114 and 116. Objects in scene 108 may also include static objects, such as furniture and periodic motion equipment (not shown). Other objects may also be present in scene 108.

[0039] The processor 104 uses signal processing techniques to analyze the echo data to determine the position of the human body. For example, in some embodiments, a range FFT is used to estimate the range component of the detected human body's position (e.g., relative to the position of the millimeter-wave radar). An angle estimation technique can be used to determine the azimuth component of the detected human body's position.

[0040] In some embodiments, a range-Doppler map (image) is generated from the echo data, and two-dimensional (2D) moving target identification (MTI) is performed on the range-Doppler map to detect moving targets.

[0041] The processor 104 can be implemented as a general-purpose processor, a controller or a digital signal processor (DSP), and the processor, for example, includes a combination circuit coupled to a memory. In some embodiments, the DSP, for example, can be implemented using an ARM architecture. In some embodiments, the processor 104 can be implemented as a customized application-specific integrated circuit (ASIC). In some embodiments, the processor 104 includes a plurality of processors, each of which has one or more processing cores. In other embodiments, the processor 104 includes a single processor with one or more processing cores. Other implementations are also feasible. Some embodiments can be implemented as a combination of a hardware accelerator and software running on a DSP or a general-purpose microcontroller.

[0042] Millimeter-wave radar 102 operates as an FMCW radar and includes millimeter-wave radar sensor circuitry, a transmit antenna, and at least two receive antennas. Millimeter-wave radar 102 transmits and receives signals in the range of 20 GHz to 122 GHz. Alternatively, frequencies outside this range, such as between 1 GHz and 20 GHz or between 122 GHz and 300 GHz, may be used.

[0043] In some embodiments, the echo signal received by the receiving antenna of the millimeter-wave radar 102 is filtered and amplified using a bandpass filter (BPF), a low-pass filter (LPF), a mixer, a low-noise amplifier (LNA), and an intermediate frequency (IF) amplifier in a manner known in the art. The echo signal is then digitized using one or more analog-to-digital converters for further processing. Other implementations are also possible.

[0044] For various reasons, it may be necessary to detect and track human targets in indoor environments. Traditional target tracking methods assume that the target is a single point in the range-Doppler image. In a traditional range-Doppler processing chain, the cluster of detections obtained is used to obtain a single bin in the range-Doppler image to determine the range and Doppler components of the detected target. This single bin is then fed to a tracker to track the target. For example, in conventional radar signal processing, range, Doppler, and angle of arrival measurements may be performed on a single point target. These components are then fed to a tracker for tracking purposes.

[0045] The motion model of the traditional tracker can be expressed as

[0046]

[0047] where k represents the discrete time step, Δt is the time between each time step, px is the position of the target in the x-direction, py is the position of the target in the y-direction, vx is the velocity of the target in the x-direction, and vy is the velocity of the target in the y-direction.

[0048] In some radar systems, such as millimeter-wave radar systems, a human target may exhibit a dual spread across range and Doppler bins when reflections are received from different parts of the human body during movement of the human target. For example, Figure 2 FIG. 2 shows a range-Doppler diagram of a moving human body captured by a millimeter wave radar system according to an embodiment of the present invention. Figure 2 As shown, the human target may exhibit peaks at different locations in the range-Doppler graph corresponding to different parts of the human target's body, such as the right foot 204 , the left foot 202 , and the torso and hands 206 .

[0049] Human targets can show spread in Doppler as well as over range bins in the range Doppler plot. For example, Figure 3A and Figure 3B Figure 1 shows fast-time versus slow-time plots of the range-Doppler response of a human target after preprocessing and after target detection and clustering, respectively, according to an embodiment of the present invention. This extension contains information about the micro-motion of the human target. Therefore, this extension can be used to improve the performance of subsequent trackers and can also be used to track micro-motion of human targets.

[0050] In one embodiment, a bounding box encloses the Doppler and range spread of a human target. The perimeter and area of ​​the bounding box are dynamically adjusted to track changes in the Doppler and range spread of the human target. One or more dimensions of the bounding box are determined and tracked to advantageously enhance detection and tracking of the human target.

[0051] A bounding box (e.g., a rectangular box) may encompass the Doppler and range extensions of a target. For example, according to an embodiment of the present invention, Figure 4A and Figure 4B The corresponding bounding boxes are shown Figure 3A and Figure 3B The fast time versus slow time diagram of the range-Doppler response of a human target. Figure 4A and Figure 4B As shown, the bounding box has a distance dimension L r and Doppler size L d , the distance dimension L r The Doppler size L extends from the minimum distance of the human target to the maximum distance. d The Doppler spread of a human target extends from the minimum Doppler velocity to the maximum Doppler velocity.

[0052] In some embodiments, the bounding box may have a shape other than a rectangle, such as a circle, an ellipse, or an arbitrary shape. For example, in some embodiments, the bounding box is a circle having a center at the centroid of the range-Doppler spread of the human target and a radius equal to the minimum radius of the range-Doppler spread that encloses the human target.

[0053] In some embodiments, one or more bounding box dimensions are determined and tracked to enhance human target detection and tracking. For example, in some embodiments, the coordinated steering motion model of a bounding box with a zero rate of change of the target's angle of arrival ω(k) may be given by

[0054]

[0055] Where k represents the discrete time step, Δt is the time between each time step, px is the position of the center of mass of the human target in the x direction, py is the position of the center of mass of the human target in the y direction, lx is the size of the bounding box in the x direction, ly is the size of the bounding box in the y direction, vlx is the rate of change of the bounding box x size, vly is the rate of change of the bounding box y size, and v c is the radial velocity of the center of mass of the human target, θ is the angle of arrival of the human target, and ω is the rate of change of the angle of arrival (AoA) of the human target.

[0056] In one embodiment, the parameters px, py, lx, ly, vlx, vly, v c , θ, and ω represent the states of a tracker (e.g., implemented using a Kalman filter such as an unscented Kalman filter) that tracks the human target using a trajectory. These states can be obtained from measurements of the millimeter wave radar system 100. For example, in one embodiment, the millimeter wave radar system 100 measures r (the distance of the target from the millimeter wave radar sensor), θ (the angle of arrival of the target), and v c (radial velocity of the target), L r (bounding box size across distance) and L d (Bounding box size across Dopplers). These measurements can be collectively referred to as

[0057] Zmeas=(r,θ,v c ,L r ,L d ) 3(3)

[0058] The Zmeas measurement can be converted to the tracker state by

[0059]

[0060] where ω, vlx, and vly are initialized to zero.

[0061] At each time step, a new set of measurements Zmeas is collected, and the trajectory is updated based on such new set of measurements Zmeas. For example, an unscented Kalman filter computes the predicted states px, py, lx, ly, vlx, vly, vc, θ, and ω for each time step (of the trajectory) (e.g., using Equation 2). Due to the Bayesian recursive approach, these states can contain information from all measurements available from time 1:k onwards - all measurements of the trajectory). Such predicted states can be converted into the form of predicted measurements by

[0062]

[0063] The Mahalanobis distance can be calculated between the predicted measurement Zpred and the new set of measurements Zmeas by the following formula

[0064]

[0065] where S is the covariance matrix between Zmeas and Zpred.

[0066] In some embodiments, if the distance Md is below a predetermined threshold, the new set of measurements Zmeas is considered to be valid measurements of the target tracked by the trajectory. The multidimensional region where the distance Md is below the predetermined threshold is called the control region or expected region associated with the trajectory.

[0067] When the rate of change of the arrival angle of the human target ω(k) is not equal to zero, the coordinated steering motion model of the bounding box can be given by

[0068]

[0069] The coordinated steering motion model including noise parameters can be given by

[0070]

[0071] where a p (k) represents the acceleration of the target's center of mass, a l (k) represents the acceleration of the bounding box size, a θ (k) represents the acceleration of the arrival angle, where

[0072]

[0073] As shown in Equations 2 through 9, the coordinated steering motion model tracks the angle of arrival and the size of the bounding box of the human target, as well as the position and velocity of the center of mass of the human target. Using a coordinated steering motion model such as that shown in Equations 2 through 9 advantageously allows for accurate tracking of human targets that move in a human manner (e.g., accelerate, decelerate, turn, change direction, extend arms and / or legs, exhibit random motion, etc.). For example, Figure 5 A trajectory 504 of the position of a human target 502 is shown according to an embodiment of the present invention.

[0074] Trajectory 504 includes detection points 508, 510, 512, and 514. Each detection point directly or indirectly includes parameter values ​​such as distance, angle of arrival, velocity, and bounding box size. For example, with respect to Equation 2, each detection point 508, 510, 512, and 514 may correspond to k values ​​0, 1, 2, and 3, and each detection point may include px, py, lx, ly, vlx, vly, and v c , θ and ω, which can be used to calculate r, θ, v c 、L r and L d The corresponding value of . Figure 5 As shown, control region 506 encompasses an area of ​​possible positions of human target 502 based on the history of trajectory 504. Control region 506 is generated based on the coordinated steering motion model, such as shown in Equations 2 through 9. For example, in some embodiments, the size of the control region can be determined based on Euclidean distance and / or Mahalanobis distance, which are calculated based on the coordinated steering motion model and the history of trajectory 504.

[0075] When determining the next position of human target 502, only targets detected within control region 506 are considered. Targets detected outside of control region 506 are ignored.

[0076] The determination of which of the targets detected within control area 506 corresponds to human target 502 (tracked by trajectory 504) is based on the history of trajectory 504. For example, in some embodiments, a data association algorithm such as nearest neighbor, Munkres algorithm, and / or probabilistic data association filter (PDAF) may be used. Other data association algorithms may also be used.

[0077] In some embodiments, a tracking algorithm (such as a Kalman filter) can be used to track the human target. Because the human target may exhibit random motion, some embodiments use an unscented Kalman filter to track the human target. The unscented Kalman filter, together with a coordinated steering model (e.g., as shown in Equations 2 to 9), advantageously allows accurate tracking of nonlinear motion of the human target (which may be difficult to track using Gaussian estimation). Other tracking algorithms may also be used.

[0078] When tracking a human target, the human target may not be detected in some cases. For example, the human target may not be detected because it moves to a position outside the radar field of view. Another reason for failure to detect a human target is that the signal strength of the signal reflected from the human target may fluctuate over time and in some cases be weaker than the detection threshold. For example, Figure 6 The signal strength of a standing human target according to an embodiment of the present invention is shown.

[0079] As shown in curve 602, even a standing human body will show a signal strength variation over time. In some embodiments, although the human target is not detected in one or more instances, the tracking of the human target still exists. For example, Figure 7A A trajectory 704 of the position of a human target 702 is shown according to an embodiment of the present invention.

[0080] like Figure 7A As shown, human target 702 was not detected in two instances (no target was detected within control region 712 and control region 714). When human target 702 was not detected, the tracking algorithm used predicted positions 706 and 708 based on the history of trajectory 704 rather than actual measurements to maintain the existence of trajectory 704. Once a target falls within the control region at the predicted position (e.g., target 710 is detected within control region 716 based on estimated position 708 of human target 702 and the history of trajectory 704), the target can be associated with trajectory 704.

[0081] Maintaining the track for multiple missed detections advantageously avoids losing the track each time a missed detection occurs, and thus helps to cope with intermittent missed detections from human targets. In some embodiments, the track remains in existence for a predetermined number of detection attempts. For example, in some embodiments, the track remains in existence for 10 unsuccessful detection attempts. Some embodiments may maintain the track in existence for a larger number of unsuccessful detection attempts (e.g., 15, 18, 24, 50, or more). Other embodiments may maintain the track in existence for a smaller number of unsuccessful detection attempts (e.g., 8, 6, or less).

[0082] In some embodiments, the track can be kept alive by using a counter. For example, in one embodiment, when the track fails to detect the corresponding human target, a countdown counter is started. When a successful detection is assigned to the track, the countdown counter is reset. If the counter reaches zero (in other words, no successful detection has been generated since the countdown counter was started), the track is terminated. In some embodiments, the countdown counter takes several seconds (e.g., 3 seconds, 15 seconds, 20 seconds or more) to reach zero. Some embodiments can use a counter that counts up, wherein the track is terminated when the counter reaches a predetermined threshold. Other methods for counting the time when successful detection is not achieved can be used.

[0083] Figure 7B 1 is a flow chart showing an exemplary method 750 for determining whether to maintain tracking of a human target according to an embodiment of the present invention. The method 750 may be implemented, for example, by the millimeter wave radar system 100. The method 750 will refer to Figure 7A However, according to the embodiment of the present invention, it should be understood that Figure 7A Non-limiting examples of trajectories of positions of human targets are shown.

[0084] During step 752, a timer is started. The timer can be implemented by a digital counter that counts up or down. In some embodiments, the timer can be implemented by an analog circuit. The implementation of a timer is well known in the art.

[0085] During step 754, a control region is determined based on the history of trajectory 704. For example, using one or more of Equations 2 to 9, based on the time t 712 The measurement history of previously occurring trajectory 704 determines the control region 712 .

[0086] During step 756, the target is detected within the control area. During step 758, it is determined whether the target is detected within the control area. For example, at time t 718 , two targets are detected within the control area 718.

[0087] If a target is detected within the control area, the timer is reset during step 760 and the trajectory 704 is updated based on the detected target during step 762. If a target is not detected within the control area, the timer is checked during step 764 to determine if the timer has expired. If it is determined during step 764 that the timer has not expired, the trajectory 704 is updated based on an estimate during step 768, where the estimate can be determined using, for example, Equations 2 through 9. If it is determined during step 764 that the timer has expired, the trajectory 704 is terminated during step 766.

[0088] For example, expiration of a timer can be indicated when the timer reaches a predetermined threshold. For example, in some embodiments, the timer expires when a down counter reaches zero. In other embodiments, the timer expires when an up counter reaches a predetermined threshold. Other implementations are also possible.

[0089] It is possible that non-human targets may appear in the radar system's field of view. Such non-human targets may be, for example, other moving objects and / or ghost targets. In some embodiments, a track is terminated when only M or fewer detections are successful in the first N attempts to detect a human target. For example, in an embodiment where N is 8 and M is 4, a track is maintained only if more than 5 successful detections are achieved in the first 8 attempts to detect a human target. For example, Figure 8A Non-limiting examples of trajectories 804 and 814 of the positions of respective human targets 802 and 812 are shown, where N is 8 and M is 4, according to an embodiment of the present invention.

[0090] like Figure 8A As shown, track 804 is terminated when fewer than 5 successful detections are achieved in the first 8 detection attempts. Track 814 remains in existence because there are at least 5 successful detections in the first 8 detection attempts.

[0091] Figure 8B 850 is a flow chart of an embodiment of a method 850 for determining whether a trajectory of a human target is maintained according to an embodiment of the present invention. The method 850 may be implemented, for example, by the millimeter wave radar system 100. The method 850 will refer to Figure 8A However, according to the embodiment of the present invention, it should be understood that Figure 8A Non-limiting examples of trajectories of positions of human targets are shown.

[0092] During step 852, a track is created (e.g., when a target is detected in the field of view of the millimeter wave radar 102). Steps 754, 756, 758, 762, 766, and 768 are related to Figure 7B The described method is performed in a similar manner.

[0093] During step 854, the number of trajectory samples (detection points) since the trajectory was created is determined. If the number of trajectory samples since the trajectory was created is below a threshold value N, the number of missed detections for the trajectory is determined during step 856. If the number of missed detections is above a threshold value M, the trajectory is terminated during step 766. In some embodiments, N is equal to 8 and M is equal to 3. Other values ​​of M and N may also be used.

[0094] When detecting and tracking human targets, multipath reflections from static objects (such as walls and chairs) may appear as real targets along with the actual human targets. Such ghost targets exhibit characteristics similar to real human targets and, therefore, may be confused with human targets. For example, Figure 9A FIG. 9 shows a range spectrum 900 of a human target moving in a zigzag manner in an indoor environment surrounded by walls after 2D MTI filtering and coherent integration according to an embodiment of the present invention. Figure 9A As shown, over time, a human target may get closer or further away from the radar sensor. The top curve 902 corresponds to the actual human target. The bottom curves 904, 906, and 908 are reflections received by the radar.

[0095] like Figure 9A As shown in Figure 1, even after applying a moving target indication (MTI) filter to remove static objects, continuous multipath reflections can still be obtained from the wall. Such multipath reflections can be difficult to distinguish from real human targets because they exhibit Doppler signatures similar to those of human targets and can be difficult to remove by conventional moving target removers (such as MTI filters).

[0096] like Figure 9A As shown, the bottom curves 904, 906, and 908 corresponding to multipath reflections are highly correlated with the top curve 902 corresponding to the actual human target (direct path signal). In an embodiment of the present invention, when the correlation between the tracks is higher than a predetermined threshold (such as, for example, 0.6), the multipath reflections are removed. Among the tracks that show high correlation with each other, the tracks that are closer to the radar (such as Figure 9A ) remains, while other trajectories (such as the bottom curve) are terminated.

[0097] Figure 9B Three bounding boxes are shown for three different tracks, respectively, according to an embodiment of the present invention. In some embodiments, slow-time data (phase) is acquired for the range bin corresponding to each bounding box. In some embodiments, the discrete cosine transform (DCT) and phase of each bounding box are evaluated over multiple consecutive time snapshots (e.g., 5 time snapshots). If a high correlation is found between the tracks (e.g., above 0.6), the track closer to the radar is kept, while the track farther from the radar is terminated. For example, in one embodiment, the correlation factor may be given by

[0098]

[0099] in

[0100]

[0101] where RDImage is the range-Doppler image (e.g., after 2D MTI filtering and coherent integration, as Figure 9A ), R1 is the slow-time data of target 1, R2 is the slow-time data of target 2, r1 is the centroid distance bin of the bounding box of target 1, and r2 is the centroid distance bin of the bounding box of target 2. In one embodiment, if F is greater than 0.6, the track associated with the target with the minimum value of r1 and r2 remains, and the other track is terminated.

[0102] Figure 10 FIG. 1 is a flow chart of an exemplary method 1000 for removing ghost tracks according to an embodiment of the present invention. The method 1000 may be implemented, for example, by the millimeter wave radar system 100. Figure 9A and Figure 9B to illustrate method 1000. However, according to an embodiment of the present invention, it should be understood that Figure 9A and Figure 9B Non-limiting examples of trajectories are shown.

[0103] During step 1002, a correlation value is obtained between each pair of existing traces (e.g., using Equation 10). For example, if traces A, B, and C are present, correlations are obtained for pairs AB, BC, and AC. During step 1004, the correlation values ​​are compared to a predetermined correlation threshold. In some embodiments, the predetermined correlation threshold may be greater than 0.6 or 0.65. In some embodiments, the predetermined correlation threshold may be less than 0.59 or 0.58.

[0104] If the correlation value is below the predetermined correlation threshold, both tracks remain present during step 1006. However, if the correlation value is above the predetermined correlation threshold, the track farthest from the radar sensor is terminated.

[0105] As a non-limiting example, if Figure 9A If method 1000 is performed on the trace shown, only the top curve remains, while the bottom curve (which is a multipath reflection of the top curve), and therefore a ghost curve, is terminated.

[0106] In addition to the advantages of detecting and tracking the size of the bounding box, detecting and tracking the spread of Doppler and range of a human target advantageously allows the use of, for example, micro-Doppler features to detect micro-motions. For example, the breathing of a human target can be detected by extracting micro-Doppler features from the Doppler and range spread of a particular trajectory. Such micro-motions can be used to continue tracking a human target that has become idle (e.g., because the human target transitions to a sitting state). For example, in some embodiments, when the trajectory of a human target indicates that the corresponding human target has stopped moving (e.g., because the human target is idly sitting or standing), the breathing of such a human target (e.g., determined based on the micro-Doppler features extracted from the corresponding spread in Doppler and range) can be used to continue detecting and tracking such a static human target. For example, in some embodiments, in step 758 (in Figure 7B or Figure 8B ), when micro-Doppler signature data extracted from the spread of distance and Doppler within the area enclosed by the bounding box indicates the presence of a human target within the control area, the human target is detected, for example, by indicating a breathing rate within a normal human level and / or a heart rate within a normal human level.

[0107] By using micro-Doppler characteristics, it is possible to maintain the trajectory of a human target that has become stationary, and to continue tracking such a human target. Respiration is an example of a micro-motion that can be monitored using Doppler and range extension. Other micro-motions, such as heart rate, can also be used.

[0108] In some embodiments, multiple human targets are monitored and tracked simultaneously via corresponding trajectories. For example, Figure 11A and Figure 11B Trajectories 1104 and 1114 are shown for tracking human targets 1102 and 1112 , respectively, according to an embodiment of the present invention.

[0109] During normal operation, data association algorithms such as the nearest neighbor, Munkres algorithm, and / or PDAF can be used to associate human targets with trajectories based on predetermined criteria (e.g., based on characteristics of the trajectories, such as location history, velocity, shape, angle, etc.). For example, PDAF is a strong data association algorithm that works well in high-clutter scenarios. For probabilistic data association, a hypothesis is considered for each measurement of each trajectory and an update is performed corresponding to each hypothesis. Therefore, if there are, for example, 5 trajectories and 10 measurements, there are 50 hypotheses to consider for PDAF.

[0110] When more than two trajectories are close to or intersecting each other, the number of hypotheses to be considered increases, thereby increasing the computational complexity of assigning measurements to trajectories using PDAF.

[0111] In one embodiment of the present invention, when two or more tracks share the same measurement, a nearest neighbor approach is used where only the track with the lowest weight for the measurement is updated, while other tracks are updated without the measurement.

[0112] For example, reference Figure 11A and Figure 11B PDAF is used to perform data association on measurements occurring between times t0 and t5. When trajectories 1104 and 1114 are close to each other (e.g., when their respective control regions touch or overlap, or are closer than a predetermined threshold), the nearest neighbor (the measurement closest to the last measurement of each trajectory) is selected. By using a fusion method that combines PDAF with the nearest neighbor, computational complexity is reduced without significantly reducing accuracy.

[0113] In some embodiments, for non-conflict scenarios (e.g., human targets move far enough from each other so that the control areas of each trajectory do not touch or overlap each other and are at least farther than a predetermined threshold), PDAF is used to perform data association, which advantageously allows data association to be performed efficiently and accurately in high clutter environments. For conflict scenarios (e.g., the control areas of the trajectories of human targets touch or overlap or are closer than a predetermined threshold), a nearest neighbor algorithm is used to perform data association, which advantageously reduces the number of hypotheses and thus reduces computational complexity without substantially affecting accuracy. In some embodiments, the predetermined threshold may be 0 meters (control areas of contact), 0.1 meters, 0.2 meters, 0.5 meters, or higher).

[0114] Figure 12 12 shows trajectories 1204 and 1214 for simultaneously tracking human targets 1202 and 1212, respectively, according to an embodiment of the present invention. Figure 12 As shown, when tracks are close to each other, tracking switches from using PDAF to using the nearest neighbor algorithm.

[0115] Figure 13 FIG. 1 is a flow chart of an exemplary method 1300 for simultaneously tracking multiple human targets according to an embodiment of the present invention. The method 1300 may be implemented, for example, by the millimeter wave radar system 100. The method 1300 will refer to Figure 11A 、 Figure 11B and Figure 12 However, according to the embodiment of the present invention, it should be understood that Figure 11A 、 Figure 11B and Figure 12 Non-limiting examples of trajectories are shown.

[0116] Steps 754, 756, 758, 764, 768, 854, and 766 can be performed in a manner similar to that described with respect to methods 750 and / or 850. During step 1302, a determination is made as to whether the control region of a track is closer than a predetermined threshold from another control region of another track (e.g., using Euclidean distance and / or Mahalanobis distance). In some embodiments, step 1302 returns "yes" only if the control regions overlap. For example, in some embodiments, during step 1302, a determination is made as to whether a detected target is within the control region of a track and also within the control region of another track. If the detected target is within the control regions of both tracks, then during step 1304, a nearest neighbor algorithm is used to assign the detected target to the closest track.

[0117] If step 1302 returns “yes”, the trajectory is updated using the nearest neighbor algorithm during step 1304. If step 1302 returns “no”, the trajectory is updated using PDAF during step 1306.

[0118] Figure 14 FIG. 1 is a flow chart illustrating an exemplary method 1400 for detecting and tracking a human target according to an embodiment of the present invention. The method 1400 may be implemented, for example, by the millimeter wave radar system 100 .

[0119] During step 1402, a range-Doppler image is generated based on the radar signal received by the millimeter-wave radar 100. For example, in one embodiment, a millimeter-wave radar having two receiving antennas (RX1 and RX2) receives radar signals from the field of view (e.g., from reflections from human targets) during steps 1404 and 1406, respectively, and generates corresponding range-Doppler maps. During steps 1408 and 1410, a 2D MTI filter is applied to the corresponding range-Doppler map, and the filtered results are coherently integrated during step 1412. The example image generated by step 1402 is Figure 3A The image shown.

[0120] After coherent integration, a range-Doppler image is generated during step 1412. During step 1414, detection of potential targets is performed. For example, in some embodiments, an order statistics (OS) constant false alarm rate (CFAR) (OS-CFAR) detector is performed during step 1416. The CFAR detector generates a detection image based on, for example, the power level of the range-Doppler image, in which, for example, a "1" represents a target and a "0" represents a non-target. For example, in some embodiments, the CFAR detector compares the power level of the range-Doppler image to a threshold, and points above the threshold are marked as targets, while points below the threshold are marked as non-targets. Although targets can be represented by 1 and non-targets can be represented by 0, it should be understood that other values ​​can be used to represent targets and non-targets.

[0121] In one embodiment, OS-CFAR uses the kth quartile / median instead of the average CA-CFAR (unit average). Using the kth quartile / median can be advantageously more robust to outliers.

[0122] During step 1418, the objects present in the detected image are clustered to generate clustered objects. For example, in an embodiment, during step 1420, the objects are associated with clusters using the density-based spatial clustering of applications with noise (DBSCAN) algorithm. The output of DBSCAN is a grouping of the detected points into specific objects. DBSCAN is a popular unsupervised algorithm that clusters objects using minimum point and minimum distance criteria and can be implemented in any manner known in the art. Other clustering algorithms can also be used.

[0123] The example image generated by step 1418 is Figure 3B The image shown.

[0124] During step 1422, parameter estimates for each target are generated, for example, using one or more of Equations 2 through 9. During step 1424, a centroid estimation of the distances of each target cluster is performed (e.g., px and py in Equations 2 through 9). During step 1426, the angle of arrival (AoA) of each target is estimated. For example, in some embodiments, the minimum variance distortionless (MVDR) technique, also known as Capon, may be used to determine the angle of arrival during step 1426. Other methods may also be used.

[0125] A velocity estimate for each target is determined during step 1428. In some embodiments, the velocity estimate corresponds to v in Equations 2 through 9. c .

[0126] During step 1430, a bounding box in range-Doppler is determined. For example, parameters of the bounding box such as size (e.g., lx and ly) and rate of change of size (e.g., vlx and vly) are determined during step 1430. The area enclosed by the bounding box includes the extension of the detected human target in Doppler and range, from which micro-Doppler feature data can be extracted. For example, in Figure 4B An example of a bounding box determined during step 1430 is shown in FIG.

[0127] like Figure 14 As shown, a list of targets and associated parameters is generated during step 1422. Tracking of the target list is performed during step 1432. Detected targets are associated with corresponding tracks during step 1434. For example, during step 1436, detected targets are associated with tracks using a method such as method 1300.

[0128] During step 1438, tracking filtering is performed. For example, in some embodiments, tracking filtering is performed using an unscented Kalman filter or a particle filter during step 1440. For example, based on the history of the trajectory (e.g., based on previous detection points) and based on the new measurements received, an unscented Kalman filter can be used to predict distances, angles, bounding box sizes, and other parameters associated with the trajectory. An unscented Kalman filter or a particle filter can be implemented in any manner known in the art.

[0129] During step 1442, track management tasks are performed, such as generating tracks and terminating tracks. For example, during step 1444, tracking initialization, re-initialization, tracking termination, and / or multipath reflection suppression may be performed using methods 750, 850, and / or 1000, and / or combinations thereof. Figure 14 As shown, during step 1432 a list of tracked targets and associated parameters is generated.

[0130] Figure 15 A flow chart of an embodiment method 1500 for detecting and tracking a human target according to an embodiment of the present invention is shown. Method 1500 can be implemented, for example, by millimeter wave radar system 100 and can be implemented in a manner similar to method 1400. However, method 1500 generates a range cross-range image during step 1502 rather than a range-Doppler image. Method 1500 also determines a bounding box in the range cross-range domain rather than a bounding box in the range-Doppler domain during step 1530. Compared to a bounding box in the range-Doppler domain (which represents / captures the micro-Doppler component and range spread of the detected target), the bounding box in the range cross-range image captures the physical form / spread of the human target.

[0131] Figure 16 16 shows a detection result 1600 of two people walking across each other in an indoor environment according to an embodiment of the present invention. Data 1 and 2 correspond to the detection of human targets 1602 and 1604, respectively. The detection result 1600 can be generated, for example, in step 1422 or 1522. Figure 16 As shown, the detection results show missed detections, such as in areas 1606 and 1608.

[0132] In some embodiments, for example, Figure 16 As shown, up to 5 human targets (e.g., data 1 to 5) can be tracked simultaneously. In other embodiments, the maximum number of human targets can be higher than 5, such as 7, 10, or 10 or higher, or lower than 5, such as 3, 2 or 1.

[0133] Figure 17 The embodiment of the present invention is shown Figure 16 The tracking results of two people are 1700. Figure 17 As shown, after executing step 1432, the missing points of region 1606 are replaced by estimated values, and track 1 associated with human target 1602 continues to track human target 1602. Figure 17 As shown, although track 2 associated with human target 1604 has multiple missed detections in region 1608 , track 2 is not terminated and continues to track human target 1604 once human target 1604 is detected again.

[0134] Figure 18 1 shows a detection result 1800 of a person walking in a zigzag manner in an indoor environment according to an embodiment of the present invention. Data 1 corresponds to the detection of a human target 1802. The detection result 1800 may be generated, for example, in step 1422 or 1522. Figure 18 As shown, the detection results show false positives, such as in area 1804.

[0135] Figure 19 The embodiment of the present invention is shown Figure 18 The tracking results of people are 1900. Figure 19 As shown, after executing step 1432, the false positives in region 1804 are removed.

[0136] Example embodiments of the present invention are summarized here. Other embodiments are also contemplated from a comprehensive view of the specification and claims filed herein.

[0137] Example 1, a method for tracking a human target, the method comprising: receiving a radar signal using a radar sensor; generating a range-Doppler map based on the received radar signal; detecting the human target based on the range-Doppler map, wherein detecting the human target comprises determining a distance and a bounding box size defining a bounding box of the detected human target, the bounding box at least partially surrounding the detected human target; and when the determined distance is within an expected area associated with a track, adding a new detection point including the determined distance and the bounding box size to the track, wherein the expected area is determined based on the bounding box size of the detection point of the track.

[0138] Example 2, a method according to Example 1, wherein the bounding box has a rectangular shape surrounding the hands, torso, and feet of the detected human target, and wherein the bounding box size is the length of the side of the rectangular shape.

[0139] Example 3, according to one of the methods of Examples 1 or 2, further comprising: predicting the distance of the detected human target based on the trajectory; and determining the expected area based on the prediction.

[0140] Example 4. The method of one of Examples 1 to 3, wherein predicting the distance of the detected human target comprises predicting the distance of the detected human target based on a change in a size of a bounding box of the detected human target.

[0141] Example 5, a method according to one of Examples 1 to 4, wherein each detection point of the trajectory includes an angle value of the detected human target.

[0142] Example 6, according to the method of one of Examples 1 to 5, further comprising: when no target is detected within the expected area, adding the new detection point to the trajectory comprises adding the new detection point to the trajectory based on a predicted distance of the detected human target.

[0143] Example 7, the method of one of Examples 1 to 6, wherein predicting the range of the detected human target includes using an unscented Kalman filter.

[0144] Example 8, according to one of the methods of Examples 1 to 7, further comprising terminating the trajectory when the first N consecutive detection points of the trajectory include fewer than M confirmed detections of a human target, where N is a positive integer greater than 1, and where M is a positive integer less than N.

[0145] Example 9, a method according to one of Examples 1 to 8, wherein N is equal to 8 and M is equal to 5.

[0146] Example 10, according to one of the methods of Examples 1 to 9, further comprising terminating the trajectory after S consecutive detection points of the trajectory are missed detections of a human target, where S is a positive integer greater than 1.

[0147] Example 11, according to one of the methods of Examples 1 to 10, further comprising determining micro-Doppler feature data of the detected human target based on the range-Doppler map, wherein detecting the human target comprises detecting the human target when the micro-Doppler feature data indicates the presence of the human target.

[0148] Example 12, a method according to one of Examples 1 to 11, wherein the micro-Doppler feature data indicates a breathing rate of the detected human target.

[0149] Example 13. The method of one of Examples 1 to 12, wherein determining the micro-Doppler signature data comprises using a bounding box size.

[0150] Example 14. The method of one of Examples 1 to 13, further comprising: detecting a second human target based on the range-Doppler map, wherein detecting the second human target comprises determining a second distance and a second bounding box size of a second bounding box; and when the determined second distance is within a second expected area associated with the second track, adding a new detection point including the determined second distance and the second bounding box size to the second track.

[0151] Example 15. The method according to one of Examples 1 to 14 further includes: generating a second range-Doppler map based on the received radar signal; detecting a third human target based on the second range-Doppler map, wherein detecting the third human target includes determining a third distance and a third bounding box size of a third bounding box; generating a new first expected area based on the trajectory; generating a new second expected area based on the second trajectory; when the determined third distance is within the new first expected area and the new second expected area, using a nearest neighbor algorithm, adding a new detection point including the determined third distance and the third bounding box size to the trajectory or the second trajectory; and when the determined third distance is within the new first expected area and outside the new second expected area, or when the determined third distance is within the new second expected area and outside the new first expected area, using a probabilistic data association filter (PDAF), adding the new detection point including the determined third distance and the third bounding box size to the trajectory or the second trajectory.

[0152] Example 16. The method according to one of Examples 1 to 15, further comprising: determining a correlation value between the track and the second track; determining a closer track and a farther track between the track and the second track, wherein the closer track is one of the track and the second track that is closer to the radar sensor, and the farther track is the other track of the track and the second track; and deleting the other track when the correlation value is greater than a correlation threshold.

[0153] Example 17, a method according to one of Examples 1 to 16, wherein the correlation threshold is approximately 0.6.

[0154] Example 18. The method of one of Examples 1 to 17, wherein determining the bounding box size comprises determining the bounding box size in the range-Doppler domain.

[0155] Example 19. The method of one of Examples 1 to 18, wherein determining the bounding box size comprises determining the bounding box size in a distance lateral distance domain.

[0156] Example 20, a millimeter wave radar system, comprising: a radar sensor; and a processor, the processor being configured to: send a radar signal using the radar sensor, receive a reflected radar signal using the radar sensor, generate a range-Doppler map based on the received reflected radar signal, and detect a human target based on the range-Doppler map, wherein detecting the human target comprises determining a distance and a bounding box size of a bounding box of the detected human target, the bounding box at least partially enclosing the detected human target, and when the determined distance is within an expected area associated with the trajectory, adding a new detection point including the determined distance and the bounding box size to the trajectory, wherein the expected area is determined based on the bounding box size of the detection point of the trajectory.

[0157] Example 21, a method for simultaneously tracking multiple human targets, the method comprising: receiving a radar signal using a radar sensor; generating a range-Doppler map based on the received radar signal; detecting the human target based on the range-Doppler map, wherein detecting the human target includes determining the distance; generating a first expected area based on a first track associated with the first human target; generating a second expected area based on a second track associated with the second human target; when the determined distance is within the first expected area and the second expected area, using a nearest neighbor algorithm, adding a new detection point including the determined distance to the first track or the second track; and when the determined distance is within the first expected area and outside the second expected area, or when the determined distance is within the second expected area and outside the first expected area, using a probabilistic data association filter (PDAF) to add the new detection point including the determined distance to the first track or the second track.

[0158] Although the present invention has been described with reference to illustrative embodiments, this description is not intended to be construed in a limiting sense. Various modifications and combinations of the illustrative embodiments, as well as other embodiments of the invention, will be apparent to those skilled in the art by reference to the description. Accordingly, the appended claims encompass any such modifications or embodiments.

Claims

1. A method for tracking a human target, the method comprising, for a new position added to a trajectory of positions of the human target: receiving radar signals using a radar sensor; generating a range-Doppler map based on the received radar signal; detecting the human target based on the range-Doppler map, wherein the detected human target exhibits a spread in Doppler and in range bins in the range-Doppler map, wherein detecting the human target comprises determining a distance and a bounding box size defining a bounding box of the detected human target, the bounding box at least partially enclosing an extension in Doppler and range bins of the detected human target; as well as adding the new position including the determined distance and the bounding box size to the track when the determined distance is within an expected area associated with the track, wherein the expected area is determined based on the bounding box size of the position of the track; Based on the trajectory, predicting the distance of the detected human target; and determining the expected area based on the prediction; Predicting the distance of the detected human target includes predicting the distance of the detected human target based on a change in the size of the bounding box of the detected human target. 2 . The method of claim 1 , wherein the bounding box has a rectangular shape surrounding the hands, torso, and feet of the detected human target, and wherein the bounding box size is the length of a side of the rectangular shape. The method according to claim 1 , wherein each position of the trajectory includes an angle value of the detected human target.

4. The method according to claim 1, further comprising: When no object is detected within the expected area, adding the new position to the trajectory includes adding the new position to the trajectory based on a predicted distance of the detected human object. The method of claim 1 , wherein predicting the distance of the detected human target comprises using an unscented Kalman filter.

6. The method of claim 1 , further comprising terminating the trajectory when the first N consecutive positions of the trajectory include fewer than M confirmed detections of the human target, where N is a positive integer greater than 1, and where M is a positive integer less than N. The method of claim 6 , wherein N is equal to 8 and M is equal to 5. 8 . The method of claim 1 , further comprising terminating the trajectory after S consecutive positions of the trajectory are missed detections of the human target, where S is a positive integer greater than 1.

9. The method of claim 1, further comprising determining micro-Doppler characteristic data of the detected human target based on the range-Doppler map, wherein detecting the human target comprises detecting the human target when the micro-Doppler characteristic data indicates presence of the human target. 10 . The method of claim 9 , wherein the micro-Doppler characteristic data indicates a breathing rate of the detected human target. The method of claim 9 , wherein determining the micro-Doppler signature data comprises using the bounding box size.

12. The method according to claim 1, further comprising: detecting a second human target based on the range-Doppler map, wherein detecting the second human target comprises determining a second distance and a second bounding box size of a second bounding box; as well as When the determined second distance is within a second expected area associated with the second track, a new detection point including the determined second distance and the second bounding box size is added to the second track.

13. The method according to claim 12, further comprising: generating a second range-Doppler map based on the received radar signal; detecting a third human target based on the second range-Doppler map, wherein detecting the third human target comprises determining a third distance and a third bounding box size of a third bounding box; generating a new first expected area based on the trajectory; generating a new second expected area based on the second trajectory; When the determined third distance is within the new first expected area and the new second expected area, using a nearest neighbor algorithm, adding a new detection point including the determined third distance and the third bounding box size to the track or the second track; as well as When the determined third distance is within the new first expected area and outside the new second expected area, or when the determined third distance is within the new second expected area and outside the new first expected area, a new detection point including the determined third distance and the third bounding box size is added to the track or the second track using a probabilistic data association filter (PDAF).

14. The method according to claim 12, further comprising: determining a correlation value between the trajectory and the second trajectory; determining a closer track and another track between the track and the second track, wherein the closer track is one of the track and the second track that is closer to the radar sensor, and the another track is the other of the track and the second track; as well as The another track is deleted when the correlation value is greater than a correlation threshold. The method according to claim 14 , wherein the correlation threshold is 0.

6.

16. The method of claim 1, wherein determining the bounding box size comprises determining the bounding box size in a range-Doppler domain.

17. The method of claim 1, wherein determining the bounding box size comprises determining the bounding box size in a distance-to-lateral distance domain.

18. A millimeter wave radar system for tracking human targets, comprising: radar sensors; as well as The processor is configured to, for a new position added to the trajectory of the position of the human body: transmitting a radar signal using the radar sensor, receiving the reflected radar signal using the radar sensor, Generate a range-Doppler map based on the received reflected radar signal, detecting a human target based on the range-Doppler map, wherein the detected human target exhibits a spread in Doppler and in range bins in the range-Doppler map, wherein the processor is configured to determine a range and a bounding box size of a bounding box of the detected human target, the bounding box at least partially enclosing an extension of the detected human target in Doppler and in range bins, and adding a new position including the determined distance and the bounding box size to the track when the determined distance is within an expected area associated with the track, wherein the processor is configured to determine the expected area based on the bounding box size of the position of the track; Based on the trajectory, predicting the distance of the detected human target; and determining the expected area based on the prediction; Predicting the distance of the detected human target includes predicting the distance of the detected human target based on a change in the size of the bounding box of the detected human target.

19. A method for simultaneously tracking multiple human targets, the method comprising: receiving radar signals using a radar sensor; generating a range-Doppler map based on the received radar signal; detecting a human target based on the range-Doppler map, wherein detecting the human target comprises determining a distance; generating a first expected region based on a first trajectory associated with a first human target; generating a second expected region based on a second trajectory associated with a second human target; When the determined distance is within the first expected area and the second expected area, adding a new position including the determined distance to the first track or the second track using a nearest neighbor algorithm; as well as When the determined distance is within the first expected area and outside the second expected area, or when the determined distance is within the second expected area and outside the first expected area, a new position including the determined distance is added to the first track or the second track using a probabilistic data association filter (PDAF).

Images