Target detection in rainfall and snowfall conditions using mmWave radar

Abstract

A radar system includes a radio frequency (RF) circuit to generate a transmit signal and to receive a corresponding receive signal from a target during rainfall or snowfall conditions, and a signal processing circuit coupled to the RF circuit to generate an adaptive filter threshold in response to the rainfall or snowfall conditions, and to generate a valid target signal in response a portion of the receive signal above the adaptive filter threshold.

Claims

1. A radar system comprising: a radio frequency (RF) circuit configured to generate a transmit signal and to receive a corresponding receive signal from a target during rainfall or snowfall conditions; and a signal processing circuit coupled to the RF circuit configured to generate an adaptive filter threshold in response to the rainfall or snowfall conditions, and to generate a valid target signal in response to a portion of the receive signal above the adaptive filter threshold, wherein the signal processing circuit comprises a plurality of pass-band Doppler filters having a corresponding plurality of range bins, and wherein the valid target signal is generated when a target range spread of three consecutive range bins of the plurality of pass-band Doppler filters is less than a predetermined length.

2. The radar system of claim 1, wherein at least one of the plurality of pass-band Doppler filters comprises a clutter map for generating the adaptive filter threshold.

3. The radar system of claim 1, wherein the signal processing circuit comprises target tracking circuitry for generating a valid approaching target signal.

4. The radar system of claim 3, further comprising an actuator driver for receiving the valid target approaching signal.

5. The radar system of claim 1, wherein the radar system further comprises a semiconductor package.

6. The radar system of claim 5, wherein the semiconductor package further comprises a plastic shield.

7. The radar system of claim 5, wherein the semiconductor package further comprises a lens.

8. The radar system of claim 1, wherein the radar system comprises further comprises a printed circuit board (PCB).

9. The radar system of claim 8, wherein the PCB further comprises a plastic shield.

10. A method for detecting a target during rainfall or snowfall conditions comprising: generating a transmit signal; receiving a receive signal from a target during rainfall or snowfall conditions, the receive signal comprising an echo of the transmit signal; generating an adaptive filter threshold in response to the rainfall or snowfall conditions; generating a valid target signal in response to a portion of the receive signal above the adaptive filter threshold; and measuring a range spread of the receive signal, wherein generating the valid target signal comprises using a plurality of pass-band Doppler filters having a corresponding plurality of range bins, and wherein the valid target signal is generated when the range spread of three consecutive range bins of the plurality of pass-band Doppler filters is less than a predetermined length.

11. The method of claim 10 further comprising tracking the target to generate a valid approaching target signal.

12. The method of claim 11 further comprising controlling an operation of an escalator with the valid approaching target signal.

13. The method of claim 11 further comprising controlling an operation of a door with the valid approaching target signal.

14. An integrated radar system comprising: a transmit antenna; a receive antenna; RF and baseband circuitry in communication with the transmit antenna and the receive antenna; and digital signal processing circuitry in communication with the RF and baseband circuitry, the digital signal processing circuitry comprising a plurality of Doppler filters each coupled to an adaptive threshold detector, wherein at least one of the Doppler filters and at least one of the adaptive threshold detectors are coupled together by a rain clutter map, wherein the plurality of Doppler filters have a corresponding plurality of range bins, and wherein a valid target signal is generated when a target range spread of three consecutive range bins of the plurality of Doppler filters is less than a predetermined length.

15. The integrated radar system of claim 14, wherein the integrated radar system is integrated into a single package.

16. The integrated radar system of claim 15, further comprising a lens.

17. The integrated radar system of claim 14, wherein the integrated radar system is integrated onto a printed circuit board.

18. The integrated radar system of claim 14, further comprising an output for controlling an actuator driver.

19. The radar system of claim 1, wherein the predetermined length is 60 cm.

20. The method of claim 10, wherein the predetermined length is 60 cm.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) For a more complete understanding of the present invention, and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:

(2) FIG. 1 is a radar system/sensor in a semiconductor package having a plastic shield according to an embodiment;

(3) FIG. 2 is a radar system/sensor on a Printed Circuit Board (PCB) having a plastic shield according to an embodiment;

(4) FIG. 3 is a radar system/sensor in a semiconductor package having a lens according to an embodiment;

(5) FIG. 4 is a diagram of radar detection of a human target in rainfall and snowfall clutter according to an embodiment;

(6) FIG. 5 is a diagram or radar detection and tracking of a human target in rainfall and snowfall clutter according to an embodiment;

(7) FIG. 6 is an overall flowchart of a radar detection and tracking method in rainfall and snowfall clutter according to an embodiment;

(8) FIG. 7 is a more detailed flowchart of the radar detection and tracking method of FIG. 6, according to an embodiment;

(9) FIG. 8 is a diagram of the equations used in user position and velocity tracking, according to an embodiment; and

(10) FIG. 9 is a flowchart of actuator logic for controlling an automatic door or escalator in rainfall or snowfall clutter, according to an embodiment.

DETAILED DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS

(11) FIG. 1 is a radar system/sensor 100 in a semiconductor package 102 having a plastic shield 114 to prevent direct exposure to rain or snow 116, according to an embodiment. Semiconductor package 102 includes a transmitter Tx1, and two receivers Rx1 and Rx2 that can be formed in a first level metal M1, in an embodiment. A second level metal M2 can be used as a ground plane, in an embodiment. A third level metal M3 can be used as an interconnect layer for interfacing with ball bonds 104. Semiconductor package 102 includes all of the circuitry for the radar system/sensor 100 including Digital Signal Processing (DSP) processing circuitry 106, memory 108, and RF and baseband circuitry 110. Semiconductor package 102 can be formed out of ceramic or plastic or other materials known in the art. Plastic shield 114 can comprise a plastic polycarbonate material or an Acrylonitrile Butadiene Styrene (ABS) material. Plastic shield 114 need not be transparent to visible light, but is transparent at millimeter radar frequencies. Plastic shield 114 can be affixed directly to package 102 or to a substrate (not shown) associated with package 102. Semiconductor package 102 can have additional metal and insulating layers, vias, interconnects, circuits, and other packaging artifacts according to embodiments.

(12) FIG. 2 is a radar system/sensor 200 on a Printed Circuit Board (PCB) 202 having a plastic shield 216 according to an embodiment. PCB 202 includes a transmitter 204, and two or more receivers 206 that can be formed in a first level metal, in an embodiment. A second level metal 214 can be used as a ground plane, in an embodiment. A third level metal (not shown) can be used as an interconnect layer for interfacing with ball bonds or other interconnect features (not shown). Semiconductor package 202 includes all of the circuitry for the radar system/sensor 200 including RF microwave integrated circuit (RF/MMIC) circuitry 208, DSP circuitry 210, and memory 212. PCB 202 can be formed out of fiberglass or other materials known in the art. Plastic shield 216 can comprise a plastic polycarbonate material or an ABS material. Plastic shield 216 need not be transparent to visible light, but is transparent at millimeter radar frequencies. Plastic shield 216 can be affixed directly to PCB 202 or to a substrate (not shown) associated with PCB 202. PCB 102 can have additional metal and insulating layers, vias, interconnects, circuits, and other such artifacts according to embodiments.

(13) FIG. 3 is a radar system/sensor 300 in a semiconductor package 302 having a lens 312 according to an embodiment. The lens 312 can be configured (shaped) to enhance beam performance, or can be used solely for protection of the sensor against environmental elements without necessarily enhancing beam performance. Semiconductor package 302 includes a transmitter Tx1, and two receivers Rx1 and Rx2 that can be formed in a first level metal M1, in an embodiment. A second level metal M2 can be used a ground plane, in an embodiment. A third level metal M3 can be used as an interconnect layer for interfacing with ball bonds 304. Semiconductor package 302 includes all of the circuitry for the radar system/sensor 100 including DSP processing circuitry 306, memory 308, and RF and baseband circuitry 310. Semiconductor package 102 can be formed out of ceramic or plastic or other materials known in the art. Lens 312 can comprise a plastic polycarbonate material or an ABS material, as well as other plastic or glass materials. Lens 312 need not be transparent to visible light, but is transparent at millimeter radar frequencies. Lens 312 can be affixed directly to package 302 as shown, in an embodiment. Semiconductor package 302 can have additional metal and insulating layers, vias, interconnects, circuits, and other packaging artifacts according to embodiments.

(14) FIGS. 1, 2, and 3 show embodiments wherein a minimum 1Tx/1Rx solution can be used for rainfall/snowfall detection or target detection during rainfall/snowfall conditions, and an additional receiver can be used to track the movement of human being targets to generate a valid target approaching signal as is described in further detail below.

(15) FIG. 4 shows an example scenario 400 of human target 406 detection using a radar 402 in rainfall or snowfall clutter conditions 404 produced by rain or snow clouds 408. FIG. 5 shows an example scenario 500 of human target tracking from location 406B to 406A using radar 402 in rainfall or snowfall clutter conditions 404 produced by rain or snow clouds 408. Human target tracking can be used to generate a valid target approaching signal that in turn can be used to activate an appliance actuator, wherein the appliance is also subjected to rainfall or snowfall conditions. An actuator is a component of an appliance that is responsible for moving and controlling a mechanism or system therein. An actuator driver is an electronic device such as an integrated circuit for appropriately energizing the actuator in response to an input signal.

(16) Rain or snow clutter can be diffusive and windblown, and thus presents a challenge to both human target detection and human target tracking. Rainfall or snowfall can have a radar signature 30+dB or more, greater than that of a human target. Rainfall or snowfall can have a small Doppler spread around zero. Thus, a single filter with a notch at zero Doppler will not adequately reject rain. A set of pass band Doppler filters using 16-32 consecutive pulses can be constructed having a notch at zero Doppler. This implementation is sufficient to reject ground clutter but is not sufficient for rain clutter or snow clutter since tangential velocity detected by the radar introduces non-zero components near zero Doppler. A set of passband filters that are away from the zero Doppler can detect targets where no rain is present.

(17) Doppler filter banks with adaptive thresholding for each range-Doppler combination are used for first level processing, according to an embodiment. The near zero-Doppler bins store rain clutter map, which are used for thresholding. The other pass band filter banks have adaptive threshold based on the mean noise floor. Second level processing measures the range spread of the detected target to discriminate between an actual human target and rain clutter.

(18) Based on the post-processing detection, a valid target (human) is fed into a tracking algorithm to monitor its direction of movement. For example, if the direction of the detected target is towards the sensor over a predetermined period of time, e.g., 1-2 seconds, an actuator is activated to enable the operation of an appliance such as an escalator or an automatic door. A separate Kalman filter track may be enabled for each valid target detected and if any of the valid targets satisfies the above time duration condition, the actuator is enabled.

(19) FIG. 6 is an overall flowchart 600 of a radar detection and tracking method in rainfall and snowfall clutter according to an embodiment. Initially, analog return radar signals are processed and converted to digital signals at step 602. Range Fast Fourier Transform (FFT) digital signal processing is performed at step 604 to change the time-based digital signals into the frequency domain. Range Gate Detection is performed at step 606 to narrow the types of targets. While not limiting, a minimum range of detection is on the order of one meter, while a maximum range of detection is on the order of six meters. Once the Range Gate Detection has been performed at step 606, the output data is captured across slow-time at step 608. In an embodiment, slow-time refers to the pulse dimension, for example the number of pulses transmitted in a pulsed radar as is known in the art.

(20) The captured data is further processed in filter bank 610, which is explained in further detail below, and in further detail with respect to the description of FIG. 7. Returning to FIG. 6, filter bank 610 includes a Doppler filter map 612, a rain clutter map 614, and an adaptive threshold detector 616. A direct connection between the Doppler filter map 612 and the adaptive threshold detector 616 is shown in FIG. 6; this is because the adaptive detection threshold is set by extracting the rain-Doppler components from the Doppler filter map 612. In FIG. 6, the Doppler filter map 612 comprises a plurality of Doppler filters corresponding to a plurality of negative, zero, and positive Doppler velocity bands as will be explained in further detail below. The adaptive threshold detector 616 comprises a plurality of threshold detectors 616 for removing rainfall and snowfall clutter information provided by the rain clutter map 614 as is described in further detail below.

(21) Integration of many pulses using low Doppler side lobes is used to reject rain clutter. Embodiment methods employ adaptive thresholding using the rain clutter map 614 to set the threshold in the adaptive threshold detector 616.

(22) High resolution clutter maps 614 are used to detect tangential targets. Various clutter map techniques are described below, according to embodiments. In general, clutter maps 614 comprise a memory that stores, for each for each range-CPI cell (Coherent Processing Interval cell) in the radar's coverage, the value of the noise and clutter echo in that cell. The clutter echoes can contain snowfall and rainfall echo. The clutter map is for instance implemented using the following recursive filter equation:

(23) $A\left(n+1\right)=\frac{1}{N}A\left(n\right)+\left(1-\frac{1}{N}\right)A\left(n-1\right)$

(24) Wherein A(n) is the amplitude of the near-zero Doppler values, N is the window of the filter, typically N=16, and n is the index of the frame (continuous running).

(25) The recursive filter in the clutter map is used to detect targets whose radial velocity is at or near zero and whose backscatter echo is greater than the clutter and/or noise amplitude stored in the clutter map. The clutter map channel offers a method of detecting targets that are not detected by the subset of the Doppler filters that are adjacent to zero Doppler and whose shape is designed to strongly reject ground echoes near zero Doppler as is described in further detail with respect to FIG. 7. Range cluster processed targets, i.e. range gate gaps of 1-2 are considered as being the same target.

(26) After processing in the filter bank 610, post-processing is performed in post-processing threshold block 618 to generate a valid target signal 626, i.e. a signal indicative of the detection of a person. The valid target signal 626 is tracked using a Kalman filter target tracking block 620 to generate a valid approaching target signal 622, i.e. a signal indicative of the detection of a person moving towards the sensor. The valid approaching target signal can be used as the input signal for an actuator driver 624 associated with an appliance such as an escalator or an automatic door that is also subject to rainfall or snowfall conditions.

(27) FIG. 7 is a more detailed flowchart 700 of the radar detection and tracking method of FIG. 6, according to an embodiment. Steps 602, 604, 606, 608, 618, 620, and 624 have been previously explained. Further details of the filter bank 610 are shown in FIG. 7. A plurality of Doppler filters each coupled to an adaptive threshold detector is shown. Three to five neighboring filters 706 include a rain clutter map 710, centered around the Doppler filter 0 (pass-band Doppler) 708. Doppler filter 708 is coupled to rain clutter map 710 and adaptive threshold detector 712. The neighboring filters 706 are similarly configured. The remaining Doppler filters do not include a rain clutter map 710. For example Doppler filter-8 (pass-band Doppler) 702 is only coupled to adaptive threshold detector 704. Doppler filter 7 (pass-band Doppler) 714 is only coupled to adaptive threshold detector 716. While Doppler velocity bands-8 through 7 are shown, other bands can be used. Similarly, while three to five neighboring filters 706 are coupled to a rain clutter map, other neighboring filters can also be coupled to a rain clutter map.

(28) Rainfall and snowfall echoes have a Doppler velocity near the Zero Doppler, for example 3 m/s to 3 m/s based on the angle of incidence of the rainfall/snowfall. However this Doppler information is unknown to the system a priorithe actual information in a particular rainfall or snowfall condition will affect the 3-5 Doppler bins/filters around the 0 Doppler bin/filter. These filters/bins are set with rain clutter map 710 as is shown in FIG. 7 by the adaptively adjusting the detection threshold based on the rainfall/snowfall these Doppler frequencies/bins encounter. A human/target above this adaptive threshold dictated by the clutter map is a valid potential target.

(29) The post-processing techniques provided by post-processing block 618 further discriminates between rain/snow clutter and human targets by identifying the spread of the target along the range or range-azimuth. Based on whether the analysis is done on a range-only dimension or a range-azimuth plane a 1D or 2D sliding window is deployed to discriminate between rain/snow and human.

(30) The spread (length) of each target is calculated using the following equation:
length=max{range-gate}min{range-gate}.

(31) Wherein max{range-gate} is the maximum allowable distance set by the range gate and min{range-gate} is the minimum allowable distance set by the range gate.

(32) If the spread (length) of the target is greater than 60 cm (range-gates depends on the range resolution), the target is classified as rain clutter and not detected as valid target. Once a valid potential target is detected from a Doppler filter bin/map, neighboring range bins are also detected to determine the validity of a target. The range spread of the detected target is determined and if it is within the 60 cm range spread, the detected target is still a valid target. If this exceeds the 60 cm range spread, the detected target and associated range bins are ignored and potentially triggered by the rainfall/snowfall itself. The post-processing step 618 can be executed before or after the Doppler processing step 610.

(33) A nearest neighbor clustering technique is for instance used for target determination, and is illustrated in further detail below with a jump of one range bin. If there is detection on range bins 2, 3, 5, 6, and 7.fwdarw.(2, 7) there is a single target with spread of 5. If there is detection on range bins 12, 15, 16, and 17.fwdarw.(12), (15, 16, 17) there are two targets with spreads of 1 and 3, respectively. If there is detection on range bins: 5, 7, 8, 9, 19, 21, 22, 29.fwdarw.(5, 7, 8, 9), (19, 21, 22) and (29) there are three targets with spreads of 4, 3 and 1, respectively.

(34) Target range spread is thus calculated by the following equation:
Range spread indices (i.e. if three consecutive range bins are excited)*range resolution=target range spread<=60 cm.fwdarw.valid potential target.

(35) User position and velocity tracking equations are shown in FIG. 8, which are used in the digital signal processing chip of the radar system, according to embodiments. The time update equations are shown in block 802, the measurement update equations are shown in block 804, and further updating equations for use with the time and measurement update equations are shown in blocks 806, 808, and 810. The equations shown in FIG. 8 are used in the algorithm shown in FIG. 6. The range and velocity are estimated from the blocks shown in FIG. 6 and are the input to block 620 (Kalman filter target tracking). The range gate is determined by a detection along the Fast Time range FFT, whereas the Doppler FFT detection is done through the Adaptive Clutter Map thresholding along the slow time/Doppler FFT.

(36) A constant velocity model is used with the following parameters =[r.sub.k, v.sub.k] and

(37) $H_k=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right]\mathrm{and}{A}_k=\left[\begin{array}{cc}1& t\\ 0& 1\end{array}\right]\mathrm{and}{R}_k=\left[\begin{array}{cc}{r}_{\mathrm{err}}^2& 0\\ 0& v_{\mathrm{err}}^2\end{array}\right]$

(38) The unknown acceleration is factored in the Q.sub.k model covariance matrix. A.sub.k is the state space matrix, and determines the change of state. H.sub.k is the measurement model. R.sub.k is the estimation error matrix (coming from the sensor estimates). Q.sub.k is the model covariance matrix (since a constant velocity model is used, target acceleration/jerks are accounted for in this matrix). The Kalman filtering described above occurs in block 620 (Kalman filter target tracking) in FIG. 6.

(39) FIG. 9 is a flowchart 900 of actuator logic for controlling an automatic door or escalator in rainfall or snowfall clutter, according to an embodiment. As previously described, a valid target approaching signal 908 is generated for a door or escalator or other appliance subjected to snowfall or rainfall conditions. The valid approaching signal 908 is used to control the actuator or controller (which in an embodiment can be a Proportional-Derivative controller) for the door or escalator at step 902. At step 904 the door or escalator is turned on for a predetermined time set in block 904. After another predetermined period of time the appliance is timed out at step 906.

(40) Application use cases include but are not limited to cases where an approaching human target is detected in snowfall or rainfall conditions and robust and reliable escalator control and automatic door opening is desired and achieved despite the escalator or automatic door being exposed to the snowfall or rainfall conditions.

(41) While this 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 persons skilled in the art upon reference to the description. It is therefore intended that the appended claims encompass any such modifications or embodiments.