TECHNIQUES FOR BAND SPECIFIC PEAK DETECTION IN A LIDAR SYSTEM

Abstract

A light detection and ranging (LIDAR) system performs a method including generating a frequency domain waveform based on a baseband electrical signal in a time domain, wherein the frequency domain waveform includes a spectrum of frequencies, separating the spectrum of frequencies in the frequency domain waveform into multiple frequency bands including at least a first frequency band and a second frequency band, and performing a first peak detection within the first frequency band. The method further includes performing a second peak detection within the second frequency band, wherein the first peak detection and second peak detection are different peak detection techniques, and selecting a peak frequency from the spectrum of frequencies in the frequency domain waveform based at least in part on the first peak detection within the first frequency band and the second peak detection within the second frequency band.

Description

FIELD

The present disclosure is related to light detection and ranging (LIDAR) systems in general, and more particularly to peak detection in frequency-modulated continuous-wave (FMCW) LIDAR systems.

BACKGROUND

Frequency-Modulated Continuous-Wave (FMCW) LIDAR systems use tunable, infrared lasers for frequency-chirped illumination of targets, and coherent receivers for detection of backscattered or reflected light from the targets that are combined with a local copy of the transmitted signal. Mixing the local copy with the return signal, delayed by the round-trip time to the target and back, generates signals at the receiver with frequencies that are proportional to the distance to each target in the field of view of the system. Human safety considerations mandate the use of low-power lasers so that reflections from objects have very low signal strength. The range and accuracy of a LIDAR system is a function of signal-to-noise ratio, yet conventional solutions fail to reliably detect targets with a weak return signal while also limiting false target detections.

SUMMARY

The present disclosure describes examples of systems and methods for peak detection in FMCW LIDAR.

A light detection and ranging (LIDAR) system includes an optical scanner to transmit an optical beam towards, and receive a return signal from, a target, an optical processing system coupled to the optical scanner to generate a baseband signal in a time domain from the return signal, the baseband signal comprising frequencies corresponding to LIDAR target ranges, and a signal processing system coupled to the optical processing system. The signal processing system includes a processor and a memory operatively coupled to the processor, the memory to store instructions that, when executed by the processor, cause the LIDAR system to generate a frequency domain waveform based on the baseband signal in the time domain, wherein the frequency domain waveform comprises a spectrum of frequencies and separate the spectrum of frequencies in the frequency domain waveform into a plurality of frequency bands comprising at least a first frequency band and a second frequency band. The instructions further cause the LIDAR system to perform a first peak detection within the first frequency band of the spectrum of frequencies, perform a second peak detection within the second frequency band of the spectrum of frequencies, wherein the first peak detection and second peak detection comprise different peak detection techniques, and select a peak frequency from the spectrum of frequencies in the frequency domain waveform based at least in part on the first peak detection within the first frequency band and the second peak detection within the second frequency band.

In some embodiments, the processor is further to determine the first peak detection for the first frequency band and the second peak detection for the second frequency band based on properties of the LIDAR system or the target. In some embodiments, the first and second peak detection each comprise at least one of thresholding and peak selection using one likelihood metric, thresholding and peak selection using separate likelihood metrics, thresholding and peak selection using a weighted likelihood metric, and frequency band filtering. In some embodiments, the processor is further to determine one or more properties of the target based at least in part on the selected peak frequency.

In some embodiments, the first peak detection selects a first intermediate peak from the first frequency band and the second peak detection selects a second intermediate peak from the second frequency band. In some embodiments, to select the peak frequency the processor is to select one of the first or second intermediate peak as the peak frequency. In some embodiments, to select the peak frequency, the processor is to perform a third peak detection on the first and second intermediate peaks. In some embodiments, to select one of the first or second intermediate peaks as the peak frequency, the processor is to determine a priority order of the first and second frequency bands and select one of the first or second intermediate peaks as the peak frequency based on the priority order of the first and second frequency bands. In some embodiments, the processor is further to determine whether a value of a likelihood metric associated with the first intermediate peak or second intermediate peak exceeds a threshold and select the peak frequency from the first intermediate peak and the second intermediate peak that has a highest value for the likelihood metric.

In some embodiments, a method, includes generating a frequency domain waveform based on a baseband electrical signal in a time domain, wherein the frequency domain waveform comprises a spectrum of frequencies and separating the spectrum of frequencies in the frequency domain waveform into a plurality of frequency bands comprising at least a first frequency band and a second frequency band. The method further includes performing a first peak detection within the first frequency band of the spectrum of frequencies, performing a second peak detection within the second frequency band of the spectrum of frequencies, wherein the first peak detection and second peak detection comprise different peak detection techniques, and selecting a peak frequency from the spectrum of frequencies in the frequency domain waveform based at least in part on the first peak detection within the first frequency band and the second peak detection within the second frequency band.

In some embodiments, a non-transitory computer-readable medium containing instructions that, when executed by a processor in a LIDAR system, cause the processor of the LIDAR system to generate a frequency domain waveform based on a baseband electrical signal in a time domain, wherein the frequency domain waveform comprises a spectrum of frequencies and separate the spectrum of frequencies in the frequency domain waveform into a plurality of frequency bands comprising at least a first frequency band and a second frequency band. The processor further performs a first peak detection within the first frequency band of the spectrum of frequencies, performs a second peak detection within the second frequency band of the spectrum of frequencies, wherein the first peak detection and second peak detection comprise different peak detection techniques, and selects a peak frequency from the spectrum of frequencies in the frequency domain waveform based at least in part on the first peak detection within the first frequency band and the second peak detection within the second frequency band.

BRIEF DESCRIPTION OF THE DRAWINGS

For a more complete understanding of various examples, reference is now made to the following detailed description taken in connection with the accompanying drawings in which like identifiers correspond to like elements:

FIG. 1 is a block diagram illustrating an example LIDAR system according to the present disclosure;

FIG. 2 is a time-frequency diagram illustrating one example of LIDAR waveforms according to the present disclosure;

FIG. 3A is a block diagram illustrating an example LIDAR system according to the present disclosure;

FIG. 3B is a block diagram illustrating an electro-optical optical system according to the present disclosure;

FIG. 4 is a block diagram of an example signal processing system according to the present disclosure;

FIG. 5A is a signal magnitude-frequency diagram illustrating an example method of peak detection according to the present disclosure;

FIG. 5B is a signal magnitude-frequency diagram illustrating a noise estimate compared to a difference between a signal spectrum and the noise estimate according to the present disclosure;

FIG. 6A is a likelihood metric-frequency diagram illustrating an example of peak detection using a likelihood metric for thresholding and peak selection, according to the present disclosure;

FIG. 6B is a likelihood metric-frequency diagram illustrating another example of peak detection using a likelihood metric for thresholding and peak selection in which no peak is selected, according to the present disclosure;

FIG. 7 is a flowchart illustrating a method of peak detection using a single likelihood metric for thresholding and peak selection according to the present disclosure

FIG. 8A is a thresholding metric-frequency diagram illustrating an example of peak thresholding using a thresholding metric different from a peak selection metric, according to the present disclosure;

FIG. 8B is a peak selection metric-frequency diagram illustrating an example of peak selection using a peak selection metric different from a thresholding metric, according to the present disclosure;

FIG. 9 is a flowchart illustrating a method of peak detection using different likelihood metrics for thresholding and peak selection, according to the present disclosure

FIG. 10 is a thresholding metric-frequency diagram illustrating an example of peak detection using a thresholding metric and a minimum peak width, according to the present disclosure;

FIG. 11 is a flowchart illustrating a method of peak detection using thresholding with a minimum peak width, according to the present disclosure

FIG. 12A is a likelihood metric-frequency diagram illustrating an example of a likelihood metric for peak detection, according to the present disclosure;

FIG. 12B is a likelihood metric-frequency diagram illustrating another example of peak detection using a weighted likelihood metric for thresholding and peak selection, according to the present disclosure;

FIG. 13 is a flowchart illustrating a method of peak detection using a weighted likelihood metric, according to the present disclosure;

FIG. 14 is a likelihood metric-frequency diagram illustrating an example of filtering out frequencies prior to peak selection, according to the present disclosure;

FIG. 15 is a flowchart illustrating a method of peak detection by filtering out frequency bands prior to peak selection, according to the present disclosure;

FIG. 16 is a likelihood metric-frequency diagram illustrating an example of using band specific peak selection strategies for peak detection, according to the present disclosure;

FIG. 17 is a flowchart illustrating a method for peak detection according to the present disclosure; and

FIG. 18 is a block diagram of an example signal processing system according to the present disclosure.

DETAILED DESCRIPTION

The present disclosure describes various examples of LIDAR systems and methods for peak detection using band specific peak detection techniques to improve target detection and reduce false detections. According to some embodiments, the described LIDAR system described herein may be implemented in any sensing market, such as, but not limited to, transportation, manufacturing, metrology, medical, virtual reality, augmented reality, and security systems. According to some embodiments, the described LIDAR system is implemented as part of a front-end frequency modulated continuous-wave (FMCW) device that assists with spatial awareness for automated driver assist systems, or self-driving vehicles.

Lidar systems described by the embodiments herein include coherent scan technology to detect a signal returned from a target to generate a coherent heterodyne signal, from which range and velocity information of the target may be extracted. The signal may be converted into one or more frequency bins, each with a magnitude of the associated frequencies within the bin. In some scenarios, a target detection may correspond to a large magnitude (i.e., a peak) for one or more frequency bins. However, selecting a peak that properly corresponds to an actual target detection may be difficult due to internal and external noise sources and other interferences that may occur. Using the techniques described herein, embodiments of the present invention can, among other things, address the issues described above by selecting and performing peak detection using different peak detection techniques for multiple different frequency bands within the same peak search. Accordingly, the probability of detecting a target is increased while the probability of false detections is increased by using one or more likelihood metrics.

FIG. 1 illustrates a LIDAR system 100 according to example implementations of the present disclosure. The LIDAR system 100 includes one or more of each of a number of components, but may include fewer or additional components than shown in FIG. 1. One or more of the components depicted in FIG. 1 can be implemented on a photonics chip, according to some embodiments. The optical circuits 101 may include a combination of active optical components and passive optical components. Active optical components may generate, amplify, and/or detect optical signals and the like. In some examples, the active optical component includes optical beams at different wavelengths, and includes one or more optical amplifiers, one or more optical detectors, or the like.

Free space optics 115 may include one or more optical waveguides to carry optical signals, and route and manipulate optical signals to appropriate input/output ports of the active optical circuit. The free space optics 115 may also include one or more optical components such as taps, wavelength division multiplexers (WDM), splitters/combiners, polarization beam splitters (PBS), collimators, couplers or the like. In some examples, the free space optics 115 may include components to transform the polarization state and direct received polarized light to optical detectors using a PBS, for example. The free space optics 115 may further include a diffractive element to deflect optical beams having different frequencies at different angles along an axis (e.g., a fast-axis).

In some examples, the LIDAR system 100 includes an optical scanner 102 that includes one or more scanning mirrors that are rotatable along an axis (e.g., a slow-axis) that is orthogonal or substantially orthogonal to the fast-axis of the diffractive element to steer optical signals to scan an environment according to a scanning pattern. For instance, the scanning mirrors may be rotatable by one or more galvanometers. Objects in the target environment may scatter an incident light into a return optical beam or a target return signal. The optical scanner 102 also collects the return optical beam or the target return signal, which may be returned to the passive optical circuit component of the optical circuits 101. For example, the return optical beam may be directed to an optical detector by a polarization beam splitter. In addition to the mirrors and galvanometers, the optical scanner 102 may include components such as a quarter-wave plate, lens, anti-reflective coated window or the like.

To control and support the optical circuits 101 and optical scanner 102, the LIDAR system 100 includes LIDAR control systems 110. The LIDAR control systems 110 may include a processing device for the LIDAR system 100. In some examples, the processing device may be one or more general-purpose processing devices such as a microprocessor, central processing unit, or the like. More particularly, the processing device may be complex instruction set computing (CISC) microprocessor, reduced instruction set computer (RISC) microprocessor, very long instruction word (VLIW) microprocessor, or processor implementing other instruction sets, or processors implementing a combination of instruction sets. The processing device may also be one or more special-purpose processing devices such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), network processor, or the like.

In some examples, the LIDAR control systems 110 may include a signal processing unit 112 such as a DSP. The LIDAR control systems 110 are configured to output digital control signals to control optical drivers 103. In some examples, the digital control signals may be converted to analog signals through signal conversion unit 106. For example, the signal conversion unit 106 may include a digital-to-analog converter. The optical drivers 103 may then provide drive signals to active optical components of optical circuits 101 to drive optical sources such as lasers and amplifiers. In some examples, several optical drivers 103 and signal conversion units 106 may be provided to drive multiple optical sources.

The LIDAR control systems 110 are also configured to output digital control signals for the optical scanner 102. A motion control system 105 may control the galvanometers of the optical scanner 102 based on control signals received from the LIDAR control systems 110. For example, a digital-to-analog converter may convert coordinate routing information from the LIDAR control systems 110 to signals interpretable by the galvanometers in the optical scanner 102. In some examples, a motion control system 105 may also return information to the LIDAR control systems 110 about the position or operation of components of the optical scanner 102. For example, an analog-to-digital converter may in turn convert information about the galvanometers' position to a signal interpretable by the LIDAR control systems 110.

The LIDAR control systems 110 are further configured to analyze incoming digital signals. In this regard, the LIDAR system 100 includes optical receivers 104 to measure one or more beams received by optical circuits 101. For example, a reference beam receiver may measure the amplitude of a reference beam from the active optical component, and an analog-to-digital converter converts signals from the reference receiver to signals interpretable by the LIDAR control systems 110. Target receivers measure the optical signal that carries information about the range and velocity of a target in the form of a beat frequency, modulated optical signal. The reflected beam may be mixed with a second signal from a local oscillator. The optical receivers 104 may include a high-speed analog-to-digital converter to convert signals from the target receiver to signals interpretable by the LIDAR control systems 110. In some examples, the signals from the optical receivers 104 may be subject to signal conditioning by signal conditioning unit 107 prior to receipt by the LIDAR control systems 110. For example, the signals from the optical receivers 104 may be provided to an operational amplifier for amplification of the received signals and the amplified signals may be provided to the LIDAR control systems 110.

In some applications, the LIDAR system 100 may additionally include one or more imaging devices 108 configured to capture images of the environment, a global positioning system 109 configured to provide a geographic location of the system, or other sensor inputs. The LIDAR system 100 may also include an image processing system 114. The image processing system 114 can be configured to receive the images and geographic location, and send the images and location or information related thereto to the LIDAR control systems 110 or other systems connected to the LIDAR system 100.

In operation according to some examples, the LIDAR system 100 is configured to use nondegenerate optical sources to simultaneously measure range and velocity across two dimensions. This capability allows for real-time, long range measurements of range, velocity, azimuth, and elevation of the surrounding environment.

In some examples, the scanning process begins with the optical drivers 103 and LIDAR control systems 110. The LIDAR control systems 110 instruct, e.g., via signal processor unit 112, the optical drivers 103 to independently modulate one or more optical beams, and these modulated signals propagate through the optical circuits 101 to the free space optics 115. The free space optics 115 directs the light at the optical scanner 102 that scans a target environment over a preprogrammed pattern defined by the motion control system 105. The optical circuits 101 may also include a polarization wave plate (PWP) to transform the polarization of the light as it leaves the optical circuits 101. In some examples, the polarization wave plate may be a quarter-wave plate or a half-wave plate. A portion of the polarized light may also be reflected back to the optical circuits 101. For example, lensing or collimating systems used in LIDAR system 100 may have natural reflective properties or a reflective coating to reflect a portion of the light back to the optical circuits 101.

Optical signals reflected back from an environment pass through the optical circuits 101 to the optical receivers 104. Because the polarization of the light has been transformed, it may be reflected by a polarization beam splitter along with the portion of polarized light that was reflected back to the optical circuits 101. In such scenarios, rather than returning to the same fiber or waveguide serving as an optical source, the reflected signals can be reflected to separate optical receivers 104. These signals interfere with one another and generate a combined signal. The combined signal can then be reflected to the optical receivers 104. Also, each beam signal that returns from the target environment may produce a time-shifted waveform. The temporal phase difference between the two waveforms generates a beat frequency measured on the optical receivers 104 (e.g., photodetectors).

The analog signals from the optical receivers 104 are converted to digital signals by the signal conditioning unit 107. These digital signals are then sent to the LIDAR control systems 110. A signal processing unit 112 may then receive the digital signals to further process and interpret them. In some embodiments, the signal processing unit 112 also receives position data from the motion control system 105 and galvanometers (not shown) as well as image data from the image processing system 114. The signal processing unit 112 can then generate 3D point cloud data that includes information about range and/or velocity points in the target environment as the optical scanner 102 scans additional points. The signal processing unit 112 can also overlay 3D point cloud data with image data to determine velocity and/or distance of objects in the surrounding area. The signal processing unit 112 also processes the satellite-based navigation location data to provide data related to a specific global location.

FIG. 2 is a time-frequency diagram 200 of an FMCW scanning signal 201 that can be used by a LIDAR system, such as system 100, to scan a target environment according to some embodiments. In one example, the scanning waveform 201, labeled as f_FM(t), is a sawtooth waveform (sawtooth “chirp”) with a chirp bandwidth Δf_C and a chirp period T_C. The slope of the sawtooth is given as k=(Δf_C/T_C). FIG. 2 also depicts target return signal 202 according to some embodiments. Target return signal 202, labeled as f_FM(t−Δt), is a time-delayed version of the scanning signal 201, where At is the round trip time to and from a target illuminated by scanning signal 201. The round trip time is given as Δt=2R/v, where R is the target range and v is the velocity of the optical beam, which is the speed of light c. The target range, R, can therefore be calculated as R=c(Δt/2). When the return signal 202 is optically mixed with the scanning signal, a range dependent difference frequency (“beat frequency”) Δf_R(t) is generated. The beat frequency Δf_R(t) is linearly related to the time delay Δt by the slope of the sawtooth k. That is, Δf_R(t)=kΔt. Since the target range R is proportional to Δt, the target range R can be calculated as R=(c/2)(Δf_R(t)/k). That is, the range R is linearly related to the beat frequency Δf_R(t). The beat frequency Δf_R(t) can be generated, for example, as an analog signal in optical receivers 104 of system 100. The beat frequency can then be digitized by an analog-to-digital converter (ADC), for example, in a signal conditioning unit such as signal conditioning unit 107 in LIDAR system 100. The digitized beat frequency signal can then be digitally processed, for example, in a signal processing unit, such as signal processing unit 112 in system 100. It should be noted that the target return signal 202 will, in general, also includes a frequency offset (Doppler shift) if the target has a velocity relative to the LIDAR system 100. The Doppler shift can be determined separately, and used to correct the frequency of the return signal, so the Doppler shift is not shown in FIG. 2 for simplicity and ease of explanation. It should also be noted that the sampling frequency of the ADC will determine the highest beat frequency that can be processed by the system without aliasing. In general, the highest frequency that can be processed is one-half of the sampling frequency (i.e., the “Nyquist limit”). In one example, and without limitation, if the sampling frequency of the ADC is 1 gigahertz, then the highest beat frequency that can be processed without aliasing (Δf_Rmax) is 500 megahertz. This limit in turn determines the maximum range of the system as R_max=(c/2)(Δf_Rmax/k) which can be adjusted by changing the chirp slope k. In one example, while the data samples from the ADC may be continuous, the subsequent digital processing described below may be partitioned into “time segments” that can be associated with some periodicity in the LIDAR system 100. In one example, and without limitation, a time segment might correspond to a predetermined number of chirp periods T, or a number of full rotations in azimuth by the optical scanner.

FIG. 3A is a block diagram illustrating an example LIDAR system 300 (e.g., an FMCW LIDAR system) according to the present disclosure. Example system 300 includes an optical scanner 301 to transmit an FMCW (frequency-modulated continuous wave) optical beam 304 and to receive a return signal 313 from reflections of the optical beam 304 from targets such as target 312 in the field of view (FOV) of the optical scanner 301. System 300 also includes an optical processing system 302 to generate a baseband electrical signal 314 in the time domain from the return signal 313, where the baseband electrical signal 314 contains frequencies (e.g., beat frequencies) corresponding to LIDAR target ranges. Optical processing system 302 may include elements of free space optics 115, optical circuits 101, optical drivers 103 and optical receivers 104 in LIDAR system 100. System 300 also includes a signal processing system 303 to measure energy of the baseband electrical signal 314 in the frequency domain, to compare the energy to an estimate of LIDAR system noise, and to determine one or more likelihood metrics for determining whether a signal peak in the frequency domain indicates a detected target. Signal processing system 303 may include elements of signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112 in LIDAR system 100.

FIG. 3B is a block diagram illustrating an example electro-optical system 350. According to some embodiments, electro-optical system 350 includes the optical scanner 301, similar to the optical scanner 102 illustrated and described in relation to FIG. 1. Electro-optical system 350 also includes the optical processing system 302, which as noted above, may include elements of free space optics 115, optical circuits 101, optical drivers 103, and optical receivers 104 in LIDAR system 100.

Electro-optical processing system 302 includes an optical source 305 to generate the frequency-modulated continuous-wave (FMCW) optical beam 304. The optical beam 304 may be directed to an optical coupler 306 that is configured to couple the optical beam 304 to a polarization beam splitter (PBS) 307 and a sample 308 of the optical beam 304 to a photodetector (PD) 309. The PBS 307 is configured to direct the optical beam 304, because of its polarization, toward the optical scanner 301. Optical scanner 301 is configured to scan a target environment with the optical beam 304, through a range of azimuth and elevation angles covering the field of view (FOV) 310 of a LIDAR window 311 in an enclosure 320 of the optical system 350. In FIG. 3B, for ease of illustration, only the azimuth scan is illustrated.

As shown in FIG. 3B, at one azimuth angle (or range of angles), the optical beam 304 passes through the LIDAR window 311 and illuminates a target 312. A return signal 313 from the target 312 passes through LIDAR window 311 and is directed by optical scanner 301 back to the PBS 307.

The return signal 313, which will have a different polarization than the optical beam 304 due to reflection from the target 312, is directed by the PBS 307 to the photodetector (PD) 309. In PD 309, the return signal 313 is optically mixed with the local sample 308 of the optical beam 304 to generate a baseband electrical signal 314 (e.g., beat signal) with a frequency that is proportional to the range of the scanned target. The baseband electrical signal 314 may be generated by the frequency difference between the local sample 308 of the optical beam 304 and the return signal 313 versus time (i.e., Δf_R(t)).

FIG. 4 is a detailed block diagram illustrating an example of the signal processing system 303 of FIG. 3A, which processes the baseband electrical signal 314, according to some embodiments. As noted above, signal processing unit 303 may include elements of signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112 in LIDAR system 100.

According to some embodiments, signal processing system 303 includes an analog-to-digital converter (ADC) 401, a time domain signal processor 402, a block sampler 403, a discrete Fourier transform processor 404, a frequency domain signal processor 405, and a peak search processor 406. The component blocks of signal processing system 303 may be implemented in hardware, firmware, software, or some combination of hardware, firmware and software.

In FIG. 4, the baseband electrical signal 314, which is a continuous analog signal in the time domain, is sampled by ADC 401 to generate a series of time domain samples 315. The time domain samples 315 are processed by the time domain module 402, which conditions the time domain samples 315 for further processing. For example, time domain module 402 may apply weighting or filtering to remove unwanted signal artifacts or to render the signal more tractable for subsequent processing. The output 316 of time domain module 402 is provided to block sampler 403. Block sampler 403 groups the time domain samples 316 into groups of N samples 317 (where N is an integer greater than 1), which are provided to DFT module 404. DFT module 404 transforms the groups of N time domain samples 317 into N frequency bins or subbands (e.g., subband signal spectrum 318) in the frequency domain, covering the bandwidth of the baseband electrical signal 314. The N subband signal spectrum 318 is provided to frequency domain module 405, which conditions the subbands for further processing. For example, frequency domain module 405 may resample and/or average the subband signal spectrum 318 for noise reduction. Frequency domain module 405 may also calculate signal statistics and system noise statistics. The processed subband signal spectrum 319 is then provided to a peak search module 406 that searches for signal peaks representing detected targets in the FOV of the LIDAR system 300.

In some embodiments, the subband signal spectrum 319 provided to the peak search module 406 is the sum of the energy in the target return 313 and all of the noise contributed by the LIDAR system 300 as the target return signal is processed. In some scenarios, electronic systems have sources of noise that limit the performance of those systems by creating a noise floor, which is the combined level of all sources of noise in the system. In order to be detected, a signal in an electronic system such as the subband signal spectrum 319, developed from the baseband electrical signal 314, must be above the noise floor absent specialized signal processing techniques such as signal integration and noise averaging.

Sources of noise in a LIDAR system, such as LIDAR system 300, may include thermal noise, 1/f noise, shot noise, impulse noise, RIN (relative intensity noise associated with lasers), TIA (trans-impedance amplifier) noise, and ADC (analog-to-digital conversion) noise. System noise may be characterized, for example, by its energy versus frequency profile across frequency bins, by its first moments (mean) across the frequency bins, by its second moments (variance) across the frequency bins, by its third moment (asymmetry) across the frequency bins, and/or by its fourth moment (kurtosis, or the sharpness of peaks) across the frequency bins of the frequency spectrum.

FIG. 5A is a diagram 500 illustrating magnitude versus frequency of the subband signal spectrum 319 that includes system noise, shown as a continuous waveform (rather than as discrete frequency bins or subbands) for ease of illustration. Diagram 500 may be generated and/or used by signal processing system 303 of FIG. 3 and peak search module 406 of FIG. 4 to detect a signal peak corresponding to a target detection. Additionally, diagram 500 may be generated and/or used by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112). The frequencies span the range from 0 to Δf_Rmax. In some scenarios, without more information about the subband signal spectrum 319, the peak search module 406 would select the highest signal peak 501 as the return signal that most likely indicates the presence of a target, and not select a lower signal peak 502, for example. However, using a calculated estimate of system noise, the peak search module 406 can be configured to compare the subband signal spectrum 319 to the system noise estimate and could make a more informed selection based on additional selection criteria (e.g., a likelihood metric).

In FIGS. 5A and 5B, signal and noise values are depicted as energy (e.g., intensity) versus frequency contours. However, as previously noted with respect to FIG. 4, the system noise may be additionally characterized by any of its first through fourth moments representing mean energy, energy variance, energy asymmetry and kurtosis versus frequency, respectively. In addition to energy alone, the signal may be characterized in terms of autocorrelation statistics across the frequency bins in the baseband and/or cross-correlation statistics between the signal and the system noise estimate across the frequency bins.

In one example, an estimate of system noise can be obtained by operating a LIDAR system, such as LIDAR system 300, in an anechoic (no-echo) calibration mode where there is no detectable return signal (e.g., return signal 313). This mode of operation generates all of the normal system noise mechanisms and results in a subband signal spectrum that includes energy only from the system noise sources. Accordingly, one or more likelihood metrics for a subband signal spectrum 319 can be generated based on the system noise and any other parameters of the LIDAR system or signal spectrum 319, such as known target reflectivity, internal reflections, known internal and external noise sources, known target locations, and so forth.

FIG. 5B is an energy versus frequency diagram 550 comparing a noise estimate 551 (e.g., the noise estimate as determined in the anechoic (no-echo) calibration mode described above) to the difference between the subband signal spectrum 319 and the noise estimate 551, diagrammed in FIG. 5B as signal minus noise (S−N) 561. In some embodiments, the signal processing system 303 may be configured to use the subband signal (e.g., subband signal spectrum 319) and the system noise estimate (e.g., system noise estimate 651) to generate one or more likelihood metrics (e.g., (S−N)/N, SNR, etc.) used to determine the likelihood that a signal peak in the frequency domain indicates a detected target and to decrease the likelihood that a signal peak in the frequency domain from a false target will be interpreted as a real target. Such likelihood metrics may be used for both detection thresholding and peak selection.

In the example of FIG. 5B, the peak search module 406 may be configured to select the signal peak with respect to one of the determined likelihood metrics. For example, the likelihood metric may be a signal minus noise to noise ratio (“(S−N)/N”). The peak search module 406 may thus select the highest non-negative signal minus noise to noise ratio (S−N)/N). Under this selection criteria, signal peak 552 with (S−N)/N 553 would be selected over signal peak 554 with (S−N)/N 555 because (S−N)/N 553 is larger than (S−N)/N 555. In another example, the peak search module 406 may use the likelihood metric of signal-minus noise (“S−N”) for peak selection in which case peak 552 would still be selected over peak 554. Any other likelihood metric may also be used to select a peak, such as SNR, raw intensity (e.g., signal 319), or other metric generated based on the LIDAR system and the subband signal spectrum 319.

FIG. 6A and 6B depict likelihood metric-frequency diagrams 600 and 650 illustrating examples of using a single likelihood metric for thresholding and peak selection. Diagrams 600 and 650 may be generated and/or used by signal processing system 303 of FIG. 3 or peak search module 406 of FIG. 4. Additionally, diagrams 600 and 650 may be generated and/or used by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112).

As discussed above, the signal processing system 303 of FIG. 3A and 4 may generate a waveform in a frequency domain in which each frequency has an associated energy value (e.g., intensity). The signal processing system 303 (e.g., via peak search module 406) may generate one or more likelihood metrics for each frequency in a spectrum of frequencies (e.g., subband frequency spectrum 319) based on the energy values. A likelihood metric may be a metric generated for each of the frequencies in the spectrum to maximize the likelihood of selecting a peak corresponding to an actual target detection while reducing the likelihood of false alarm detection (e.g., due to noise). For example, the likelihood metric may be a signal-to-noise ratio (SNR), intensity, signal-minus-noise, signal-minus-noise to noise ratio, etc., as described above with respect to FIGS. 5A-B. Whatever likelihood metric is selected, a peak search module (e.g., peak search module 406 of FIG. 4.) may select the frequency, or frequency bin, in the spectrum that has the highest value (e.g., the highest peak) with respect to the likelihood metric and that also exceeds a threshold value for the likelihood metric.

As depicted in FIG. 6A, a threshold M_T(f) 605 for the likelihood metric M(f) may be set at a single threshold value across all the frequencies in the spectrum of frequencies. The likelihood metric M(f) may be set at a value to maximize probability of target detection while minimizing the number of false alarm detections (i.e., detections that are not actual targets). In another example, the threshold 605 may vary across the frequencies of the spectrum. For example, the threshold 605 may be adjusted for frequencies corresponding to previously detected targets or interferers. According to the depicted example, the selected peak 610 may be selected because it is both the highest peak in the diagram with respect to the likelihood metric M(f) and also exceeds the threshold M_T(f) 605. Thus, the selected peak 610 is selected as the frequency associated with a target detection. The frequency of the selected peak 610 can then be used to calculate properties of the target such as distance to the target, velocity of the target, and reflectivity of the target. In one embodiment the peak may be selected based on the following equation: P=arg max_fM(f) subject to M(f)≥M_T(f). In other words, the frequency corresponding to the maximum value for the likelihood metric M(f) is selected for which the likelihood metric M(f) also is equal to or greater than the threshold at that frequency.

In the example depicted by FIG. 6B, the threshold M_T(f) 615 is set at a value that none of the frequencies of the spectrum exceed with respect to the likelihood metric M(f). Accordingly, none of the peaks in the diagram 650 will result in a peak detection. In one example, the peak selection algorithm may first apply the threshold M_T(f) to the diagram to determine if any of the frequencies exceed the threshold for the likelihood metric M(f). The peak selection algorithm may then select a peak from the frequencies that exceed the threshold. In another example, the peak selection algorithm may first select the highest peak with respect to the likelihood metric M(f) and then apply the threshold M_T(f) to determine if the selected peak exceeds the threshold M_T(f). Therefore, only the highest peak that is also above the threshold is selected and used for target detection. Thus, as depicted in diagram 650 of FIG. 6B, after the highest peak is initially selected, the peak will not be used for target detection if it does not exceed the threshold M_T(f) 615.

FIG. 7 flowchart illustrating a method 700 of peak detection using a single likelihood metric for thresholding and peak selection in a LIDAR system, such as LIDAR system 100 or LIDAR system 300. Method 700 may be performed by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112) and/or peak search module 406 of FIG. 4.

Method 700 begins at operation 710, where processing logic (e.g., peak search module 406) determines a likelihood metric for a spectrum of frequencies in a frequency domain waveform (e.g., subband signal spectrum 319). The likelihood metric may be an intensity, an SNR, a signal minus noise to noise ratio ((S−N)/N), or any other likelihood ratio used to increase target detections and reduce false alarm detections.

At operation 720, the processing logic (e.g., peak search module 406) identifies one or more frequencies in the frequency domain waveform that exceed a threshold value for the likelihood metric. In one example, the processing logic may filter out the frequencies with a value for the likelihood metric that is below the threshold value. Thus, the remaining frequencies with a value for the likelihood metric that exceed the threshold value may remain and be used at operation 730 for peak selection.

At operation 730, the processing logic selects a peak frequency from the frequency domain waveform corresponding to the frequency with the highest value for the likelihood metric, as depicted in FIG. 6B. In one example, the same likelihood metric is used for thresholding at operation 720 and for peak selection at operation 730. In some embodiments, the processing logic may perform the thresholding of operation 720 prior to performing the peak selection of operation 730. In such embodiments, the peak frequency may be selected from the frequencies remaining after thresholding. In alternative embodiments, the processing logic may first select a peak at operation 730 and then perform the thresholding of operation 720 to determine whether the selected peak exceeds the threshold. If the selected peak exceeds the threshold, the peak is selected for use in target detection. If the selected peak does not exceed the threshold, then no peak is selected from the current frequency domain waveform. Thus, no target would be detected during the processing of the frequencies included in the particular time domain being processed.

FIGS. 8A and 8B depict metric-frequency diagrams 800 and 850 illustrating an example method of peak detection using a first likelihood metric for thresholding and a second different likelihood metric for performing peak selection. Diagrams 800 and 850 may be generated and/or used by signal processing system 303 of FIG. 3 or peak search module 406 of FIG. 4. Additionally, diagrams 800 and 850 may be generated and/or used by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112).

In some embodiments, a peak search module may first perform a thresholding operation on the frequencies in the waveform to filter out any frequencies that are below a threshold value for a thresholding metric. The thresholding metric may be intensity, SNR, or any other likelihood metric. The peak search module may then determine a different likelihood metric to perform peak selection on the remaining frequencies that were not filtered out by the thresholding operation. The frequency of the remaining frequencies that has the highest value for the peak selection metric is selected for target detection.

As depicted in diagram 800 of FIG. 8A, the frequencies in the frequency domain waveform that are less than the threshold M_T(f) 805 are filtered out. However, the peak is not selected using the thresholding metric M₂(f). Instead, a peak signal frequency may be selected based on the peak selection metric M₁(f), as shown in diagram 850 of FIG. 8B, which may be a different metric than the thresholding metric M₂(f). For example, peak 802A and 804A may each include frequencies with signals that have a value for the thresholding metric M₂(f) that is above the threshold M_T(f). Therefore, the frequencies of the frequency spectrum corresponding to peak 802A and 804A may be used for peak selection. However, the peak selection metric M₁(f) may be determined for the frequencies corresponding to peaks 802A and 804A. For example, the peak 802B with respect to the peak selection metric M₁(f) may correspond to the frequencies of peak 802A with respect to the thresholding metric M₂(f). Similarly, the selected peak 804B with respect to the peak selection metric M₁(f) may correspond to the frequencies of 804A with respect to the thresholding metric M₂(f). Therefore, although peak 802A may be the highest peak for the thresholding metric M₂(f), peak 804B is the highest peak with respect to the peak selection metric M₁(f) for the remaining frequencies that were not filtered out during thresholding. Accordingly, peak 804B would be selected for peak detection. Thus, the highest peak with respect to the thresholding metric M₂(f) may not be the resulting selected peak with respect to the peak selection metric M₁(f).

In one embodiment, the peak thresholding and selection using different likelihood metrics may be represented by the following equation: P=arg max_fM₁(f) subject to M₂(f)≥M_T(f). In other words, the peak that has the maximum value for the peak selection metric M1(f) is selected that also has a value of the thresholding metric M2(f) that is greater than or equal to the threshold at that frequency. In another embodiment, different peak selection metrics may be defined based on different thresholds. For example, if no peaks exceed the initial threshold MT(f) then a lower threshold may be selected and a different peak selection metric used for peak selection. In some embodiments, the additional threshold values and peak selection metric may be represented by the following equation: P=arg max_fM₁(f) subject to M₂(f)≥M_T(f), however if there are no f such that M2(f)≥M_T(f), then P=arg max_fM₃(f) subject to M_T(f)≥M_T(f) where MT(f)≥M_T(f). This may be generalized for any number of thresholds and any number of additional likelihood metrics.

FIG. 9 flowchart illustrating a method 900 of peak detection using different likelihood metrics for thresholding and peak selection in a LIDAR system, such as LIDAR system 100 or LIDAR system 300. Method 700 may be performed by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112) and/or peak search module 406 of FIG. 4.

Method 900 begins at operation 910, where processing logic (e.g., peak search module 406) determines a first likelihood metric for a spectrum of frequencies in a frequency domain waveform. The first likelihood metric may be signal-to-noise ratio (SNR), intensity, signal-minus-noise, signal-minus-noise to noise ratio, or any other likelihood metric determined for the spectrum of signal frequencies. The first likelihood metric may be used for detection thresholding to filter out signals that correspond to noise and to bias peak detection for actual targets based on known information about the targets. For example, if targets of interest for the LIDAR system have a known minimum reflectivity then the first likelihood metric may be selected as intensity and a threshold value selected so that any noise that is below the minimum reflectivity are filtered out from peak selection (e.g., because it is known that targets should have an intensity above the threshold). In another example, if the LIDAR system has large variations in the estimated noise level then the first likelihood metric may be selected as SNR to filter out peaks corresponding to noise events.

At operation 920, the processing logic (e.g., peak search module 406) identifies one or more frequencies in the frequency domain waveform that exceed a threshold value for the first likelihood metric. In one example, the processing logic may filter out any frequencies that do not exceed the threshold value. In another example, the processing logic may select each of the frequencies that are equal to or exceed the threshold value. Thus, the remaining frequencies are the frequencies that are equal to or exceed the threshold value with respect to the first likelihood metric.

At operation 930, the processing logic (e.g., peak search module 406) determines a second likelihood metric for the spectrum of frequencies in the frequency domain waveform. Similar to the first likelihood metric, the second likelihood metric may be signal-to-noise ratio (SNR), intensity, signal-minus-noise, signal-minus-noise to noise ratio, or any other likelihood metric. However, depending on the selection for the first likelihood metric, the second likelihood metric may be selected to be a different likelihood metric than the first likelihood metric. For example, if the first likelihood metric is selected to be intensity then the second likelihood metric may be selected as SNR. Alternatively, if the first likelihood metric is selected as SNR then the second likelihood metric may be selected as intensity.

At operation 940, the processing logic (e.g., peak search module 406) selects a peak signal frequency from the frequency domain waveform corresponding to the frequency with the highest value for the second likelihood metric based on the one or more frequencies in the frequency domain waveform that exceed the threshold value for the first likelihood metric. Accordingly, the peak frequency is selected from the frequencies that were either selected, or not filtered out, during detection thresholding (operation 920). Because a different likelihood metric is used for peak detection than was used for thresholding, any biases that may be present in the first likelihood metric may be offset by the second likelihood metric. Accordingly, using two separate likelihood metrics for thresholding and peak selection may increase the probability of proper target detections and reduce false alarms.

FIG. 10 depicts a metric-frequency diagram 1000 illustrating an example method of peak detection using thresholding of peaks with minimum threshold widths. Diagram 1000 may be generated and/or used by signal processing system 303 of FIG. 3 or peak search module 406 of FIG. 4. Additionally, diagram 1000 may be generated and/or used by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112).

In some embodiments, during the thresholding operation as described above with respect to FIGS. 6A and 8A, an additional constraint of peak width may be added to filter out uncorrelated peaks (e.g., peaks with small widths that are likely to be noise events). For example, as depicted in diagram 1000 of FIG. 10, the peak search module may include a minimum peak width that exceeds a threshold M_T(f) for a thresholding metric M₂(f).

In one example, the peak search module may first identify a peak that exceeds the threshold M_T(f) value for the thresholding metric M₂(f). The peak search module may then determine if additional frequencies adjacent to the identified peak are also above the threshold M_T(f). For example, the peak search module may determine whether a particular range of frequencies [delta f] (e.g., a certain number of frequency bins) on either side of the identified peak. Accordingly, the identified peak may be filtered from the peak selection unless the peak has a minimum width of 2Δf that exceeds the threshold M_T(f), where f is an arbitrary frequency range or number of frequency bins. As described above with respect to FIG. 6B and 8B, the peak search module may then select a peak from the remaining frequencies that were not filtered out during thresholding (i.e., that exceed the threshold with the minimum width). The addition of multi-frequency bin thresholding may be represented by the following equation: P=arg max_f M₁(f) subject to M₂(f′)≥M_T(f′), and f′ includes the set [f−Δf, f+Δf]. In other words, a band [f−Δf, f+Δf] around the peak needs to be above the threshold M_T(f).

FIG. 11 flowchart illustrating a method 1100 of peak detection using thresholding with a minimum peak width in a LIDAR system, such as LIDAR system 100 or LIDAR system 300. Method 1100 may be performed by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112) and/or peak search module 406 of FIG. 4.

Method 1100 begins at operation 1110, where processing logic (e.g., peak search module 406) determines a first likelihood metric for frequencies in a frequency domain waveform. The likelihood metric may be an intensity, an SNR, a signal minus noise to noise ratio ((S−N)/N), or any other likelihood ratio used to increase target detections and reduce false alarm detections as described herein.

At operation 1120, the processing logic (e.g., peak search module 406) identifies one or more frequencies in the frequency domain waveform that exceed a threshold value for the first likelihood metric. In one example, the processing logic may filter out any frequencies that do not exceed the threshold value. In another example, the processing logic may select each of the frequencies that are equal to or exceed the threshold value. Thus, the remaining frequencies are the frequencies that are equal to or exceed the threshold value with respect to the first likelihood metric.

At operation 1130, the processing logic (e.g., peak search module 406) selects a peak frequency from the frequency domain waveform corresponding to the frequency with the highest value for the first likelihood metric or a second likelihood metric. In one example, the same first likelihood metric may be used to select the peak. In another example, a second likelihood metric, different from the first likelihood metric, may be used for peak selection.

At operation 1140, the processing logic (e.g., peak search module 406) determines whether a minimum band of frequencies adjacent to the peak frequency exceeds the threshold value for the first likelihood metric. In other words, the processing logic determines whether the width of the selected peak exceeds a threshold width. In one embodiment, the processing logic determines whether the width of the portion of the peak that exceeds the threshold is wider than a minimum width. If the band of frequencies corresponding to the peak has a width larger than the minimum width, the peak is selected for target detection. Otherwise, if the peak has a width less than the minimum width the peak is not selected. In some embodiments, a wider peak may be correlated with a detected target while a narrow peak may be correlated with a noise event. Thus, only selecting peaks that exceed a particular width may reduce false alarm detections.

FIGS. 12A and 12B depict likelihood metric-frequency diagrams 1200 and 1250 illustrating examples of peak detection using a weighted likelihood metric for peak selection. Diagrams 1200 and 1250 may be generated and/or used by signal processing system 303 of FIG. 3 or peak search module 406 of FIG. 4. Additionally, diagrams 1200 and 1250 may be generated and/or used by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112).

As discussed above, the signal processing system 303 of FIG. 3A and 4 may generate a waveform in a frequency domain in which each frequency of a spectrum of frequencies has an associated energy value (e.g., intensity). The signal processing system 303 (e.g., via peak search module 406) may generate one or more likelihood metrics for each frequency in the diagram based on the energy values. A likelihood metric may be a metric generated for each of the frequencies to increase the likelihood of peak detection while reducing the likelihood of false alarm detection (e.g., due to noise). For example, the likelihood metric may be a signal-to-noise ratio (SNR), intensity, signal-minus-noise, signal-minus-noise ratio, etc., as described above with respect to FIGS. 5A-B. Whatever likelihood metric is selected, in some embodiments a peak search module (e.g., peak search module 406) may select the frequency, or frequency bin, with the maximum peak with respect to the likelihood metric.

As depicted in FIG. 12A, a likelihood metric M(f) may be generated in a frequency domain for a set of frequencies corresponding to a distance resolution of the LIDAR system. The likelihood metric may result in one or more peaks (e.g., peaks 1205 and 1210) that may correspond to a target detection or potential noise. For example, peak 1210 may correspond to a target detection while peak 1205 may be a noise event. However, because the two peaks 1205 and 1210 are substantially the same magnitude with respect to the likelihood metric M(f) either of the peaks may be selected for use in target detection and ranging. Therefore, in such an instance the likelihood metric M(f) may be unreliable for peak selection. Thus, the likelihood metric M(f) may be modified based on known properties of the LIDAR system of target to increase the likelihood of proper target detection.

Referring to FIG. 12B, the likelihood metric M(f) of diagram 1200 may be weighted to generate a weighted likelihood matric M_W(f) to bias the peak selection toward higher frequencies (e.g., peak 1210 over peak 1205). For example, previous detections or know attributes of the LIDAR system may indicate that peaks at higher frequencies are more likely to correspond to actual target detections while peaks at lower frequencies are more likely to be a noise event or other interferer. Thus, the weighted likelihood metric M_W(f) may be weighted in a manner that that biases peak selection toward higher frequencies despite similar peaks for the original likelihood metric M(f) of diagram 1200. In an another example, if lower frequencies are more likely to be actual target detections while higher frequencies are more likely noise events, the weighted likelihood metric M_W(f) may bias selection toward lower frequencies, in effect reversing diagram 1250.

Although depicted in diagram 1250 of FIG. 12B as linearly weighted (e.g., M_W(f)−M(f)+Δf) based on frequency, the likelihood metric M(f) may be weighted in any other manner to bias detections based on a-priori information. For example, the likelihood metric M(f) may be weighted in a multiplicative manner (e.g., M_W(f)=M(f)*W(f), where W(f) is a weighting for the frequency f) or using any other general weighting function (e.g., M_W(f)=W(f, M(f)), where W(f, M(f)) is a weighting function with parameters of frequency f and likelihood metric M(f)). Additionally, the weighting function may depend on any variables of the frequency domain diagram such as peak width or based on whether the scene is dynamic or stationary.

The general weighting function may also be extended to use additional parameters, such as higher order information such as known interferers, previous target detections, or estimated target locations, to weight the likelihood metric. For example, the weighting function may multiply the likelihood metric by zero for frequencies corresponding to known noise events or interferers. Similarly, the weighting function may increase the weighting of the likelihood metric for estimated target locations which may be based in part on previous target detections to increase detection likelihood for such frequencies.

FIG. 13 flowchart illustrating a method 1300 of peak detection using a weighted likelihood metric for peak selection in a LIDAR system, such as LIDAR system 100 or LIDAR system 300. Method 1300 may be performed by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112) and/or peak search module 406 of FIG. 4.

Method 1300 begins at operation 1310, where processing logic (e.g., peak search module 406) determines a likelihood metric for a spectrum of frequencies in a frequency domain waveform. The likelihood metric may be an intensity, an SNR, a signal minus noise to noise ratio ((S−N)/N), or any other likelihood metric used to increase target detections and reduce false alarm detections. The likelihood metric may be determined based on the initial frequency domain waveform and noise estimates as depicted in FIGS. 5A and 5B.

At operation 1320, the processing logic (e.g., peak search module 406) applies a weight to the likelihood metric for each frequency in the spectrum of frequencies of the frequency domain waveform. The weighting of the likelihood metric may bias or favor certain detections that are known to have a higher probability of being a proper target detection. For example, the LIDAR system may consistently have noise events in lower frequencies while target detections are generally in the higher frequency range. Thus, the processing logic may bias peak selection toward the higher frequencies by weighting higher frequencies more heavily than lower frequencies. In another example, where previous detections or other information indicate frequencies where a target detection is likely, the processing logic may weight a frequency band corresponding to the previously detected target more heavily. Weights may be applied according to any other scenario to bias detections based on known information about the LIDAR system or expected targets.

In some embodiments, the weight may be a simple offset adding a frequency dependent offset to the likelihood metric for each frequency. In another example, the weight may be a multiplicative weight in which the likelihood metric value for each frequency is multiplied by a frequency dependent weight. In another example, a general weighting function may be applied to the frequency spectrum where the general weighting function takes a particular frequency and the likelihood metric for that frequency as inputs and generates a weighted likelihood metric for the frequency. In some embodiments, the applied weight may be dependent on variables other than frequency, such as peak width or a correlation metric, etc. In some embodiments, the processing logic may determine whether the scene viewed by the LIDAR system is stationary or dynamic and apply a weight function accordingly. For example, if the scene is dynamic, such as if the LIDAR system is in motion, then the processing logic may weight frequencies corresponding to closer targets more heavily to favor detection of targets that are imminently in contact with the LIDAR system (e.g., such as obstacles of an automated driving vehicle). Additionally, based on known information about a scene or the LIDAR system, certain frequencies may be known to correspond to interferers or other noise around which the processing logic may reduce the weight, or multiply the metric by zero, for such frequencies to avoid detections of the interferer. Similarly, based on higher order information such as previous detections, known target locations, etc. the processing logic may determine the weighting function to be higher for those frequencies to bias detection toward such frequencies.

At operation 1330, the processing logic (e.g., peak search module 406) selects a peak frequency from the frequency domain waveform corresponding to the frequency with the highest value for the weighted likelihood metric. In some embodiments, the weighted likelihood metric may result in a highest peak that is different from the highest peak in the frequency domain waveform using the original likelihood metric.

FIG. 14 depicts a likelihood metric-frequency diagram 1400 illustrating an example method of peak detection with filtering of frequency bands prior to peak selection. Diagram 1400 may be generated and/or used by signal processing system 303 of FIG. 3 or peak search module of FIGS. 4 and 5. Additionally, diagram 1400 may be generated and/or used by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112).

In some embodiments, a peak search module (e.g., peak search module 406 of FIG. 4) may first perform a thresholding operation on the frequencies in the waveform to filter out any frequencies that are below a threshold value for a thresholding metric (e.g., a likelihood metric for thresholding). The thresholding metric may be intensity, SNR, or any other likelihood metric. The peak search module may then perform peak selection based on a peak selection metric (e.g., a likelihood metric for peak selection) on the remaining frequencies that were not filtered out by the thresholding operation. The frequency of the remaining frequencies that has the highest value for the peak selection metric may be selected for target detection.

As depicted in diagram 1400 of FIG. 14, several frequency bands may be filtered out from peak selection. For example, a minimum frequency 1402 may be defined and the frequencies below the minimum frequency may be removed (e.g., filtered out) from peak selection. The minimum frequency may be defined to filter out detections due to particles accumulated on the sensor window or low frequency noise from optical or electrical sources. In another example, a band around a previously selected peak (e.g., previously selected peak 1404) may be filtered out so that the side lobes of the peak are not selected as an additional detection. In another example, a band around a known interferer (e.g., known interferer 1406) that may be known based on an estimated noise level may be filtered out from peak selection. The known interferer may have an intensity that is unknown or time-varying and therefore may need to be filtered out prior to peak selection. The center frequency (e.g., peak) of the known interferer may vary over time and thus the filtered frequency band may dynamically move with the interferer. In yet another example, a maximum frequency (e.g., max frequency 1408) may be defined and the frequencies above the maximum frequency may be filtered out for peak selection. Accordingly, only peaks in the frequency ranges that are not filtered out may be selected for target detection. As depicted, in diagram 1400 of FIG. 14, no peak would be selected because the previously selected peak 1404 and the known interferer 1406 would be filtered out, thus avoiding a false alarm detection or redundant detection.

FIG. 15 flowchart illustrating a method 1500 of peak detection via frequency band filtering, such as LIDAR system 100 or LIDAR system 300. Method 1500 may be performed by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112) and/or peak search module 406 of FIG. 4.

Method 1500 begins at operation 1510, where processing logic (e.g., peak search module 406) determines a likelihood metric for the spectrum of frequencies of the frequency domain waveform. The likelihood metric may be an intensity, an SNR, a signal minus noise to noise ratio ((S−N)/N), or any other likelihood ratio used to increase target detections and reduce false alarm detections.

At operation 1520, the processing logic (e.g., peak search module 406) filters out a band of frequencies in the spectrum of frequencies based on previously collected information associated with the target or the LIDAR system. The processing logic may filter out one or more frequency bands. The frequency bands filtered out may be statically selected (e.g., the selected frequencies filtered remain the same over time) or they may be dynamically determined over time where each successive peak selection may operate on a different set of frequencies. The previously collected information associated with the target may include known properties of the target such as target reflectivity, target position (e.g., based on previous detections of the target), whether the target is a moving target, or any other target properties. The previously collected information for the LIDAR system may include noise properties of the LIDAR system (e.g., estimated noise levels), known interferers of the LIDAR system, or any other properties of the LIDAR system that may have an effect on detected signals.

For example, the processing logic may determine a minimum frequency below which frequencies are filtered out to prevent detections from particulars accumulated on the sensor window or form low-frequency noise from optical or electrical sources. In another example, a band of frequencies around a known interferer whose intensity is unknown or time-varying may be filtered out to avoid selecting the interferer as a target detection. In another example, the processing logic may filter out a band of frequencies around a previously selected peak to avoid selecting side lobes of that peak as a separate peak.

At operation 1530, the processing logic (e.g., peak search module 406) selects a peak frequency from remaining frequencies of the spectrum of frequencies that were not filtered out. The peak frequency may correspond to a frequency with the highest value for the likelihood metric.

FIG. 16 depicts a metric-frequency diagram 1600 illustrating an example of using band specific peak detection techniques for frequency bands in a peak search. Diagram 1600 may be generated and/or used by signal processing system 303 of FIG. 3 or peak search module 406 of FIG. 4. Additionally, diagram 1000 may be generated and/or used by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112) for target detection and ranging.

As depicted in FIG. 16, a spectrum of frequencies of a waveform diagram may be separated into two or more frequency bands in which a separate peak detection technique may be used to identify a peak within that band. For example, a first peak detection method may be performed within the first frequency band 1604, a second peak detection method may be performed within the second frequency band 1606, and a third peak detection method may be performed within the third frequency band 1608. Any one of the peak detection methods described herein may be performed within each band. Additionally, any combination of the peak detection methods described above may be performed within or across each of the frequency bands. Additionally, although depicted as including three frequency bands, the frequency spectrum may be separated into any number of frequency bands. Furthermore, one or more frequency bands may be filtered out in conjunction with performing different peak detection method in the remaining frequency bands. For example, as depicted in FIG. 16, frequencies below a minimum frequency may be filtered out (e.g., filtered frequency 1602). Any other frequency band, as described above, may be filtered out based on a-priori information about the LIDAR system or targets of the LIDAR system.

In some embodiments, the peak search module may use a-priori information about the LIDAR system or targets of the LIDAR system to determine the peak detection strategies to be used for each frequency band of the spectrum. For example, targets within a certain range (e.g., the first frequency band 1604) may have a minimum target intensity. Therefore, the thresholding metric and peak selection metric may be selected in the manner described with respect to FIGS. 8A and 8B (e.g., set the thresholding metric as intensity with a threshold value corresponding to the minimum intensity of the targets in the range corresponding to the first frequency band 1604). Furthermore, targets in a different range (e.g., frequency band 1606), such as beyond the range discussed above, may have a different minimum intensity. The threshold for the thresholding metric may therefore be set at a different value than the threshold set for the first frequency band 1604.

In some embodiments, the estimated noise levels of the frequency spectrum may have different properties in the different frequency bands. Thus, the peak detection method for each band may further be selected based on the noise properties within each frequency band. For example, a known interferer may be located within a frequency band (e.g., the second frequency band 1604). Accordingly, the peak search module may filter out, attenuate, or otherwise suppress that frequency band, use multi-frequency bin thresholding (e.g., require a minimum peak width for detection), select the thresholding and peak selection metrics to avoid selection of the interferer, etc. Any further combination of the peak detection methods, different thresholding and peak selection metrics, minimum peak width thresholds, or any other variables of the peak detection methods described above may be used for peak detection in the different frequency bands of the frequency spectrum.

Using different peak detection methods in several different frequency bands may result in multiple peaks being detected (e.g., one peak for each frequency band). Therefore, the peak search module may then select a final overall peak from the peaks detected in each frequency band. As described in more detail below with respect to FIG. 17, the peak search module may use any of the peak detection methods described herein to select the final peak or define other ways to prioritize the final peak selection.

FIG. 17 is a flowchart illustrating a method 1700 of peak detection using band specific peak detection techniques for frequency bands in a peak search of a LIDAR system, such as LIDAR system 100 or LIDAR system 300. Method 1700 may be performed by one or more of the elements of LIDAR system 100 of FIG. 1 (e.g., signal conversion unit 106, signal conditioning unit 107, LIDAR control systems 110 and signal processing unit 112) and/or peak search module 406 of FIG. 4.

Method 1700 begins at operation 1710, where processing logic (e.g., peak search module 406) generates a frequency domain waveform based on a baseband electrical signal. The frequency domain waveform may include an energy (e.g., intensity) for a spectrum of frequencies. For example, the frequency domain waveform may include several frequency bins, each of which has an associated energy or intensity value.

At operation 1720, the processing logic (e.g., peak search module 406) separates the spectrum of frequencies in the frequency domain waveform into a plurality of frequency bands comprising at least a first frequency band and a second frequency band. Each of the frequency bands may include a subset of the frequencies (e.g., frequency bins) included in the frequency domain waveform. The processing logic may determine the frequency ranges for each of the frequency bands based on information known about the LIDAR system and one or more intended targets of the LIDAR system.

At operation 1730, the processing logic performs a first peak detection method within the first frequency band of the frequency spectrum. The first peak detection method may include any of the above described peak detection methods, such as thresholding and peak selection using one likelihood metric, thresholding and peak selection using separate likelihood metrics, thresholding and peak selection using a weighted likelihood metric, and frequency band filtering, or any other peak detection technique. The first peak detection method may select a first intermediate peak frequency from within the first frequency band. The first intermediate peak frequency may correspond to a frequency within the first frequency band that has the highest frequency and also satisfies all thresholds associated with the first peak detection technique.

At operation 1740, the processing logic (e.g., peak search module 406) performs a second peak detection method within the second frequency band of the spectrum of frequencies. The second peak detection method may include any of the above described peak detection methods, such as thresholding and peak selection using one likelihood metric, thresholding and peak selection using separate likelihood metrics, thresholding and peak selection using a weighted likelihood metric, and frequency band filtering. The second peak detection method may select a second intermediate peak frequency from within the second frequency band. The first and second peak detection techniques may be selected for their corresponding frequency bands based on properties of the LIDAR system and/or the targets of the LIDAR system.

At operation 1750, the processing logic (e.g., peak search module 406) selects a peak frequency from the spectrum of frequencies in the frequency domain waveform based at least in part on the first peak detection method within the first frequency band and the second peak detection method within the second frequency band. Once the peak detection techniques have been performed within each of the frequency bands, multiple intermediate peaks will have been selected, one for each distinct band. Thus, the processing logic may further select a final peak from the intermediate peaks.

In one example, the processing logic may select the final peak based on a predefined priority of frequency bands (e.g., lowest frequency to highest, highest to lowest, etc.). In another embodiment, the processing logic may perform a third peak detection (e.g., one of the above described techniques) on the intermediate peaks to select the final peak. For example, the processing logic may determine whether a value of a likelihood metric associated with the first intermediate peak or second intermediate peak exceeds a threshold and then select the peak frequency from the first intermediate peak and the second intermediate peak that has a highest value for the likelihood metric. As such, the final selected peak may be the highest peak with respect to whichever likelihood metric is used for performing the final peak selection. The processing logic may then determine one or more properties (e.g., range and velocity) of a target based at least in part on the selected peak frequency.

FIG. 18 is a block diagram of a processing system 1800 (e.g., similar to signal processing system 303 illustrated and described above with respect to FIG. 4) in a LIDAR system such as LIDAR system 100 or LIDAR system 300. Processing system 1800 includes a processing device 1801, which may be any type of general purpose processing device or special purpose processing device designed for use in the LIDAR system. Processing device 1801 is coupled with a memory 1802, which can be any type of non-transitory computer-readable medium (e.g., RAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic disk memory or optical disk memory) containing instructions that, when executed by processing device 1801 in the LIDAR system, cause the LIDAR system to perform the method described herein. In particular, memory 1802 includes instructions 1804 to generate a frequency domain waveform based on the baseband signal in the time domain, wherein the frequency domain waveform comprises a spectrum of frequencies; instructions 1806 to separate the spectrum of frequencies in the frequency domain waveform into a plurality of frequency bands comprising at least a first frequency band and a second frequency band; instructions 1808 to perform a first peak detection within the first frequency band of the spectrum of frequencies; instructions 1810 to perform a second peak detection within the second frequency band of the spectrum of frequencies, wherein the first peak detection and second peak detection comprise different peak detection techniques; and instructions 1812 to select a peak frequency from the spectrum of frequencies in the frequency domain waveform based at least in part on the first peak detection within the first frequency band and the second peak detection within the second frequency band.

The preceding description sets forth numerous specific details such as examples of specific systems, components, methods, and so forth, in order to provide a thorough understanding of several examples in the present disclosure. It will be apparent to one skilled in the art, however, that at least some examples of the present disclosure may be practiced without these specific details. In other instances, well-known components or methods are not described in detail or are presented in simple block diagram form in order to avoid unnecessarily obscuring the present disclosure. Thus, the specific details set forth are merely exemplary. Particular examples may vary from these exemplary details and still be contemplated to be within the scope of the present disclosure.

Any reference throughout this specification to “one example” or “an example” means that a particular feature, structure, or characteristic described in connection with the examples are included in at least one example. Therefore, the appearances of the phrase “in one example” or “in an example” in various places throughout this specification are not necessarily all referring to the same example.

Although the operations of the methods herein are shown and described in a particular order, the order of the operations of each method may be altered so that certain operations may be performed in an inverse order or so that certain operation may be performed, at least in part, concurrently with other operations. Instructions or sub-operations of distinct operations may be performed in an intermittent or alternating manner.

The above description of illustrated implementations of the invention, including what is described in the Abstract, is not intended to be exhaustive or to limit the invention to the precise forms disclosed. While specific implementations of, and examples for, the invention are described herein for illustrative purposes, various equivalent modifications are possible within the scope of the invention, as those skilled in the relevant art will recognize. The words “example” or “exemplary” are used herein to mean serving as an example, instance, or illustration. Any aspect or design described herein as “example” or “exemplary” is not necessarily to be construed as preferred or advantageous over other aspects or designs. Rather, use of the words “example” or “exemplary” is intended to present concepts in a concrete fashion. As used in this application, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or”. That is, unless specified otherwise, or clear from context, “X includes A or B” is intended to mean any of the natural inclusive permutations. That is, if X includes A; X includes B; or X includes both A and B, then “X includes A or B” is satisfied under any of the foregoing instances. In addition, the articles “a” and “an” as used in this application and the appended claims should generally be construed to mean “one or more” unless specified otherwise or clear from context to be directed to a singular form. Furthermore, the terms “first,” “second,” “third,” “fourth,” etc. as used herein are meant as labels to distinguish among different elements and may not necessarily have an ordinal meaning according to their numerical designation.

Claims

1. A light detection and ranging (LIDAR) system, comprising: an optical scanner to transmit an optical beam towards, and receive a return signal from, a target;an optical processing system coupled to the optical scanner to generate a baseband signal in a time domain from the return signal, the baseband signal comprising frequencies corresponding to LIDAR target ranges; anda signal processing system coupled to the optical processing system, comprising: a processor; anda memory operatively coupled to the processor, the memory to store instructions that, when executed by the processor, cause the LIDAR system to: generate a frequency domain waveform based on the baseband signal in the time domain, wherein the frequency domain waveform comprises a spectrum of frequencies;separate the spectrum of frequencies in the frequency domain waveform into a plurality of frequency bands comprising at least a first frequency band and a second frequency band;perform a first peak detection within the first frequency band of the spectrum of frequencies;perform a second peak detection within the second frequency band of the spectrum of frequencies, wherein the first peak detection and second peak detection comprise different peak detection techniques; andselect a peak frequency from the spectrum of frequencies in the frequency domain waveform based at least in part on the first peak detection within the first frequency band and the second peak detection within the second frequency band.

2. The LIDAR system of claim 1, wherein the processor is further to: determine the first peak detection for the first frequency band and the second peak detection for the second frequency band based on properties of the LIDAR system or the target.

3. The LIDAR system of claim 1, wherein the first and second peak detection each comprise at least one of thresholding and peak selection using one likelihood metric, thresholding and peak selection using separate likelihood metrics, thresholding and peak selection using a weighted likelihood metric, and frequency band filtering.

4. The LIDAR system of claim 1, wherein the processor is further to: determine one or more properties of the target based at least in part on the selected peak frequency.

5. The LIDAR system of claim 1, wherein the first peak detection selects a first intermediate peak from the first frequency band and the second peak detection selects a second intermediate peak from the second frequency band.

6. The LIDAR system of claim 5, wherein to select the peak frequency the processor is to: select one of the first or second intermediate peak as the peak frequency.

7. The LIDAR system of claim 6, wherein to select the peak frequency, the processor is to: perform a third peak detection on the first and second intermediate peaks.

8. The LIDAR system of claim 6, wherein to select one of the first or second intermediate peaks as the peak frequency, the processor is to: determine a priority order of the first and second frequency bands; andselect one of the first or second intermediate peaks as the peak frequency based on the priority order of the first and second frequency bands.

9. The LIDAR system of claim 6, wherein the processor is further to: determine whether a value of a likelihood metric associated with the first intermediate peak or second intermediate peak exceeds a threshold; andselect the peak frequency from the first intermediate peak and the second intermediate peak that has a highest value for the likelihood metric.

10. A method comprising: generating a frequency domain waveform based on a baseband electrical signal in a time domain, wherein the frequency domain waveform comprises a spectrum of frequencies;separating, by a processor of a LIDAR system, the spectrum of frequencies in the frequency domain waveform into a plurality of frequency bands comprising at least a first frequency band and a second frequency band;performing, by the processor, a first peak detection within the first frequency band of the spectrum of frequencies;performing, by the processor, a second peak detection within the second frequency band of the spectrum of frequencies, wherein the first peak detection and second peak detection comprise different peak detection techniques; andselecting, by the processor, a peak frequency from the spectrum of frequencies in the frequency domain waveform based at least in part on the first peak detection within the first frequency band and the second peak detection within the second frequency band.

11. The method of claim 10, further comprising: determining the first peak detection for the first frequency band and the second peak detection for the second frequency band based on properties of the LIDAR system or a target.

12. The method of claim 10, wherein the first and second peak detection each comprise at least one of thresholding and peak selection using one likelihood metric, thresholding and peak selection using separate likelihood metrics, thresholding and peak selection using a weighted likelihood metric, and frequency band filtering.

13. The method of claim 10, further comprising: determining one or more properties of a target based at least in part on the selected peak frequency.

14. The method of claim 10, wherein the first peak detection selects a first intermediate peak from the first frequency band and the second peak detection selects a second intermediate peak from the second frequency band.

15. The method of claim 14, wherein selecting the peak frequency comprises: selecting one of the first or second intermediate peak as the peak frequency.

16. The method of claim 15, wherein selecting the peak frequency comprises: performing a third peak detection on the first and second intermediate peaks.

17. The method of claim 15, wherein selecting one of the first or second intermediate peaks as the peak frequency comprises: determining a priority order of the first and second frequency bands; andselecting one of the first or second intermediate peaks as the peak frequency based on the priority order of the first and second frequency bands.

18. The method of claim 15, further comprising: determining whether a value of a likelihood metric associated with the first intermediate peak or second intermediate peak exceeds a threshold; andselecting the peak frequency from the first intermediate peak and the second intermediate peak that has a highest value for the likelihood metric.

19. A light detection and ranging (LIDAR) system, comprising: an optical scanner to transmit an optical beam towards, and receive a return signal from, a target;an optical processing system coupled to the optical scanner to generate a baseband electrical signal in a time domain based on the return signal, the baseband electrical signal comprising a plurality of frequencies corresponding to LIDAR target ranges; anda signal processing system coupled to the optical processing system, comprising: circuitry; anda memory operatively coupled to the circuitry, the memory to store instructions that, when executed by the circuitry, cause the LIDAR system to: determine one or more metrics for each of a plurality of frequencies based on the baseband electrical signal in the time domain;separate the plurality of frequencies into a plurality of frequency bands;perform a first peak detection within a first frequency band of the plurality of frequency bands;perform a second peak detection within a second frequency band of the plurality of frequency bands, wherein the first peak detection and second peak detection comprise different peak detection techniques; andselect a peak frequency from the plurality of frequencies based at least in part on the first peak detection within the first frequency band and the second peak detection within the second frequency band.

20. The LIDAR system of claim 19, wherein the first and second peak detection each comprise at least one of thresholding and peak selection using one likelihood metric, thresholding and peak selection using separate likelihood metrics, thresholding and peak selection using a weighted likelihood metric, and frequency band filtering.