Measuring distance of object using light time-of-flight method

Abstract

The invention relates to measuring the distance of an object using a light time-of-flight method. There is provided a photoelectric sensor for measuring a distance of an object in a detection area using a light time-of-flight method, the sensor comprising: a light emitter for emitting a light signal into the detection area; a light receiver having a first plurality of avalanche photodiodes in Geiger mode for detecting received light from the detection region; a second plurality of light time-of-flight measurement units for determining a single light time-of-flight between the emission of the light signal and the triggering of the probe event in the avalanche photodiode; and a control and evaluation unit which is designed to collect the individual light time-of-flight in the histogram, to locate the useful light signal in the histogram on the basis of a threshold value, and to determine a distance value to the object as a function of the useful light signal. An extraneous light level is first estimated from the histogram, and then a threshold is determined based on the extraneous light level such that the threshold lies above the expected number of exponentially decaying noise and extraneous light events with a safety margin.

Description

[0001] The invention relates to a photoelectric sensor and a method for measuring the distance to an object in a detection area using the light time-of-flight method according to the preambles of claims 1 and 11.

[0002] Distance measurement or ranging can be used in various areas, such as factory automation, logistics automation or security technology. According to a variation of the time-of-flight principle (dToF, direct Time of Flight), short light pulses are emitted and the time until the diffusely reflected (remittiert) or reflected (reflektiert) light pulse is detected is measured. Possible applications of distance measurement are improved light curtains that monitor the distance between their transmitter and receiver or reflector or switching systems with binary object presence recognition, in which the switching state depends on whether the object is within a certain distance range. The latter type of sensor is also called a background suppression light detector. By using corresponding spatially resolved receivers, single-beam or one-dimensional distance measurement can be extended to linear or planar distance measurement. Laser scanners are also based on time-of-flight measurement to determine the distance of corresponding angular positions.

[0003] In many applications, the detection sensitivity of a simple photodiode as a light receiver is insufficient. In an avalanche photodiode (APD), incident light triggers a controlled avalanche breakdown (avalanche effect). The charge carriers generated by the incident photons are thus multiplied, generating a photocurrent that is proportional to the received light intensity but significantly larger than in a simple PIN diode. In the so-called Geiger mode, a bias voltage (bias) is applied to the avalanche photodiode that is higher than the breakdown voltage (breakdown voltage), so that even a single charge carrier released by a single photon can trigger an avalanche, which then recruits all available charge carriers due to the high field intensity. Thus, an avalanche photodiode counts single events, just like the so-called Geiger counter. An avalanche photodiode in Geiger mode is also called a SPAD (Single-Photon Avalanche Diode).

[0004] Geiger APDs, or SPADs, are very fast, highly sensitive semiconductor-based photodiodes. The drawback of this high sensitivity is that avalanche breakdown can be triggered not only by useful photons, but also by weak interference events caused by extraneous light, optical crosstalk, or dark noise. Consequently, each SPAD, in addition to measuring the desired time-of-flight of the light, will also mistakenly detect interference events as a false time-of-flight. The interference event then affects the measurement result with a signal as strong as the received useful light and cannot be distinguished from the signal. Following avalanche breakdown, regardless of its cause, the sensitivity of the avalanche photodiode drops sharply within a dead time, or recovery time, of approximately 5 to 100 ns, rendering it virtually incapable of further measurements. Consequently, under the influence of extraneous light, the SPAD pool available for measurement decreases (pile-up effect).

[0005] Therefore, for the measurement, it is necessary to distinguish useful events from interference events. To this end, multiple single measurements are usually performed with multiple SPADs and / or repeated measurements are performed to enable statistical evaluation, in particular by collecting the single measurement results in a histogram. In the histogram, useful events form peaks, which can in principle be detected using threshold values. However, the difference between useful light and interference events depends greatly on the respective measurement situation. Slower drift (for example, degradation of the optical transmitter) as well as manufacturing variations and component tolerances also have an impact. Therefore, determining the threshold value is very challenging, and even the best-adjusted static threshold value is only valid under strictly controlled measurement conditions.

[0006] DE 10 2021 118 660 A1 describes a laser scanner using a single-photon avalanche diode and histogram evaluation with the aid of an optimization filter.

[0007] EP 3 428 683 A1 discloses a photosensor with a light receiver comprising multiple SPADs. Some of these SPADs are connected to a light time-of-flight measurement unit in a 1:1 configuration. This allows for the selection of a region of interest for the light receiver, based on which the light time-of-flight measurement is performed. However, this document does not provide a solution to the threshold problem described.

[0008] In EP 3 418 767 B1, in another general-purpose photoelectric sensor, a parameter describing the exponentially decaying frequency of background events is estimated based on a binomial model and further filtered using a median filter. This estimation is based on a corresponding histogram, in which measurement events and interference events appear mixed, and the median filter achieves separation. On the one hand, a single histogram has a very limited statistical basis. Furthermore, in certain measurement situations, such as when an object enters the detection area from the side and hits an edge, quite random and multiple peaks appear, which the median filter cannot process in the expected manner. Furthermore, the computational cost of the median filter is relatively high, as it requires a sorting operation, which reduces the sensor's response time on limited, especially embedded, hardware.

[0009] Maik Beer's doctoral thesis "SPAD-based Sensors for Distance Measurement" was published in 2018 at the University of Duisburg-Essen in Duisburg and Essen. ” deals with the effects of background light on distance measurements of general sensors.

[0010] The object of the present invention is therefore to further improve the light time-of-flight measurement of a SPAD-based distance measurement sensor.

[0011] This object is achieved by a photosensor, in particular a photosensor of the type mentioned in the introduction, for measuring the distance to an object in a detection area using the time-of-flight method according to claims 1 and 11. As is common in time-of-flight measurement, a light transmitter emits a light signal, which is received by a light receiver after diffuse or direct reflection from the object whose distance is to be measured. Preferably, the light signal has short pulses, enabling the use of pulse-based methods (direct time-of-flight). The light receiver comprises a first plurality of avalanche photodiodes or pixels, which can be operated in Geiger mode, in which they are biased with a voltage greater than their breakdown voltage so as to trigger an avalanche event upon receiving light. The term SPAD will also be frequently used hereinafter for avalanche photodiodes in Geiger mode. A second plurality of time-of-flight measurement cells determines the respective individual light flight times between the emission and reception of the light signal. In this case, the time-of-flight measurement cells are associated with specific avalanche photodiodes; depending on the embodiment, the relationship can be 1:n, 1:1, or n:1, and can be of different or identical group sizes. In particular, as described in EP 3 428 683 A1, only the avalanche photodiodes that are fully evaluated by the light time-of-flight measurement unit can be selected in order to determine the region of interest in the optical receiver. A single light time-of-flight measurement result is initially an uninterpreted measurement result that can correspond to both useful events and interference events in the light signal returning from the object.

[0012] The control and evaluation unit evaluates the individual light flight times. To this end, the individual light flight times are first collected in a histogram using multiple light flight measurement units and / or repeated measurements, each time emitting a light signal and waiting for the measurement duration. The histogram classifies the individual light flight times into discrete measurement periods (time bins) and calculates the frequency (counts per bin) of the corresponding individual light flight times. The useful light signal or peak is then located in the histogram based on a threshold value. The distance value is determined based on the temporal position of the useful light signal.

[0013] The basic concept of the present invention is to first estimate the ambient light level as information about the current measurement situation, so that the threshold value can be dynamically adjusted. The determined threshold value takes into account the model assumption that background events decay exponentially over time, and maintains a buffer distance from interference events (i.e., noise events and ambient light events) defined by a safety margin. The threshold curve is parameterized based on the currently measured ambient light level. In specific implementations, it may be advantageous to use a more easily computable function (e.g., a polynomial) as an approximation of the exponential function.

[0014] The advantage of the present invention is that it enables reliable measurements even under the influence of ambient light. Due to the dynamic adjustment of the threshold, low-diffuse-reflecting objects at a distance can be detected in low ambient light conditions, while the threshold setting can avoid false detections in the presence of increased ambient light. The threshold used for the current ambient light level can be carefully prepared and defined during development, possibly with a significant investment of time and an extensive database. This also allows for consideration of drift or degradation processes of individual sensors across the entire product range, as well as sample dispersion. At runtime, the appropriate threshold is selected based on the current ambient light level, using only minimal computing resources.

[0015] Preferably, the light time-of-flight measurement unit has a TDC (Time-to-Digital Converter). This is a well-known and relatively simple component that can determine the flight time of a single light beam with high temporal resolution. The TDC can be monolithically integrated directly into the crystal of the optical receiver. Preferably, the corresponding TDC starts at the emission time and stops at the reception time due to a single received light pulse or in the event of an interference event caused by extraneous light or dark noise. Other operating modes are conceivable, such as starting the TDC each time an avalanche is triggered and then stopping it at a known point in time, such as the end of a measurement period.

[0016] Preferably, the control and evaluation unit is designed to estimate the extraneous light level by accumulating the first interval of the histogram. This is used because no useful light has yet returned here, and the pile-up effect causes only a small number of avalanche photodiodes to enter their dead time. It should be noted that the avalanche photodiodes are preferably inactive before the measurement begins, so that at least initially, a full pool of avalanche photodiodes is still available. Ideally, the decaying exponential function can be reconstructed from two intervals. In practice, the measurement is a random experiment, so a larger number of intervals should be used. Nevertheless, when accumulating the first interval, a partial selection of intervals, such as the second, fifth, and eighth intervals, should also be included, and preferably only all intervals 1 through N should be considered. Regardless, the counts in the first interval are highly correlated with the extraneous light level, and this relationship is used to determine the extraneous light level. Alternatively, the extraneous light level can be determined using an additional dedicated extraneous light receiver or at least one avalanche photodiode reserved for this purpose in the optical receiver.

[0017] Preferably, the control and evaluation unit stores a first lookup table, which correlates the number of accumulated detection events with the ambient light level. "Stored" means that the first lookup table (LUT) is stored in the control and evaluation unit or in a memory accessible to the control and evaluation unit. The determination of the ambient light level at runtime is then limited to an accumulation of the first intervals and a simple table access in order to obtain a matching value for the ambient light level from the sum of the first intervals. This is a very fast and easy to use implementation that requires minimal hardware resources. Intermediate values ​​that are missing from the first lookup table can be obtained by interpolation. For this purpose, preferably a fixed given function is used, in particular a polynomial, more preferably a function that has already been used for calibrating the first lookup table, as will be explained below. However, a simple linear interpolation is usually sufficient.

[0018] Preferably, before distance measurement, the first lookup table is calibrated by repeatedly exposing the light receiver to defined ambient light levels with the light transmitter inactive, and determining the corresponding sums of the first bins of the histogram generated during this period. In particular, the ambient light is generated by an additional light source. Calibration is performed before the actual measurement operation, for example, during development or factory production. The light receiver is systematically exposed to varying ambient light levels of known intensity, but without the superposition of useful light due to the inactive light transmitter. The histograms are recorded separately, and their first bins are summed. This generates data pairs that relate the sums of the first bins to the ambient light level. From this, the relationship between the ambient light level and the sums of the first bins can be derived from the first lookup table with any desired accuracy, either by averaging or, preferably, by fitting a function or polynomial. The fitted function can be stored in the sensor for later use in interpolation.

[0019] Preferably, the control and evaluation unit is designed to delay the emission of the light signal relative to the start signal so that no useful light signal is recorded in the first interval of the histogram. This allows the distance measurement range to begin immediately in front of the sensor while still ensuring that the first interval is free of useful light, from which the ambient light level can be estimated. The additional delay can be easily calculated from the measured light time-of-flight. Appropriate calibration is generally required for unavoidable internal signal delays, which can replace or supplement artificial delays.

[0020] The control and evaluation unit preferably stores a second lookup table that associates at least one suitable parameter of a calculation rule for the threshold value with the ambient light level. Thus, the threshold value is specified as a calculation rule with at least one previously undetermined parameter, in particular a component with a decaying exponential function and a component preferably also based on an exponential function for a safety margin. At least one parameter is then read from the second lookup table based on the ambient light level, thereby dynamically adapting the threshold value to the currently prevailing ambient light level. The second lookup table is identical to the first lookup table with regard to storage location, type of use, possible interpolation, and advantages.

[0021] The second lookup table is preferably calibrated prior to the distance measurement by repeatedly exposing the light receiver to defined ambient light levels without the light transmitter being activated and determining the corresponding sum of the histograms generated during this period. and the center of gravity where Histogram(i) represents the interval i = 1 ... M in the histogram. The sum HS and the center of gravity λ are then used as or in the parameters of the threshold. The calibration scenario is similar to that of the first lookup table, in particular, both lookup tables can be calibrated simultaneously.

[0022] Preferably, the control and evaluation unit is designed to The threshold is determined using parameters a and b, as well as a scaling factor S for a safety margin. This is a specific calculation rule that closely simulates the behavior of the optical receiver being used. Parameters a and b are set based on the prevailing ambient light level and are preferably simply read from a second lookup table. The first term models the expected curve of the noise margin. To account for the discreteness of individual measurements, the threshold is shifted upward by the second term as a buffer. Unlike a and b, the safety margin S is not dependent on the ambient light level; the component below the square root already accounts for this dependence. Instead, the safety margin S represents a trade-off between the Type I and Type II errors—that is, the preference for mistaking objects for noise or, conversely, for falsely detecting rare, high-noise events as objects. The sensor is configured accordingly by setting S at the factory or at the operating site.

[0023] For parameters a and b, preferably apply and In this case, λ is preferably derived from the center of gravity of at least one histogram determined at a defined ambient light level. These parameters are derived from a modeling of the light receiver behavior, which has proven to be very effective in practice.

[0024] Preferably, the control and evaluation unit is designed to compare the number of individual light flight times for each bin of the histogram with a threshold value and to associate the bins that exceed the threshold value with the useful light signal. Thus, the threshold value is applied bin by bin in order to find those bins in which the returned emitted light is detected. This localizes the useful light signal in time. In many measurement situations, there is only one such cluster of bins corresponding to a single useful light peak. The light flight time can then be determined as the maximum value, center of gravity or a similar measure of these bins, or by fitting a peak function in these bins. If there are multiple, non-adjacent, consecutive bins that exceed the threshold value, multiple distance values ​​can be output, or, for example, the first peak, the last peak or the most pronounced peak can be used as the basis for the distance value.

[0025] The method according to the invention can be further developed in a similar manner and exhibits similar advantages. Advantageous features are described exemplarily but not exhaustively in the dependent claims which are dependent on the independent claim. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Other features and advantages of the present invention will be described in more detail below based on exemplary embodiments and with reference to the accompanying drawings. In the accompanying drawings:

[0027] Figure 1 shows a schematic diagram of a photosensor using light time-of-flight measurement;

[0028] Figure 2 A schematic diagram showing an optical receiver and downstream components used for measurement evaluation;

[0029] Figure 3 An exemplary histogram of the flight time of a single light beam measured in the case of pure external light incidence without useful light is shown, wherein the accumulation of the first interval is illustrated;

[0030] Figure 4 Shows something like Figure 3 Graphs of six exemplary histograms for different ambient light levels;

[0031] Figure 5 a diagram showing a relationship between the accumulated first interval and the ambient light level;

[0032] Figure 6 A diagram showing modeling of the extraneous light portion of a histogram and a threshold value with a safety margin thereon;

[0033] Figure 7 Shows something like Figure 6 A diagram of six exemplary threshold values ​​for different ambient light levels; and

[0034] Figure 8 Shows something like Figure 6 , but now with useful light peaks identified by thresholding.

[0035] Figure 1 A schematic diagram of a photoelectric sensor 10 for distance measurement based on the time-of-flight principle is shown in a one-dimensional embodiment. Sensor 10 is described as representative of other optical time-of-flight measurement sensors, in particular as mentioned in the introduction. A light emitter 12 (e.g., an LED or laser light source) emits a light signal 14 into a monitoring area 16. If an object 18 is present in the monitoring area, some of the light is diffusely reflected or reflected and returns to sensor 10 as a diffusely reflected light signal 20, where it is recorded by a light receiver 22.

[0036] The light receiver 22 includes a plurality of pixel elements 24, also known as SPADs (single photon avalanche diodes), which can be operated in Geiger mode so as to trigger an avalanche event upon receiving light by applying a bias voltage greater than the breakdown voltage to the pixel elements. Some basic SPAD characteristics have been described in the introduction. The pixel elements 24 are preferably arranged in a matrix. The number of pixel elements 24 can vary, and the matrix can be, for example, a square or rectangular arrangement with dozens, hundreds, or even thousands or more pixel elements 24.

[0037] The light receiver 22 is connected to the sensor control block 26. Figure 1 This is only briefly shown in FIG. 2 , and the possible structure of the sensor control block 26 will be referred to in FIG. Figure 2 The sensor control block 26 controls the light emitter 12 so that the light signal 14 is emitted, preferably in the form of short pulses in the nanosecond or even picosecond range. The time point at which the light signal 14 is triggered can be used as a reference for the time-of-flight measurement. In other embodiments, a part of the light signal 14 can be used internally as an optical reference. The sensor control block 26 processes the signals of the pixel elements 24 and evaluates these signals to determine the time of flight from the emission time point of the emitted light signal 14 to the reception time point of the diffusely reflected light signal 20. The time of flight can be converted into a distance with the aid of the speed of light. This will be referred to later. Figures 2 to 8 Determination of the reception time point will be described.

[0038] In practice, the sensor 10 comprises other elements, in particular transmitting and receiving optics and interfaces, which are known per se and are omitted here for the sake of simplicity. Figure 1The light receiver 22 and the sensor control block 26 are shown as separate, which is possible in a practical embodiment, but is mainly for illustrative purposes. Preferably, these components are at least partially integrated on a common chip, the surface of which is shared by the pixel elements 24 and the circuits assigned or assignable to the pixel elements 24 or groups of pixel elements 24 for controlling and evaluating the pixel elements.

[0039] exist Figure 1 A coaxial arrangement is shown in FIG, wherein the light emitter 12 is arranged in front of the light receiver 22. Other coaxial arrangements are also possible, for example with the aid of a beam splitter. A biaxial arrangement or a triangulated arrangement is also conceivable, wherein the light emitter 12 and the light receiver 22 are arranged side by side in a mutually staggered manner. The sensor 10 can be Figure 1 One-dimensional sensor of the type shown. Other non-exhaustive embodiments are light curtains, light barriers and laser scanners. The sensor 10 can output or display a distance value or can also act as a switch, i.e. trigger a switching event when an object within a certain distance range (including deviations from the expected distance range) is detected. A plurality of sensors 10 can be combined, for example to form a light barrier for distance measurement or distance monitoring. Mobile systems are also conceivable, in which the sensor 10 is mounted movably, or scanning systems, in which the emitted light signal 14 is swept across the monitoring area 16 by means of a movable mirror or by movement of the measuring system, in particular by a rotational movement.

[0040] Figure 2 A schematic diagram of the downstream components of the light receiver 22 and the sensor control block 26 is shown. The light receiver 22 is again shown as a SPAD matrix with a plurality of pixel elements 24. Some of the pixel elements 24 are connected to a time-of-flight measurement unit 28, which in this embodiment is a time-to-digital converter (TDC). A switching device 30 determines the connection, i.e. which pixel elements 24 are selected for evaluation and by which light time-of-flight measurement unit the pixel elements are evaluated respectively. The light time-of-flight information generated by the light time-of-flight measurement unit 28 is accumulated and evaluated by a control and evaluation unit 32, preferably after storing the accumulated light time-of-flight information in a memory not shown separately, and preferably in the form of a histogram. The result of the evaluation is a distance value, which can serve as a basis for further evaluation.

[0041] The switching device 30 can be designed as a programmable matrix or in other ways so as to connect selected pixel elements 24 to selected time-of-flight measurement units 28 in a 1:1 or n:1 scheme. It is not necessary to evaluate all pixel elements 24. The reason for forming only selected connections is that a large number of light time-of-flight measurement units 28 corresponding to the number of pixel elements 24 may not be possible, or at least the cost is too high and requires too large a chip area. Therefore, preferably, the number of light time-of-flight measurement units 28 is only a fraction of the number of pixel elements 24. In addition, the signal-to-noise ratio can be improved by preferentially selecting pixel elements 24 within a region of interest (ROI) that actually receives the diffusely reflected light signal 20.

[0042] The time-of-flight measurement unit 28 measures the corresponding individual light flight times between the emission of the emitted light signal 14 and the reception of the diffusely reflected light signal 20. In one embodiment, the light flight time measurement unit 28 is activated with the emission of the light signal 14 and is stopped due to an avalanche event in the connected pixel element 24. In another embodiment, the light flight time measurement unit is activated due to an avalanche event and is stopped at a reference time point, wherein the offset between the emission time point and the reference time point is compensated by calculation. Each individual light flight time is highly unreliable by itself, because the measured avalanche breakdown may be caused by ambient light or dark noise rather than by the diffusely reflected light signal 20, and therefore the corresponding individual light flight time may be completely unrelated to the distance to be measured.

[0043] Therefore, the individual light flight times of the light flight time measurement unit 28 are accumulated in a histogram, preferably through repeated measurements for better statistics, and evaluated by the control and evaluation unit 32. The histogram divides the measurement period into time intervals, and in each interval, the number of individual light flight times falling within the time interval of the interval is counted. When the diffusely reflected light signal 20 is received, a peak is formed in the histogram, and the time position of the peak determines the reception time point.

[0044] As mentioned above, the components of the optical receiver 22 and the sensor control block 26 can be integrated on the same chip. In a preferred embodiment, the optical receiver 22, the light flight time measurement unit 28 and the switching device 30 are part of an ASIC (Application-Specific Integrated Circuit), while the control and evaluation unit 32 is implemented on a microprocessor. In another embodiment, the control and evaluation unit 32 is also at least partially integrated into the ASIC. The memory for the histogram can be part of the ASIC, the microprocessor or a separate component. This is only a preferred hardware implementation. Alternatively, the functions of the sensor control block 26 can also be implemented on one or more arbitrary hardware components, such as an ASIC, a CPU (Central Processing Unit), an FPGA (Field Programmable Gate Array), a DSP (Digital Signal Processor), etc.

[0045] Figure 3 An exemplary histogram of individual light flight times measured in the case of pure extraneous light incidence of useful light without diffusely reflected light signal 20 is shown. The corresponding number (count) of individual light flight times is displayed on the Y-axis as the height of the bar of the corresponding interval on the X-axis. In the absence of useful light, the histogram contains only an exponentially decaying noise portion or background events (PileUp - pile-up effect), which is caused by the dead time of the pixel element 24 after the avalanche event. There are a total of M intervals, corresponding to the measurement period or the maximum detectable distance. Due to the unavoidable and / or intentional delay from the internal trigger of the emission light signal 14 to the actual emission time point, the useful light reception can only start from interval N+1 at the earliest. Therefore, the first N intervals can be used for pure extraneous light estimation at runtime, and then other noise events are simply classified as extraneous light parts. The accumulated first N intervals highlighted in box 34, i.e. where Histogramm(i) is the count in the i-th bin, which is closely related to the ambient light level.

[0046] Figure 4 Shows something like Figure 3 Graphs of six exemplary histograms for different ambient light levels. By exposing the sensor 10 or its light receiver 22 to such known and defined ambient light levels before a measurement operation (e.g. during development, manufacturing or commissioning), a desired number of data pairs (FS ) for different ambient light levels (e.g., FP0=0 klux, FP1=1 klux, ...) can be obtained. n ,FP n ).

[0047] Figure 5 The relationship between FP and FS is shown, that is, the relationship between the external light level and the accumulated first N intervals. n ,FP n ) can be used to fit functions such as straight lines or FS Fit (x) = p0 + p1x + p2x 2 For a particularly efficient implementation, a first lookup table LUT1 can be generated therefrom with any desired precision. Alternatively, only the coefficients p0, p1, and p2 can be stored. During measurement operation, the accumulated measurements of the first N intervals FS are converted into an estimated ambient light level FP using the first lookup table LUT1. Intermediate values ​​may be rounded or interpolated.

[0048] As an alternative to a dynamic estimation of the ambient light level, a fixed configuration is conceivable, in which a specific ambient light resistance is set at the factory or at the customer's site, and the FP is pre-specified as a parameter. Furthermore, as an alternative to estimating the ambient light level by accumulating the FS of the first N intervals, it is conceivable to use an additional receiving element or at least one pixel element 24, in which only the ambient light is measured, separated from the useful light channel.

[0049] Figure 6 A diagram showing the modelling of an exponential curve 36 of the extraneous light portion of the histogram and a threshold value 38 with a safety margin thereon is shown. Determining the threshold value 38 is the real purpose, estimating the extraneous light level FP is an intermediate step to achieve this.

[0050] For each external light level FP, an exponential curve 36 can be estimated using several parameters. The general form of the density function of the exponential distribution is:

[0051]

[0052] For each measured ambient light histogram, the model parameters λ can be estimated using the maximum likelihood method, taking into account all measured values. The maximum likelihood method enables robust parameter estimation using all available measured data in the histogram.

[0053] For an exponential distribution, a reliable estimate of λ is the mean or centroid of all the individual light flight times measured in the histogram, i.e.:

[0054]

[0055] is the sum of the numbers of all bins in the histogram.

[0056] The exponential curve 36 of the extraneous light portion is the expected value E(i) of the number of background events in each histogram bin i and is calculated as follows:

[0057]

[0058] This is not yet the threshold 38, because there are at least some intervals 40, 42 with a number (count) of random increases compared to the expected value E(i) that are still higher than the expected value. Therefore, a safety margin between the two must also be considered.

[0059] Assuming that the extraneous light events are Poisson distributed on the interval , the standard deviation can be calculated directly, since the expected value and variance are the same here:

[0060]

[0061] By scaling the standard deviation by the desired factor S (e.g., a safety margin in the range S=2...6), the relevant threshold value 38 can now be determined, which is referred to as the noise margin RG(i) of the corresponding i-th interval in the following formula:

[0062] RG(i)=E(i)+S*Std(i).

[0063] Now if for a particular external light level FP n , the parameter λ is determined from the extraneous light histogram by the maximum likelihood method as described above n and HS n , then the threshold 38 applies as follows:

[0064]

[0065] Now change the parameter (a n ,b n ) is defined as:

[0066]

[0067] This simplifies the threshold value of 38 to:

[0068]

[0069] A second lookup table LUT2 may be formed, which converts the corresponding external light level FP n With parameter (a n ,b n ) is associated. In the measurement operation, it is only necessary to find the current external light level FP n Parameters (a n ,b n ), you can use the above RG n(i) is used to calculate the threshold value 38 for each interval i. As mentioned above, the external light level FP n In turn, it can be obtained from the first lookup table LUT1 or measured in another way. In order to reduce the workload of calculating exponential functions and square roots, an approximation method, for example in the form of a polynomial, can be used.

[0070] Figure 7 Shows something like Figure 6 3 is a diagram of six exemplary threshold values ​​38 for different extraneous light levels FP, which are determined according to the method just described.

[0071] Figure 8 Shows something like Figure 6 , for illustrating the further evaluation using a suitable threshold value 38. In contrast to the histograms shown so far, here also useful light peaks 44 are recorded, which are reliably distinguished from the background by the threshold value 38. For example, the number of each bin i is related to the associated threshold value RG n (i) Comparison. The intervals with a number exceeding the threshold contain information about the temporal position of the useful light peak 44 being searched for. If there is more than one interval exceeding the threshold, for example, a maximum, a center of gravity or a comparable measure can be formed from the intervals exceeding the threshold, in particular weighted with an exponential decay, in order to determine the reception time point. Alternatively, a curve of the transmitted pulse can be fitted, which can also be fitted in a simplified form (e.g., a parabola). If there are multiple discontinuous intervals exceeding the threshold, the first, the most obvious, the last or a cluster pre-specified according to a rule is selected, or multiple reception time points are calculated for multiple clusters. Such multi-target measurements occur, for example, when an object is behind another (semi-)transparent object.

[0072] In a particularly preferred embodiment, the two lookup tables LUT1 and LUT2 are determined as described above during development, sometimes requiring considerable time and measurement effort, and are stored in the corresponding sensor 10. In operation, it is only necessary to calculate the sum FS of the first N bins of the histogram and read out the extraneous light level FP from the first lookup table LUT1. n In turn, the corresponding noise boundary RG is read from the second lookup table LUT2 n , or according to several parameters of the second lookup table LUT2 (for example, (a n ,b n )), a predefined calculation rule for the threshold value 38 is adjusted. This threshold value 38 is then used to reliably identify the useful light peak 44 against the background of the current ambient light situation.

Claims

1. A photoelectric sensor (10) for measuring the distance of an object (18) in a detection area (16) using a time-of-flight method, wherein: The sensor (10) comprises a light transmitter (12) for transmitting a light signal (14) into the detection region (16); a light receiver (22) having a first plurality of avalanche photodiodes (24) in Geiger mode for detecting received light (20) from the detection region (16); a second plurality of light flight time measurement units (28) for determining individual light flight times between the emission of the light signal (14) and the triggering of a detection event in the avalanche photodiodes (24); and a control and evaluation unit (32) designed to collect the individual light flight times in a histogram, to locate a useful light signal (44) in the histogram based on a threshold value (38), and to determine a distance value to the object (18) based on the useful light signal (44). It is characterized by: The control and evaluation unit (32) is further designed to first estimate the extraneous light level from the histogram and then determine the threshold value (38) based on the extraneous light level such that the threshold value lies above an expected exponentially decaying number of noise and extraneous light events with a safety margin.

2. The sensor (10) according to claim 1, wherein The control and evaluation unit (32) is designed to estimate the ambient light level by accumulating a first bin of the histogram.

3. The sensor (10) according to claim 2, wherein The control and evaluation unit (32) stores a first lookup table which relates the number of accumulated detection events to the ambient light level.

4. The sensor (10) according to claim 3, wherein Prior to distance measurement, the first lookup table is calibrated by repeatedly exposing the light receiver (22) to defined ambient light levels with the light transmitter (12) inactive and determining the corresponding sums of the first bins of the histogram generated during this period.

5. The sensor (10) according to any one of the preceding claims, wherein The control and evaluation unit (32) is designed to delay the emission of the light signal (14) relative to the start signal so that no useful light signal is recorded in the first interval of the histogram.

6. The sensor (10) according to any one of the preceding claims, wherein The control and evaluation unit (32) stores a second lookup table that associates at least one suitable parameter of the calculation rule for the threshold value (38) with the ambient light level.

7. The sensor (10) according to claim 6, wherein Prior to distance measurement, the second lookup table is calibrated by repeatedly exposing the light receiver (22) to defined ambient light levels with the light transmitter (12) inactive and determining the corresponding sum of the histograms generated during this period. and the center of gravity Here, Histogram(i) represents the interval i=1…M in the histogram.

8. The sensor (10) according to any one of the preceding claims, wherein The control and evaluation unit (32) is designed to The threshold value (38) is determined, which includes the parameters a and b and a scaling factor S for the safety margin.

9. The sensor (10) according to claim 8, wherein For the parameters a and b, apply and Here, λ is derived from the center of gravity of at least one histogram determined at a defined ambient light level.

10. The sensor (10) according to any one of the preceding claims, wherein The control and evaluation unit (32) is designed to compare the number of individual light flight times per bin of the histogram with the threshold value (38) and to associate bins exceeding the threshold value with the useful light signal (44).

11. A method for measuring the distance of an object (18) in a detection area (16) using a time-of-flight method, wherein: a light signal (14) is emitted into the detection region (16), a light receiver (22) having a first plurality of avalanche photodiodes (24) in Geiger mode detects the received light (20) from the detection region (16), a second plurality of light flight time measurement units (28) determines a single light flight time between the emission of the light signal (14) and the triggering of a detection event in the avalanche photodiodes (24), the single light flight times are collected in a histogram, a useful light signal (44) is located in the histogram based on a threshold value (38), and a distance value to the object is determined from the useful light signal (44), It is characterized by: The extraneous light level is first estimated from the histogram, and then the threshold (38) is determined based on the extraneous light level such that the threshold (38) lies above an expected exponentially decaying number of noise and extraneous light events with a safety margin.

Images