WO2022206031A1

WO2022206031A1 - Method for determining noise level, lidar, and ranging method

Info

Publication number: WO2022206031A1
Application number: PCT/CN2021/138328
Authority: WO
Inventors: 章洪燕; 许帅骑; 杨晋; 向少卿
Original assignee: 上海禾赛科技有限公司
Priority date: 2021-03-31
Filing date: 2021-12-15
Publication date: 2022-10-06
Also published as: CN115144863A

Abstract

The present invention provides a method for determining a noise level, comprising: S101, obtaining an intensity information-time information curve of an optical signal; S102, clamping the intensity information-time information curve by means of an estimated noise threshold; and S103, determining a noise level by means of the clamped intensity information-time information curve. The present invention further provides a method for ranging by means of a lidar, and a lidar.

Description

Methods for determining noise levels, lidar, and ranging methods

technical field

The present disclosure relates to the technical field of photoelectric detection, and in particular, to a method for determining a noise level, a laser radar, and a method for ranging using the laser radar.

Background technique

Lidar is a radar system that emits a laser beam to detect the position, speed and other characteristics of the target. It is an advanced detection method that combines laser technology with photoelectric detection technology. Because of its high resolution, good concealment, strong anti-active interference ability, good low-altitude detection performance, small size and light weight, lidar is widely used in autonomous driving, transportation and communication, unmanned aerial vehicles, intelligent robots, resource exploration and other fields. The environment in which the lidar is located is usually full of various noises or background light, such as sunlight, the lights of other vehicles, the lights of buildings, and the laser light emitted by the lidar on other vehicles. In order to accurately calculate the distance of the target, the lidar needs to obtain the noise level in the surrounding environment as accurately as possible, and distinguish the echo generated by the laser beam emitted by itself from the surrounding noise. In the high-precision time-to-digital converter (TDC) measurement system in lidar, the detection signal and noise are superimposed on each other, and accurate noise information cannot be obtained on the basis of ensuring ranging accuracy and efficiency.

A single photon avalanche diode (SPAD) can be used as a detector in the lidar, and the SPAD can be triggered by a single photon to avalanche. In some applications, the output terminals of multiple SPADs are connected to the same TDC, as a macro pixel (pixel), the TDC provides the trigger moment and the number of SPADs that are simultaneously triggered in the macro pixel.

In order to obtain the current noise information, it is often necessary to reserve a period of time not to actively emit light, and use the SPAD(s) trigger information to measure the noise, which will waste a lot of measurement time. Other noise acquisition methods, one is to fix the noise information, that is, the detection of all cycles in several cycles or even a period of time adopts the same fixed noise threshold, so that real-time noise information cannot be obtained; Data other than the arrival time of the signal is used to calculate the noise, but in this way, the noise and the signal will affect each other, and it is difficult to obtain accurate noise information.

The contents in the Background section are merely technologies known to the disclosed person, and do not of course represent the prior art in the field.

SUMMARY OF THE INVENTION

In view of at least one defect of the prior art, the present invention provides a method for determining a noise level, including:

S101: Obtain an intensity information-time information curve of an optical signal;

S102: Clamp the intensity information-time information curve with an estimated noise threshold; and

S103: Determine the noise level using the clamped intensity information-time information curve.

According to one aspect of the present invention, the intensity information-time information curve is a photon number histogram.

According to an aspect of the present invention, the step S101 includes: scanning a field of view multiple times, and superimposing the intensity information versus time information curves of the multiple scans as the intensity information-time information curve.

According to an aspect of the present invention, the method further includes: acquiring the total intensity of the optical signal, calculating an estimated noise intensity using the total intensity, comparing the estimated noise intensity with a preset noise intensity, and obtaining a smaller value, Take the smaller value as the estimated noise threshold.

According to an aspect of the present invention, the step of calculating the estimated noise intensity includes: multiplying the total number of photons S by a preset ratio K to obtain the estimated noise photon number, as the estimated noise intensity, the preset ratio K is between 0-1.

According to an aspect of the present invention, the step S101 further includes: receiving an optical signal through a detection unit, where the detection unit includes a plurality of single-photon avalanche diodes, and the preset noise intensity is based on the single-photon avalanche diodes in the detection unit. One or more of the number, the dead time of the single photon avalanche diode is determined.

According to an aspect of the present invention, the step S103 includes: obtaining the total intensity of the clamped intensity information-time information curve, relative to the average value of the time span, as the noise level.

According to an aspect of the present invention, in step S102, for the intensity information corresponding to any time information, the smaller value of the intensity information and the estimated noise threshold is taken as the clamped intensity information.

The present invention also provides a method for ranging by means of laser radar, comprising:

S201: Calculate the noise level by the method described above;

S202: Perform noise filtering processing on the intensity information-time information curve based on the noise level; and

S203: Calculate the distance of the target object based on the intensity information-time information curve after noise filtering.

According to an aspect of the present invention, the step S202 includes: judging whether the peak value of the intensity information-time information curve is higher than the noise level; when the peak value is higher than the noise level, based on the noise level, Perform noise filtering processing on the intensity information-time information curve.

According to an aspect of the present invention, the step S203 includes: calculating the center of gravity of the intensity information relative to the time information according to the intensity information-time information curve after noise filtering, and the time information corresponding to the center of gravity is taken as the flight time.

According to an aspect of the present invention, the step S203 includes: calculating the leading edge time of the echo pulse according to the filtered intensity information-time information curve, and using the leading edge time as the flight time, wherein the leading edge time is the time information corresponding to the intensity information on the leading edge of the echo pulse equal to the preset threshold.

The present invention also provides a laser radar, comprising:

a light emission module, configured to emit a detection laser beam for detecting a target;

a light detection module configured to receive echoes of the detection laser beam reflected on the target and convert them into electrical signals; and

The control module is coupled to the light emitting unit and the light detection module, and is configured to execute the method as described above to calculate the distance of the target.

According to an aspect of the present invention, the light detection module includes a plurality of detection units, each detection unit includes a plurality of single-photon avalanche diodes to receive the echoes.

Description of drawings

The accompanying drawings constituting a part of the present disclosure are used to provide further understanding of the present disclosure, and the exemplary embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation of the present disclosure. In the attached image:

FIG. 1 shows a method for determining a noise level according to an embodiment of the present invention;

Figure 2 shows an intensity information-time information curve according to an embodiment of the present invention;

Figure 3 shows the intensity information-time information curve after clamping;

Figure 4 shows a schematic diagram of the number of echo photons obtained by multiple scans in one measurement;

5 shows a schematic diagram of a histogram obtained by accumulating the number of echo photons of multiple scans;

Figure 6 shows a detection unit according to an embodiment of the present invention;

Fig. 7 shows the curve waveform obtained after the intensity information-time information noise filtering processing;

FIG. 8 shows a method for ranging by a lidar according to a preferred embodiment of the present invention;

Figure 9 shows a data storage method according to the prior art;

10 and 11 show a specific schematic diagram of a storage mode according to a preferred embodiment of the present invention;

Figure 12 shows a block diagram of a lidar according to one embodiment of the present invention.

Detailed ways

In the following, only certain exemplary embodiments are briefly described. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. Accordingly, the drawings and description are to be regarded as illustrative in nature and not restrictive.

In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "top", "bottom", "front", " Or The positional relationship is based on the orientation or positional relationship shown in the accompanying drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, Therefore, it should not be construed as a limitation of the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes, and should not be understood as indicating or implying relative importance or implying the number of indicated technical features. Thus, features defined as "first", "second" may expressly or implicitly include one or more of said features. In the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

In the description of the present invention, it should be noted that, unless otherwise expressly specified and limited, the terms "installation", "connection" and "connection" should be understood in a broad sense, for example, it may be a fixed connection or a detachable connection Connection, or integral connection: it can be a mechanical connection, an electrical connection or can communicate with each other; it can be directly connected or indirectly connected through an intermediate medium, it can be the internal communication of two elements or the interaction of two elements relation. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

In the present invention, unless otherwise expressly specified and limited, a first feature "on" or "under" a second feature may include direct contact between the first and second features, or may include the first and second features Not directly but through additional features between them. Also, the first feature being "above", "over" and "above" the second feature includes that the first feature is directly above and diagonally above the second feature, or simply means that the first feature is level higher than the second feature. The first feature "below", "below" and "beneath" the second feature includes the first feature being directly above and obliquely above the second feature, or simply means that the first feature has a lower level than the second feature.

The following disclosure provides many different embodiments or examples for implementing different structures of the present invention. In order to simplify the disclosure of the present invention, the components and arrangements of specific examples are described below. Of course, they are only examples and are not intended to limit the invention. Furthermore, the present disclosure may repeat reference numerals and/or reference letters in different instances for the purpose of simplicity and clarity and not in itself indicative of a relationship between the various embodiments and/or arrangements discussed. In addition, the present disclosure provides examples of various specific processes and materials, but one of ordinary skill in the art will recognize the application of other processes and/or the use of other materials.

The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, but not to limit the present invention.

first

FIG. 1 shows a method 100 for determining a noise level according to an embodiment of the present invention, which can be used to determine a noise level or threshold in a lidar, which will be described in detail below with reference to FIG. 1 .

In step S101 : obtaining the intensity information-time information curve of the optical signal.

FIG. 2 shows an intensity information-time information curve according to an embodiment of the present invention, wherein the abscissa is the time information, and the ordinate is the parameter value representing the optical signal intensity information. It is easy to understand that the abscissa in FIG. 2 is not an absolute time coordinate, and the abscissa is the time difference between the time when the laser radar transmits the light pulse and the time when the light signal is received. The intensity information-time information curve shown in Fig. 2 can be the intensity information-time information curve obtained in one scan of the lidar for a certain field of view, or it can be obtained in multiple scans of a certain field of view Curves formed by superimposing the intensity information of the optical signal with respect to the time information are all within the protection scope of the present invention. According to a preferred embodiment of the present invention, the ordinate in FIG. 2 can use the number of photons to represent the intensity of the optical signal.

In step S102: the intensity information-time information curve is clamped by using an estimated noise threshold.

Estimated noise thresholds are shown in FIG. 2 (shown by the line of “Noise Thresholds” in FIG. 2 ). For the intensity information corresponding to any time information, the smaller value of the intensity information and the estimated noise threshold is taken as the clamped intensity information, and a clamped intensity information-time information curve is obtained. In this embodiment, the clamping refers to comparing the intensity information corresponding to any time scale with the estimated noise threshold, and taking the smaller value as the intensity information after clamping. That is, for the part of the intensity information-time information curve where the intensity information is higher than the estimated noise threshold, take the estimated noise threshold; for the intensity information-time information curve, the intensity information is lower than the estimated noise For the part of the threshold, take the actual intensity information. In the total time span, the intensity information corresponding to each time scale is compared in turn to obtain a clamped intensity information-time information curve. For the intensity information-time information curve shown in FIG. 2 , the estimated noise threshold set in FIG. 2 is used for clamping, and the clamped intensity information-time information curve shown in FIG. 3 is obtained.

In step S103: the noise level is determined using the clamped intensity information-time information curve.

On the basis of the clamped intensity information-time information curve shown in FIG. 3 , preferably an average value of the clamped intensity information relative to the time information can be obtained as the noise level. For example, the area of the area covered by the clamped intensity information-time information curve can be obtained, and then divided by the total time span, that is, the time span of the clamped intensity information-time information curve on the horizontal axis, and the clamped intensity information-time information curve can be obtained. The intensity information relative to the time span is averaged as the noise level.

Preferably, the clamped intensity information shown in FIG. 3 is characterized by the number of photons, and the total number of photons in the total time span is calculated and divided by the total time span to obtain the noise level characterized by the number of photons.

Through the above method, the noise value can be calculated directly by using the intensity information-time information curve obtained by lidar detection, and it is not necessary to reserve a special detection time for noise measurement; and after obtaining an intensity information-time information curve, it can be The noise level corresponding to the detection data can be obtained by calculating the detection data, that is, the real-time noise can be acquired while measuring the distance, and the accuracy of the noise level can be improved. Taking the noise level calculated in real time as the corresponding ranging noise threshold can improve the ranging accuracy.

In the detection process of lidar, take the detector composed of single-photon avalanche diode SPAD(s) as an example, because SPAD can be triggered by a single photon to avalanche effect, so it is easily affected by environmental noise; The photon detection efficiency (PDE) of the commonly used detection light band of lidar is low, and the signal intensity obtained by a single detection is very weak. It is in Geiger mode within the time window, that is, the state where the avalanche effect can be triggered by photons. This time window is called the "detection time window". Only a few triggers occur within the time window, and it is impossible to distinguish whether it is triggered by the echo signal reflected by the target or Triggered by ambient noise. According to a preferred embodiment of the present invention, in order to improve the distance measurement performance of the lidar and reduce the influence of noise, as shown in FIG. 4 , the same field of view range can be repeatedly scanned multiple times. For each scan, the light source at the transmitting end emits a light pulse for detection, and the controller of the lidar records the emission time t1 when the light pulse is emitted. The light pulse encounters an external obstacle, is reflected by the obstacle, and returns to the The lidar is received by the photodetector at the receiving end. When the photodetector is an array of SPAD(s), ambient light may also cause the SPAD to be triggered avalanche. Once the SPAD receives the photon, an avalanche electrical signal is generated, which is transmitted to the time-to-digital converter TDC. The TDC outputs the time signal of the SPAD trigger and the signal of the number of SPADs that are triggered at the same time, and the subsequent memory stores the SPAD trigger time minus the launch time t1. The timestamp timestamp (that is, the time information represented by the horizontal axis in Figures 2 and 3) and the trigger number of the timestamp (hereinafter referred to as cnt) signal.

The trigger number cnt obtained by each measurement is stored in the corresponding memory location according to the timestamp timestamp. When a new trigger number cnt arrives at the corresponding position of a timestamp timestamp, the original stored value is compared with the new trigger number cnt. Accumulate and then update to this position. After multiple measurements are superimposed, a histogram is saved in the memory. As shown in Figure 5, the histogram reflects the sum of the number of triggers cnt corresponding to different timestamps on the time axis. In this way, the histogram is used to calculate The time of flight corresponding to the echo pulse is obtained by operations such as the center of gravity, and then the ranging result is obtained.

Therefore, according to the above-mentioned embodiment, in one time-of-flight measurement of the lidar for a field of view, the lidar actually performs multiple scans (multiple transmit-receive cycles), and the number of scans can range from dozens to several times. Hundreds of times, scan a field of view (or approximately one target point) within a period of time, and superimpose the curve of the intensity information received by the detector relative to the time information in the multiple scans as the intensity information - Time information curve. For example, as shown in Figure 4, in the 1st, 2nd, ..., nth scans, only a very limited number of echoes or photons are received in each scan, but after the detection results of n scans are superimposed, Obtain the photon number histogram of one time-of-flight measurement in this field of view, as shown in Figure 5, where the scale of the abscissa is time information, and the scale width on the time axis is usually equal to the resolution of the time-to-digital converter in the lidar, That is, the resolution of the detection time of the lidar. The total number of photons S can be obtained by accumulating the number of photons corresponding to each scale on the photon number histogram, which is used to represent the total intensity information.

In the context of the present invention, a distinction is made between "measurement" and "scanning". Specifically, a "measurement" corresponds to the time-of-flight measurement of a certain field of view within a detection period of the lidar (that is, within the period of generating a frame of point cloud) to generate a point cloud in a frame. Or multiple (one or more columns or one block) "points", a complete frame of point cloud is obtained after the measurement of all fields of view is completed; and "scanning" refers to the laser in a detection channel during a measurement process The process of completing a transmission and the detector completing the corresponding reception. A "measurement" may include a single "scan" or multiple "scans" of the same target point, eg hundreds of times.

According to an embodiment of the present invention, the method 100 further includes: acquiring the total intensity of the optical signal, calculating an estimated noise intensity (for example, represented by the number of photons) according to the total intensity, and setting parameters according to the receiving end of the lidar. Set the preset noise level.

The total intensity of the optical signal can be characterized by the total number of photons S received in the total time span.

The total time span corresponds to the total time during which the intensity information is recorded on the curve shown in FIG. 2 , that is, the difference between the maximum value and the minimum value of the time information. The time range corresponding to the maximum detection distance and the minimum detection distance of the lidar can be used as the total time span. Or alternatively, as shown in FIG. 2 , only the intensity information corresponding to part of the time information is recorded. As a specific embodiment, the time period during which the reflected echo of the target can be expected to be obtained is used as the total time span of the intensity information-time information curve. The total intensity of the optical signal can be characterized by the area covered by the intensity information-time information curve in FIG. 2 . According to a preferred embodiment of the present invention, the total intensity can be characterized by the total number of photons S received in the total time span.

To calculate the estimated noise intensity, the total number of photons S in the total time span can be multiplied by a preset ratio K to obtain the estimated noise photon number, which is used as the estimated noise intensity, and the preset ratio K is between 0- between 1. The preset ratio K∈(0,1), the empirical value of the ratio of the noise number to the total photon number can be obtained as K through simulation or actual measurement.

In a specific embodiment of the present invention, the receiving end parameters of the lidar include the total number of detection units (pixels), the number of SPADs in each detection unit, and the dead time of the SPAD, according to which the SPAD (unit time) is calculated. ) of the average maximum noise trigger photon number S _max , which is taken as the preset noise intensity. Regarding how to calculate S _max , specific examples will be given below.

Further, the estimated noise intensity calculated according to the intensity information-time information curve is compared with the preset noise intensity S _max , a smaller value is obtained, and the smaller value is used as the estimated noise threshold.

According to a preferred embodiment of the present invention, in step S102, for the intensity information corresponding to any time information, the smaller value of the intensity information and the estimated noise threshold is taken as the clamped intensity information. The step S101 further includes: receiving the echo through a detection unit, where the detection unit includes a plurality of single-photon avalanche diodes. The preset noise intensity is determined based on one or more of the number of single-photon avalanche diodes in each detection unit and the dead time of single-photon avalanche diodes. FIG. 6 shows an embodiment of the detection unit. As shown in the figure, the detection module includes a plurality of detection units, and the detection units P1, P2 and P3 are shown in the figure, and each detection unit includes nine single-photon avalanche diodes respectively. , the nine single-photon avalanche diodes are all connected to a time-to-digital converter TDC, so that the TDC can obtain the time signal of the SPAD triggered in the detection unit and the signal of the number of SPADs triggered at the same time, and store them in the memory.

For the detection unit shown in Figure 6, a detection unit can trigger up to 9 times within the SPAD dead time. The dead time of the SPAD in the detection unit shown in Figure 6 is set to be 20 ns, and the same SPAD can be triggered at most once within 20 ns. Assuming that one measurement includes N scans, one detection unit can be triggered at most within the time of one measurement. Triggered (9*N*detection time/20ns) times, S _max =9*N/20, that is, the maximum number of noise trigger photons per unit time caused by noise in the total flight time, as the preset noise intensity.

In the above embodiment, the noise level can be calculated directly by using the intensity information-time information curve (such as a histogram) obtained by detection, and it is not necessary to reserve a detection time for noise measurement; and real-time noise can be obtained while ranging, which can be used as Ranging noise threshold. In the data processing process of lidar, the noise threshold can be used to judge whether the echo is valid, that is, the echo pulse signal reflected from the target object or the noise signal. If the measured echo strength is lower than the noise threshold, it is judged as noise and filtered. If the set noise threshold is higher than the actual noise level, the actual detected echo will be misjudged as noise, resulting in data loss; if the set noise threshold is lower than the actual noise level, some noise signals will be judged as targets Objects reflect echoes, creating noise in the point cloud. In addition, the environmental noise around the lidar is also constantly changing, and the inability to obtain the noise in real time will also cause lost points or noise. The above embodiments of the present invention can acquire the noise level in real time, improve the signal-to-noise ratio of the laser radar, and thus improve the ranging accuracy.

The present invention also provides a method 200 for ranging by using a laser radar, including:

S201: Calculate the noise level through the method 100 as described above.

S202: Perform noise filtering processing on the intensity information-time information curve based on the noise level.

As a preferred embodiment of the present invention, a peak value is obtained on the intensity information-time information curve, that is, the maximum value of the intensity information on the total flight time span, and it is determined whether the peak value is higher than the noise level, and if it is higher than the noise level level, the intensity information-time information curve is subjected to noise filtering processing.

The noise filtering process, for example, compares the intensity information corresponding to each time scale with the noise level, retains the intensity information data higher than the noise level, and subtracts the noise level from the intensity information data to obtain the intensity corresponding to the actual detection signal. Information-time information curve.

According to a preferred embodiment of the present invention, the step S203 includes: according to the time information corresponding to the center of gravity of the intensity information-time information curve after noise filtering processing on the time axis, as the flight time, calculating the time of the target object. distance.

As shown in Figure 7, the curve waveform obtained after the noise filtering process is shown. On this basis, the center of gravity position of the curve waveform is obtained, and the flight time Tof1 corresponding to the center of gravity position is taken as the time of flight in this measurement process. Time of flight, which is used to calculate the distance to the target and generate a data point in the point cloud.

On the other hand, from the intensity information-time information curve after noise filtering, the maximum value of the intensity information is obtained, and as the echo intensity, the detection light intensity emitted by the lidar is obtained, and the target object can be calculated based on the echo intensity and the detection light intensity. reflectivity.

Fig. 8 shows a method for ranging by a lidar according to a preferred embodiment of the present invention, wherein the detection methods shown in Figs. 4-7 are combined. This is described in detail below with reference to FIG. 8 .

In the embodiment of FIG. 8 , in one measurement of the lidar, the lidar performs multiple scans, taking 500 times as an example.

In step S301, the i-th scan is performed, and i is initially 1. The laser of the lidar emits the detected light pulse, and the controller of the lidar records the emission time t1 of the emitted light pulse. The light pulse encounters an external obstacle, is reflected by the obstacle and returns to the lidar, and is detected by the photoelectric detection of the receiving end. received by the device. When the photodetector is an array of SPAD(s), ambient light may also cause the SPAD to be triggered avalanche. Once the SPAD receives the photon, it generates an avalanche electrical signal, which is transmitted to the time-to-digital converter TDC. The TDC outputs the time signal of the SPAD trigger and the signal of the number of SPADs that are triggered at the same time, and stores the SPAD trigger time minus the time stamp of the launch time t1. timestamp (that is, time information) and the trigger number of the timestamp (hereinafter referred to as cnt).

In step S302, the detection results of the i-th scan and the detection results of the previous i-1 times are accumulated. When a new trigger number cnt arrives at the position corresponding to a certain timestamp timestamp, the original stored value and the new trigger number cnt are accumulated and then updated to the position. At the same time, all cnt values are accumulated (time stamps are not distinguished) to obtain the total number of SPAD triggers, which are used to obtain the total intensity of the optical signal for one measurement.

In step S303, it is determined whether i has reached a preset number of scans N, for example, 500. If it has not been reached yet, accumulate i, and go back to step S301 to continue scanning detection. If the preset number of scans N has been reached, proceed to step S304.

In step S304, a histogram is generated. Since N scans have been completed, the number of triggers cnt obtained in each scan is stored in the corresponding memory location according to the time stamp timestamp. After multiple measurements are superimposed, a histogram is stored in the memory, as shown in Figure 5. The histogram reflects the sum of the number of triggers cnt corresponding to different timestamps on the timeline. The data in memory can be read out as a histogram. At the same time, all the cnt values are accumulated to obtain the total intensity of the optical signal represented by the number of photons in one measurement.

In step S305, the data of the histogram is filtered. After that, steps S306 and S307 are performed respectively.

In step S306, the noise level is obtained, for example, the noise threshold around the lidar is obtained according to the method 100 described above.

In step S307, the peak value of the signal is found according to the filtered histogram, that is, the point with the largest ordinate in the histogram is found.

In step S308, according to the noise level obtained in step S306, it is determined whether the signal peak value found in step S307 is valid, that is, whether the signal peak value is greater than the noise level. When the signal peak value is greater than the noise level, the signal peak value is valid, and then step 309 is performed. When the signal peak value is lower than the noise level, the signal peak value is invalid and not processed.

In step S309, ranging information is obtained.

As an embodiment of the present invention, the center of gravity of the intensity information in the total time span is calculated, and the time information corresponding to the center of gravity is used as the flight time to calculate the distance of the target object.

As another embodiment of the present invention, the distance to the target object is calculated using the echo pulse leading edge time. Specifically, the intensity information of the front edge of the echo pulse is compared with a preset threshold, and the time information corresponding to the intensity information whose intensity is equal to the preset threshold is used as the flight time.

As an embodiment, the above-mentioned preset threshold is a noise threshold.

As an embodiment, the above-mentioned preset threshold is an average value of the signal peak value and the noise threshold value.

In step 309, the above-mentioned methods of S202-S203 may be used to perform noise filtering processing on the intensity information-time information curve based on the noise level, and then based on the intensity information-time information curve after the noise filtering processing, Calculate the distance and/or reflectivity of the target.

In the above-mentioned embodiment of the present invention, during a measurement process of the lidar, the detection results of N scans are superimposed and stored. After all scans are completed, the data in the memory can be read out as a histogram, and the histogram data is filtered. , look for signal peaks. Using the noise calculation method of the present invention, the real-time noise value of the histogram can be obtained as the current noise threshold. When it is judged that the peak value of the histogram signal is greater than the noise threshold, the peak value is an effective peak value, and then the echo arrival time is calculated. Therefore, the noise level and ranging information of the current measurement can be obtained almost simultaneously after one measurement is completed. Compared with the solution of setting a uniform noise threshold for the lidar, the solution of dynamically calculating the noise level of the embodiment of the present invention is more accurate.

The noise level is calculated by using the scheme of the embodiment of the present invention, and the distance and/or reflectivity of the target object are calculated after the intensity information is filtered out of the noise. The filtered intensity information can reflect a more real echo pulse signal. Conducive to improving the measurement accuracy.

the second aspect

The second aspect of the present invention relates to a method for storing detection data of a lidar, which can be combined with the method 100 for determining a noise level applied to the first aspect of the present invention.

In some lidar time-to-digital converters, each time scale of its time resolution needs to have a corresponding storage location, and all the triggered SPAD number information cnt obtained by multiple measurements are stored in the same time as the time. In the corresponding storage location, the time resolution of the time-to-digital converter TDC can reach the order of picoseconds ps, so a register with a very large space is required. The specific explanation is as follows.

SPAD can be triggered by a single photon to avalanche effect, so it is easily affected by ambient light noise; on the other hand, SPAD has low photon detection efficiency (PDE) for the commonly used detection light band of lidar, and the signal obtained by a single detection The intensity is very weak, as shown in Figure 4. In a detection scan, only a few triggers may occur within the detection time window (two triggers in Figure 4), and it is impossible to distinguish whether it is an echo signal or ambient light noise. In order to improve the range-finding performance of the lidar and reduce the influence of noise, as shown in Figure 4, the lidar can perform multiple repeated measurements during one detection of the same field of view (one measurement is called a sweep). , the number of repetitions can reach 400-500 times, or more or less), the results of multiple measurements or scans are accumulated to obtain a histogram, and the distance is measured from this, and then a point cloud on the lidar is obtained. point.

For one scan, the controller of the lidar gates some (a row or column or any shape of interest) macro pixels by supplying high voltage to the SPAD, and then sends a synchronization signal to notify the laser at the transmitter that it can emit light, and the laser at the transmitter At time t _a (a represents the a-th scan), a light pulse for detection is sent out. The light pulse encounters an external obstacle, is reflected by the obstacle and returns to the lidar, and can be received by the photodetector at the receiving end. . When the photodetector is a SPAD(s) array, once the SPAD receives the photons, an avalanche electrical signal is generated, which is transmitted to the time-to-digital converter, and the time-to-digital converter outputs the time signal t _1a triggered by the SPAD and the SPAD triggered at the same time. The number signal cnt _1a (here 1a represents the first trigger of the a-th scan), the timestamp _1a (hereinafter referred to as tp _1a ) of t _1a -t _a is calculated by the subtraction procedure, and tp _1a and the time The number of triggers for stamping the cnt _1a signal is transmitted and stored in memory. A macro pixel includes multiple SPADs, and SPADs can be detected again after the dead time, so in one scan, SPAD triggering may occur at another time, and the memory stores tp _2a and cnt _2a of this trigger (2a is Represents the 2nd trigger of the a-th probe). Multiple triggers in one scan need to be stored as time information.

In the next scan b, the controller of the lidar sends a signal again according to the preset program to control the transmitting end to send out the detection light pulse at time t _b . Once the SPAD receives the photon, the avalanche electrical signal is transmitted to the time-to-digital converter TDC, and the time-to-digital converter TDC outputs the time signal t _1b triggered by the SPAD and the signal cnt _1b of the number of SPADs triggered at the same time (the b-th detection 1 trigger), the subsequent memory stores the timestamp timestamp _1b (hereinafter referred to as tp _1b ) of the SPAD trigger time t _1b -t _b and the trigger number cnt _1b signal of the timestamp. A macro pixel includes multiple SPADs, and the SPADs can be detected again after the dead time, so in one scan, the SPAD may be triggered again at another time, and the memory stores tp _2b and cnt _2b of this trigger.

In hundreds of measurements, the trigger number cnt obtained by each measurement is stored in the corresponding memory location according to the timestamp. When a new trigger number cnt arrives at the corresponding location of the same timestamp, the original stored value is compared with The new trigger number cnt is accumulated and then updated to this position. After multiple scans and stacking, a histogram is saved in the memory. As shown in Figure 5, the histogram reflects the sum of the trigger numbers cnt corresponding to different timestamps on the time axis. In this way, the histogram is used to calculate the center of gravity or the frontier time and other operations to obtain the time information corresponding to the echo, which is used as the flight time for distance calculation to generate a point on the point cloud.

A data storage method is shown in Figure 9. The abscissa is time t, the scale interval of the abscissa is the resolution of TDC, and each time scale corresponds to a storage location R (register). For example, in a certain detection scan a, a SPAD trigger occurs at time scale 0. According to the transmission time and the trigger time of TDC transmission, the timestamp tp ₁ (trigger time - the current transmission time) and the trigger number information cnt _1a are obtained by calculation, and cnt _1a is stored. At the corresponding storage location R1 at time tp ₁ ; if a SPAD trigger occurs at time scale 4, obtain time information tp ₅ and cnt _5a , and store cn _5a in the storage location R5 corresponding to tp ₅ . In another detection scan _b , SPAD triggering also occurs at time scale 4, and time information tp ₅ and _cnt _5b are obtained, and _cnt _5b also corresponds to the storage location R5. The added value is updated to R5. (combined with Fig. 9, a represents the a-th detection, b is the b-th detection, and the numbers represent the corresponding time scale and the corresponding storage location; the storage location R corresponds to the time scale one-to-one, the memory only stores the trigger number cnt, and the data processing When the circuit reads the data, the time corresponding to the trigger number cnt can be known according to the storage location).

Referring to Figure 9, it can be seen that a histogram is obtained by accumulating the data of many detection scans (400-500 times), and the detection results of hundreds of scans are superimposed into a histogram to obtain a point in the point cloud. The storage location corresponding to the time scale stores the accumulated sum of all the trigger quantities cnt triggered at this moment. Although SPAD triggering does not occur at every time scale in a scan, as shown in Figure 9, a histogram data is superimposed by many detection results, and each time scale may be triggered at a certain scan A SPAD trigger occurs during the process, so that the memory receives the corresponding data. Therefore, for a TDC, each time scale needs to have a corresponding storage location, and all the trigger numbers cnt obtained from multiple measurements are stored in the storage location corresponding to the time. The time interval of tp, that is, the resolution of TDC The rate is on the order of ps, requiring very large registers.

With such a storage and ranging method, since the accuracy of the timestamp is in the order of ps, when a long tof detection is required, it requires a great deal of memory and storage space to store a complete histogram. In particular, in order to improve the distance measuring ability, it is necessary to increase the measurement duration and the number of repeated measurements, and the requirements for storage space are also increasing.

The inventor of the present application has conceived that it is not necessary to set a corresponding storage location for each time scale of the TDC time resolution. The weight of the time information to store the intensity information. The invention adopts the data storage method of weighted accumulation, compresses the original signal under the condition of keeping the ranging precision, and greatly reduces the storage space required for storing the histogram. Specifically, the weighted accumulation data storage method can reduce the total storage space to 1/10 of the original range.

Specifically, the time precision of the stored intensity information in the present invention is the first time precision, and the first time precision may be n times the time resolution of the time-to-digital converter TDC. The intensity information refers to the optical signal intensity information corresponding to the time information. For different photodetectors, different parameters can be used to represent the optical signal intensity: for example, the detector is a SPAD array, and the SPAD that is triggered at the same time corresponding to the time information can be used. The quantity is used as the intensity information; if the detector is SiPM, the output level/current intensity corresponding to the time information can be used to represent the optical signal intensity information.

A detailed description will be given below with reference to the accompanying drawings.

First, the detection data of the radar includes time information and intensity information corresponding to the time information.

Taking the detection unit shown in FIG. 6 as an example, the time information is the time when one or more single-photon avalanche diode SPADs in the detection unit (P1, P2, P3...) are triggered, and the intensity information is the triggering time. The number of single-photon avalanche diodes SPADs that are triggered by time, that is, the intensity of the optical signal is characterized by the number of single-photon avalanche diodes SPADs that are triggered. According to a preferred embodiment of the present invention, the time information is the time stamp timestamp of the triggering of the single-photon avalanche diode SPAD, that is, the time difference t between the time t a emitted from the laser and the time t _1a _when the single-photon avalanche diode SPAD is triggered _1a -t _a .

In the embodiment of FIG. 6 , the single-photon avalanche diode SPAD is used as an example for illustration. Those skilled in the art can easily understand that the present invention is not limited to this, and other types of photodetectors can also be used, including but not limited to avalanche photodiode APD. , silicon photomultiplier tube SiPM, etc.

The data storage method of the present invention is specifically as follows: with a first time precision, according to the weight of the time information, the intensity information is stored; the first time precision is the time interval between any two adjacent first time scales , and is n times the time resolution of the radar detection data, where n>1; the weight is associated with the time information and the time interval of at least one first time scale.

Figures 10 and 11 show a specific schematic diagram of a storage method according to a preferred embodiment of the present invention, and the implementation of the data storage method of the present invention will be described in detail below with reference to Figures 10 and 11 .

In Figure 10, the abscissa is the flight time, and the interval of the time scale of the abscissa is, for example, the time resolution of the lidar, such as the time resolution of the time-to-digital converter TDC, which can reach the order of picoseconds ps. As shown in Figure 10, the first time scale is set based on the time resolution of the lidar. As shown in A and A+1 in Figure 10, the interval between two adjacent first time scales spans 16 interval of the time resolution of the lidar. When a photon is detected at time x (eg, one or more single-photon avalanche diodes SPAD in a detection unit shown in FIG. 6 are triggered), the detected intensity value is stored according to the weight of said time x. Time x represents the time resolution of radar detection data where the time interval between the time and the first time scale A adjacent to the left of the time is x times.

Those skilled in the art can easily understand that since the time resolution of lidar is small and the interval of the first time scale is large, the time scale corresponding to the time resolution of lidar can also be called "fine scale". Also known as "coarse scale".

As shown in FIG. 10 , the weight of the time x includes a first weight and a second weight, the first weight is associated with the time interval between the time x and one of the adjacent first time scales, and the second weight Associated with the time interval between said instant x and another adjacent first time scale. After the first weight and the second weight are determined, the intensity information is stored according to the first weight and the second weight, respectively, with the first time precision.

According to a preferred embodiment of the present invention, the first weight is associated with the time interval between the moment x and the first time scale A adjacent to the left thereof, and the first weight is, for example, (16-x), The second weight is associated with the time interval between the moment x and the first time scale A+1 adjacent to the right side thereof, and the second weight is, for example, x. Therefore, the moment x is represented by its weight at two adjacent coarse scales (A, A+1), where the weight of x at coarse scale A is (16-x), and at coarse scale A+ The weight of 1 is x (x represents the distance from A at this moment), which is equivalent to the fine scale of x at this moment. In other words, by using x as a weight, the data at the fine scale is stored at the addresses corresponding to two adjacent thick scales to represent the value of the scale x, rather than storing the scale x itself. This process is represented by the equation as follows:

A*(16-x)+(A+1)*x=A*16+x

In the formula, the left side of the equal sign is the sum of the weights applied according to the coarse scale storage, the start value and the end value of the coarse scale, and the right side of the equal sign is the specific value of the timestamp. It can be seen that the storage method of coarse scale + weight can accurately represent the specific value of the timestamp.

Similarly, when the signal obtained by triggering includes the number of triggers cnt indicating the number of triggers or information such as intensity in addition to the timestamp, the newly added intensity information on the coarse scale A is cnt*(16-x), and the coarse scale A+1 The newly added intensity information is cnt*x, which can be accumulated separately in multiple scans. A detailed description will be made with reference to FIG. 11 . The fine scale represents the time-to-digital converter TDC time resolution. For a timestamp, the starting value of its coarse scale is A, and its fine scale is at the corresponding 0-15 fine scale x scale in its coarse scale.

Referring to FIG. 11 , a register is allocated for each coarse scale, the coarse scale interval of the abscissa is 16 times the TDC resolution, and each coarse scale corresponds to a register. During a certain scan a, a SPAD trigger occurs at time scale 0, time information tp ₁ (corresponding to x _1a =0) and trigger quantity information cnt _1a are obtained, and cnt _1a is stored in register A corresponding to coarse scale A, respectively *(16-x _1a ), store cnt _1a *x _1a in register A+1 corresponding to coarse scale A+1; at another time scale 5, obtain time information tp ₆ (corresponding to x6 _a = 5) and the number of triggers For the information cnt _6a , read out the data stored in the register A corresponding to the coarse scale A, add cnt _6a *(16-x _6a ), and then store it in the register A; read the register A+1 corresponding to the coarse scale A+1 The data read out, add cnt _6a *x _6a and re-store it in register A+1. In a coarse scale time (fine scale 0 to 15), all trigger quantity information cnt is weighted, and stored in the registers corresponding to storage positions A and A+1 after summing with the original data. The trigger quantity information cnt in the next coarse scale time is weighted and stored in the registers corresponding to the coarse scales A+1 and A+2. For example, when a SPAD trigger occurs at time 2', the time information tp ₃ ' and cnt _3a ' are obtained. , then add cnt _3a '*(16-x _3a ') to the data stored in register A+1 corresponding to coarse scale A+1, and store cnt _3a '* in register A+2 corresponding to coarse scale A+2 x _3a '.

In the process of the next scan b, the received signals tp ₂ and cnt _2b are assigned weights cnt _2b *(16-x _2b ) and cnt _2b *x _2b on the coarse scale A and A+1 respectively, which are respectively the same as the original stored data. After the summation is stored in the registers corresponding to the coarse scale A and A+1. A histogram is obtained by accumulating the data of many scans. In several scans, all the trigger numbers cnt corresponding to the triggers at times 0 to 15 are stored in the registers corresponding to the coarse scales A and A+1.

Compared with the scheme that requires a register for data storage at each fine scale, the present invention adopts the weighted accumulation storage method, only needs to set registers corresponding to the coarse scales from 0 to n+1, and the required number of registers is reduced to the original one. 1/16, although the storage bit width of each register increases, the space occupied becomes larger, but because the storage location to be allocated is greatly reduced, the weighted accumulation data storage method can reduce the total storage space to the original 1/ 10 ranges.

In the embodiments of FIGS. 10-11 , the time interval between adjacent first time scales (coarse scales) is 16 times the time resolution (fine scale) of radar detection data, that is, data compression is performed with a weight of 16. Those skilled in the art can easily understand that the present invention is not limited thereto, and the weight here can be any large positive integer. Preferably, the time interval of the coarse scale is 2 ^m times that of the fine scale, where m is a positive integer, so as to facilitate implementation in an FPGA or an ASIC.

In the above embodiment, the first weight is (16-x), the second weight is x, the present invention is not limited to this, the first weight may be x, and the second weight is (16-x ), or the first weight may be 1-(x/n), and the second preset weight may be x/n, as long as the first weight is equal to the time x and one of the adjacent first time scales The second weight may be associated with the time interval between the moment x and another adjacent first time scale.

The storage method of the second aspect of the present invention can be applied to the method 100 of determining the noise level of the first aspect of the present invention. For example, in step S101 of the method 100, an intensity information-time information curve of an optical signal is obtained, and the intensity information-time information curve is, for example, a photon number histogram. The curves of intensity information relative to time information of multiple scans are superimposed to obtain, and the data obtained from each scan can be weighted and stored according to the storage method of the second aspect of the present invention, and finally a photon number histogram is obtained.

In addition, in the method 100, when calculating the total intensity S of the optical signal, it can be performed according to different ways. According to an embodiment of the present invention, the total intensity S of the optical signal is the accumulated sum of the original trigger number cnt data before weighting, and the histogram is two parallel steps. Or alternatively, the total intensity S of the optical signal can be calculated from the histogram, all of which are within the scope of the present invention.

According to the photon number histogram obtained by the storage method of the second aspect of the present invention, when the noise level is calculated, since the time precision of the histogram data storage is n times the time resolution of the radar detection data, that is, the number of scales on the abscissa of the histogram becomes The intensity information corresponding to each time scale is processed in steps S102, S103, and the required calculation amount is correspondingly greatly reduced, which improves the data processing and calculation efficiency while maintaining the calculation accuracy unchanged.

the third aspect

The present invention also relates to a laser radar 300, as shown in FIG. 12, including a light emission module 301, a light detection module 302 and a control module 303, wherein the light emission module 301 refers to the part of the laser radar for laser emission, which can be Contains circuits, devices, structures, etc., configured to emit a detection laser beam for detection of a target. The light detection module 302 refers to the part of the laser radar used to detect the echo signal of the laser, which may include circuits, devices, structures, etc., and is configured to receive the echo reflected by the detection laser beam on the target and convert it into electric signal. The control module 303 is coupled to the light emission module 301 and the light detection module 302, and is configured to execute the

method

100, 200 or 300 as described above to calculate the distance of the target object. According to a preferred embodiment of the present invention, the light detection module 302 includes a plurality of detection units as shown in FIG. 6 , and each detection unit includes a plurality of single-photon avalanche diodes for receiving echoes.

The light emitting module 301 includes a light emitting array 3011, which is implemented by, for example, a Vertical-Cavity Surface-Emitting Laser (VCSEL) array. The light emitting array 3011 includes multiple rows and multiple columns. Wherein, each row and each column are respectively provided with a plurality of light emitting units 30111, and each light emitting unit 30111 includes at least one light emitter; the light emitting array 3011 is also correspondingly configured with an emitting array driving circuit, which is coupled to each light emitting unit 30111. The transmitter is used to drive the light transmitter to work.

The photodetection module 302 includes a photodetection array 3021, and the photodetection array 3021 may use, for example, a detection unit as shown in FIG. 6 to receive the optical echo signal after the detection beam reaches the target OB. In addition, the photodetection module 302 may further include a signal readout circuit 3022 for reading out and transmitting the signals generated by the photodetection array 3021 to the control module 303 .

In the lidar 300, a transmitting lens group 304 may be further provided, which is located on the outgoing optical path of the light transmitting array 3011; May be located on the focal plane of the receiving lens group 305 .

The control module 303 is coupled to the light emission array 3011 and the photoelectric detection array 3021; respectively controls the light emission arrays to emit detection beams according to a certain order and power, and the corresponding light detection arrays receive echo signals.

Finally, it should be noted that the above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, the The technical solutions described in the foregoing embodiments may be modified, or some technical features thereof may be equivalently replaced. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

A method of determining noise levels, including:

S101: Obtain an intensity information-time information curve of an optical signal;

S102: Clamp the intensity information-time information curve with an estimated noise threshold; and

S103: Determine the noise level using the clamped intensity information-time information curve.
The method of claim 1, wherein the intensity information-time information curve is a photon number histogram.
The method according to claim 1, wherein the step S101 comprises: scanning a field of view multiple times, and superimposing the curves of the intensity information relative to the time information of the multiple scans as the intensity information-time information curve.
The method according to claim 2, further comprising: acquiring a total intensity of the optical signal, calculating an estimated noise intensity using the total intensity, comparing the estimated noise intensity with a preset noise intensity, obtaining a smaller value, and calculating The smaller value is used as the estimated noise threshold.
The method according to claim 4, wherein the step of calculating the estimated noise intensity comprises: multiplying the total photon number S by a preset ratio K to obtain the estimated noise photon number, as the estimated noise intensity, the The preset ratio K is between 0-1.
The method according to claim 4 or 5, wherein the step S101 further comprises: receiving an optical signal through a detection unit, the detection unit includes a plurality of single-photon avalanche diodes, and the preset noise intensity is based on the One or more of the number of single photon avalanche diodes, the dead time of the single photon avalanche diodes is determined.
The method according to any one of claims 1-4, wherein the step S103 comprises: obtaining the total intensity of the clamped intensity information-time information curve, relative to the average value of the time span, as the noise level.
The method according to any one of claims 1-4, wherein in step S102, for the intensity information corresponding to any time information, the smaller value between the intensity information and the estimated noise threshold is taken as the clamp strength information after.
A method for ranging by lidar, comprising:

S201: Calculate the noise level by the method according to any one of claims 1-8;

S202: Perform noise filtering processing on the intensity information-time information curve based on the noise level; and

S203: Calculate the distance of the target object based on the intensity information-time information curve after noise filtering.
The method according to claim 9, wherein the step S202 comprises: judging whether the peak value of the intensity information-time information curve is higher than the noise level; when the peak value is higher than the noise level, based on the Noise level, performing noise filtering processing on the intensity information-time information curve.
The method according to any one of claims 9 or 10, wherein the step S203 comprises: calculating the center of gravity of the intensity information relative to the time information according to the intensity information-time information curve after noise filtering, and the center of gravity corresponds to time information as flight time.
The method according to any one of claims 9 or 10, wherein the step S203 comprises: calculating the leading edge time of the echo pulse according to the filtered intensity information-time information curve, and using the leading edge time as Time of flight, wherein the leading edge time is the time information corresponding to the intensity information on the leading edge of the echo pulse equal to a preset threshold.
A lidar comprising:

a light emission module, configured to emit a detection laser beam for detecting a target;

a light detection module configured to receive echoes of the detection laser beam reflected on the target and convert them into electrical signals; and

A control module, coupled to the light emitting unit and the light detection module, and configured to perform the method of any one of claims 9-12 to calculate the distance of the target.
14. The lidar of claim 13, wherein the light detection module includes a plurality of detection units, each detection unit including a plurality of single-photon avalanche diodes to receive the echoes.