WO2020083780A1

WO2020083780A1 - Time-of-flight ranging using modulated pulse trains of laser pulses

Info

Publication number: WO2020083780A1
Application number: PCT/EP2019/078391
Authority: WO
Inventors: Mathias Müller; Gerald ZAHLHEIMER; Jonas FAUSER; Andreas BOLLU
Original assignee: Blickfeld GmbH
Priority date: 2018-10-24
Filing date: 2019-10-18
Publication date: 2020-04-30
Also published as: DE102018126522A1

Abstract

A method of time-of-flight ranging comprises two correlations, one at a coarse time resolution and one at a fine time resolutions. Thereby, the computational resources required to determine a distance to an object can be reduced. LIDAR beam steering may be used.

Description

Time-of-Fliqht Ranging using Modulated Pulse Trains of Laser Pulses

TECHNICAL FIELD

Various examples generally relate to time-of-flight (TOF) ranging by transmitting a pulse train of laser pulses. Light Detection and Ranging (LIDAR) techniques are described. Various examples specifically relate to the processing of measured photon events.

BACKGROUND

Light detection and ranging (LIDAR; sometimes also referred to as laser ranging or LADAR) allows to provide a 3-D point cloud of a scene. Objects can be accurately detected. Ranging is possible. Pulsed or continuous-wave laser light is transmitted along a transmit beam and, after reflection at an object, detected along a receive beam. This allows to determine the distance to the object (z-position). Determining the z-position is also referred to as TOF ranging.

To enhance the measurement range of the LIDAR, techniques of using modulation of pulses are known. See, e.g.: US2018238998A; US2017269209A; WO2017132703; WO2018050906.

It has been observed that the amount of data required to be processed when employing modulated pulse trains is very high.

SUMMARY

Therefore a need exists for advanced techniques of TOF ranging using modulated pulse trains of laser pulses. Specifically, a need exists for advanced techniques that facilitate efficient and fast processing of measured photon events.

This need is met by the features of the independent claims. The features of the dependent claims define embodiments.

According to an example, a method of TOF ranging includes transmitting a pulse train of laser pulses. The pulse train is modulated in accordance with a modulation reference code. The method also includes measuring a sequence of photon events over a measurement time duration, using a detector. The method further includes binning the measured photon events into a sequence of bins. Each bin of the sequence of bins has a binning time duration. The method also includes performing a first correlation between a photon count of each bin of the sequence of bins and a correlation code. The correlation code is associated with the modulation reference code. The method further includes performing one or more second correlations between one or more sub- sequences of photon events and the correlation code. Here, the one or more sub-sequences of photon events are cropped from the sequence of photon events based on one or more correlation maximums of the first correlation. The method further includes determining at least one distance based on the one or second correlations.

For example, a detector having built-in amplification, e.g., an avalanche detector may be used. For example, an array detector may be used.

For example, the method may be implemented by a compute unit, e.g., by a Field Programmable Gated Array (FPGA).

A computer program, a computer program product, and a computer-readable medium are provided that include program code. The program code can be executed by at least one computing unit. Executing the program code causes the at least one computing unit to perform a method of TOF ranging which includes transmitting a pulse train of laser pulses. The pulse train is modulated in accordance with a modulation reference code. The method also includes measuring a sequence of photon events over a measurement time duration, using a detector. The method further includes binning the measured photon events into a sequence of bins. Each bin of the sequence of bins has a binning time duration. The method also includes performing a first correlation between a photon count of each bin of the sequence of bins and a correlation code associated with the modulation reference code. The method further includes performing one or more second correlations between one or more sub- sequences of photon events and the correlation code. Here, the one or more sub-sequences of photon events are cropped from the sequence of photon events based on one or more correlation maximums of the first correlation. The method further includes determining at least one distance based on the one or second correlations.

A device comprises a storage medium and at least one computing unit. The at least one computing unit can access the storage medium to load program code. Executing the program code causes the at least one computing unit to perform a method of TOF ranging which includes transmitting a pulse train of laser pulses. The pulse train is modulated in accordance with a modulation reference code. The method also includes measuring a sequence of photon events over a measurement time duration, using an array detector. The method further includes binning the measured photon events into a sequence of bins. Each bin of the sequence of bins has a binning time duration. The method also includes performing a first correlation between a photon count of each bin of the sequence of bins and a correlation code associated with the modulation reference code. The method further includes performing one or more second correlations between one or more sub-sequences of photon events and the correlation code. Here, the one or more sub-sequences of photon events are cropped from the sequence of photon events based on one or more correlation maximums of the first correlation. The method further includes determining at least one distance based on the one or second correlations.

It is to be understood that the features mentioned above and those yet to be explained below may be used not only in the respective combinations indicated, but also in other combinations or in isolation without departing from the scope of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 schematically illustrates a scan system including a coaxial optical setup and a frictionless steering mirror according to various examples.

FIG. 2 is a flowchart of a method according to various examples.

FIG. 3 schematically illustrates a transmit phase and a receive phase according to various examples.

FIG. 4 schematically illustrates a laser pulse transmitted during the transmit phase according to various examples.

FIG. 5 schematically illustrates a modulation reference code for a pulse train including multiple laser pulses according to various examples.

FIG. 6 schematically illustrates a sequence of photon events according to various examples.

FIG. 7 is a flowchart of a method according to various examples.

FIG. 8 schematically illustrates a sequence of bins into which the sequence of photon events are binned according to various examples. FIG. 9 schematically illustrates correlation maximums obtained from a first correlation between the modulation reference code and the sequence of bins according to various examples.

FIG. 10 is a flowchart of a method according to various examples.

DETAILED DESCRIPTION OF EMBODIMENTS

In the following, embodiments of the invention will be described in detail with reference to the accompanying drawings. It is to be understood that the following description of embodiments is not to be taken in a limiting sense. The scope of the invention is not intended to be limited by the embodiments described hereinafter or by the drawings, which are taken to be illustrative only.

The drawings are to be regarded as being schematic representations and elements illustrated in the drawings are not necessarily shown to scale. Rather, the various elements are represented such that their function and general purpose become apparent to a person skilled in the art. Any connection or coupling between functional blocks, devices, components, or other physical or functional units shown in the drawings or described herein may also be implemented by an indirect connection or coupling. A coupling between components may also be established over a wireless connection. Functional blocks may be implemented in hardware, firmware, software, or a combination thereof.

Hereinafter, techniques of TOF ranging and LIDAR are described. Using these techniques, it becomes possible to determine the distance towards an object in the environment using the round-trip time (RTT) of photons of a laser pulse. Thus, an active TOF ranging technique is used.

The techniques described herein can also be used to determine an intensity of reflected light, and, thereby a reflectivity of an object that is ranged. It is also possible to determine multiple distances, in case of multiple reflections.

According to various examples, a modulated pulse train of laser pulses is employed. For example, a Code Division Multiple Access (CDMA) modulation, e.g., using On-Off-Keying (OOK) may be used. For example, a pulse-time modulation can be employed where the inter- pulse distance between pulses of the pulse train is varied. Here, as a general rule, various types of respective modulation reference codes may be employed. To give an example, a so- called Gold sequence may be employed. Other examples include Kasami sequences, Hadamard sequences or Zadoff-Chu sequences. Different modulation references codes of the same type of code may be mutually orthogonal to each other. This helps to provide for interference resilience and facilitates CDMA.

Some of the techniques described herein benefit from a combination of one or more modulation schemes with beam steering. Here, different pulse trains are transmitted into different directions. For this, a steering mirror to steer light can be employed. Alternatively or additionally, it would also be possible to rotate or otherwise move the light source or light sources. For example, a FLASH-Lidar system may benefit from the techniques described herein where multiple directions are illuminated with primary laser light at the same time instance.

For example, the techniques described herein may facilitate 1 -D or 2-D steering of light. Scanning of light can correspond to repetitively redirecting light using different transmission angles. A scan rate or repetition rate of the scanning can be defined by steering cycles. For this, the steering mirror can be deflected accordingly.

It is possible to implement transmission angles by deflecting the steering mirror in accordance with one or more degrees of freedom of motion of a mass-spring system formed by an elastic mount and the steering mirror. For example, the steering mirror may be rotated, tilted, shifted, etc.. Examples of the degrees of freedom of motion that can be used to steer light include flexure and torsion of the at least one spring of the mass-spring system.

According to some examples, resonant motion of the mass-spring system is possible. Specifically, the respective freedom of motion may exhibit a respective resonance characteristic - sometimes also referred to as frequency response, i.e., deflection as a function of frequency. The resonance characteristic may have a peak of certain width in frequency domain. It is possible to select a driving force to exhibit a frequency within this resonance peak. For this, an actuator can be appropriately controlled. By resonant motion, large changes in the transmission angles can be achieved. Large scanning areas can be implemented.

Example actuators that can be controlled to drive the elastic scan unit resonantly include piezoelectric comb drives, magnetic drives, piezoelectric actuators, etc.. The steering mirror and the elastic mount can be part of a scan unit. A scan system (or, simply, scanner) may include the scan unit, a light source configured to emit light to be scanned, and/or a detector configured to receive secondary light. The scan system can also include one or more actuators to actuate the elastic mount, to thereby deflect the steering mirror.

To increase the measurement range and/or daylight suppression, spatial filtering can be employed. For example, using spatial filtering, daylight can be suppressed to a level that the individual return or daylight photon events can be counted and recorded individually, without crossing a limit on event recording bandwidth of the detector or digital circuit. Spatial filtering relates to a scenario where primary laser light - i.e., the transmitted laser pulses - is transmitted along the transmit beam and via the steering mirror towards the object; and secondary light reflected by the object is collected along a receive beam and via the same steering mirror. In other words, the transmit beam is at least partly overlapping with the receive beam. Specifically, the transmit beam and the receive beam can be overlapping in an overlap section which includes the reflective surface of the steering mirror.

Sometimes, such a technique of spatial filtering is referred to as a coaxial optical setup, because the OA of the transmit beam and the OA of the receive beam are co-axially aligned. As a general rule, there may be an offset between the transmit beam and the receive beam for a coaxial optical setup, as long as there is an overlap between the beam profiles of the transmit beam and the receive beam. In a coaxial optical setup, the transmit beam and the receive beam may share at least one common aperture, e.g., defined by the steering mirror.

Coaxial optical setups generally allow to restrict the solid angle from which light is collected and focused onto the detector. Thereby, background noise is reduced, by reducing the number of background photons, e.g., due to sunshine.

Various techniques described herein relate to the post-processing of the measurement data obtained from the detector. Various techniques are based on the finding that the number of calculations associated with performing a comparison between the measured data and expected data (based on the correlation) can be very high. According to various examples, this is mitigated by employing a two-step approach: first, a coarse search for candidate ranges; and second, one or more fine searches at the candidate ranges, to resolve ambiguities and determine the range at greater accuracy. FIG. 1 schematically illustrates a scan system 100 according to various examples. The scan system 100 includes a compute unit 90, a laser diode 101 , and a detector 102. The scan system 100 defines a coaxial optical setup.

The compute unit 90 can be implemented by an application-specific integrated circuit (ASIC) and/or a field-programmable array (FPGA) and/or a general purpose processor. The compute unit 90 may include an analog-to-digital converter and/or a time-to-digital (TDC) converter.

The laser diode 101 is controlled by the compute unit 90 to transmit primary laser light 1 1 1 , along a transmit beam 121. The transmit beam 121 passes a beam splitter 130, then passes a steering mirror 150, then passes a housing window 151 of the scan unit and then reaches a surrounding 190.

The primary laser light may be scattered at an object (not illustrated in FIG. 1 ), to generate non-coherent secondary light 1 12. The secondary light 1 12 - as well as background light and/or interfering light - travels along a receive beam 122 through the window 151 , passes the mirror 150 and then reaches the beam splitter 130. The beam splitter deflects the secondary laser light 1 12 towards a detector 102.

The OA of the transmit beam 121 and the OA of the receive beam 122 are aligned; hence, a coaxial optical setup is implemented. Both, the transmitter aperture of the transmit beam 121 , as well as the detector aperture of the receive beam 122 are defined by the steering mirror 150.

As a general rule, the detector 102 may be an avalanche photo diode (APD) detector or a single-photon APD detector (SPAD). Another example would be a Silicon photomultiplier (SiPM). The detector 102 may have built-in physical amplification, e.g., may be an avalanche detector. Another example would be a Silicon Photomultiplier(SiPM), conventional photon multiplier tube or any detector, which might be invented in the future, capable of single photon detection.

FIG. 1 illustrates an overlap section 125 of the transmit beam 121 and the receive beam 122. The overlap section 125 has and end at the beam splitter 130.

The compute unit 90 is configured to actuate the steering mirror 150 to implement different scanning angles. For this, an actuator 901 is coupled with the compute unit 90. Example actuators 901 include electrostatic drives, magnetic drives, or piezoelectric drives. The actuator 901 is configured to exert a force onto a first end of an elastic mount. Thereby, reversible deformation of a spring of the elastic mount 902 is caused, deflecting the steering mirror 150 that is coupled to a second end of the elastic mount 902 which is opposite from the first end. A spring element can extend between the first and second end (not illustrated in FIG. 1 )·

Example implementations of the elastic mount 902 are described in US20180143322A1 : FIG. 62 and in DE 10 2016 014 001 A1 and in DE102009058762A1 and in US20100290142A1 .

The compute unit 90 can read out the detector 102.

In the example of FIG. 1 , the optical setup as a so-called pre-scanner optical setup. This is because there are no lenses etc. having a significant optical impact on the transmit beam 121 arranged upstream of the steering mirror 150 along the transmit beam 121 (as would be the case for a post-scanner optical setup, not illustrated herein). Thus, the divergence of the transmit beam 121 present at the steering mirror 150 corresponds to the divergence of the transmit beam 121 leaving the scan system 100 towards the surrounding / environment 190. Additional beam shaping behind the steering mirror 150 is not provided for in the pre-scanner optical setup.

FIG. 2 is a flowchart of an example method that may be implemented by the scan system 100 according to the example of FIG. 1 . For example, the method of FIG. 2 may be implemented by the compute unit 90.

Initially, at block 1001 , a pulse train of laser pulses is transmitted. The pulse train may be modulated in accordance with one or more modulation schemes. For example, the pulse train may be modulated in accordance with a modulation reference code. For instance, the compute unit 90 may appropriately control the laser diode 101 , to transmit the pulse train.

Block 1001 may be generally referred to as a transmit phase.

Next, at block 1002, photon events are detected. For example, the compute unit 90 may appropriately control the array detector 102. Photon events relate to the detection of one or more photons per sample time interval. For example, a typical sample time interval may be in the range of 20 picoseconds, e.g., in the range of 10 ps to 30 ps.

Block 1002 may be generally referred to as a measurement phase. As a general rule, blocks 1001 and 1002 may be executed at least partly in parallel. Hence, the pulse train of laser pulses may be transmitted in block 1001 while already detecting photon events in block 1002.

Next, at block 1003, the photon events are processed, in accordance with the modulation scheme. Block 1003 may be generally referred to as a postprocessing phase.

For example, the postprocessing may be implemented by the compute unit 90. Specifically, one or more correlations between an expected signal pattern - according to the modulation scheme - and the measured signal pattern of the photon events can be performed. Then, one or more correlation maximums are obtained which indicate the RTT of the photons. Based on the one or more correlation maximums, the distance to the object can be determined.

As a general rule, it would be possible that (i) blocks 1001 and/or 1002, and (ii) block 1003 are at least partially implemented in parallel. Hence, while still transmitting pulses and/or measuring for further photon events, it is possible to initiate the postprocessing.

Further details with respect to the transmit phase and the measurement phase are illustrated in connection with FIG. 3.

FIG. 3 illustrates aspects with respect to the timing of transmitting a pulse train of laser pulses and measuring a sequence of photon events. Specifically, as illustrated in FIG. 3, there is a transmit phase 201 (cf. FIG. 2, block 1001 ) during which a pulse train of primary light is transmitted (the pulse train is not illustrated in FIG. 3 for sake of simplicity). The pulse train has a certain pulse train duration that corresponds to the transmit phase duration 202.

FIG. 3 also illustrates aspects with respect to measuring photon events that are caused by secondary light of laser pulses. The photon events are triggered by secondary light and ambient or interfering light, i.e., background photons, e.g., from the sun or other light sources. FIG. 3 schematically illustrates a measurement phase 205 that has a measurement phase duration 206. During the measurement phase 205, the detector 102 detects photon events and outputs respective measurement signals that are later on analyzed.

In the example of FIG. 3, the measurement phase 205 and the transmit phase 201 overlap partly in time domain. For instance, the transmit phase duration 202 can be in the range of 200 ns to 600 ns. This corresponds to distances of 30 meters to 90 meters. On the other hand, it is desired to detect objects that are as close as 10 m to 30 m. The detector 102 is activated appropriately early. Hence, it is possible that the first reflected secondary light is detected as a photon event before the last pulse of primary light has been fired.

A priori, it cannot be said at which RTT and range the object is situated. Thus, the measurement phase duration 206 has a significant length. Hence, it is required to measure for photon events over the extended measurement phase duration 206 - which corresponds to being able to detect objects within a significant range of, e.g., up to 200 meters or 300 meters or even up to 600 meters. For example, for closer objects, the respective photon events will occur earlier during the measurement phase 205; while for more remote objects, the respective photon events will occur later during the measurement phase 205.

An optional post-processing step is explained next. As illustrated in FIG. 3, sometimes, a photon event 516 may be detected comparably early in the measurement phase 205 (early photon event). Specifically, as illustrated in FIG. 3, the early photon event is detected at a point in time at which the transmit phase 201 is still ongoing. The early photon event 516 has a comparably large signal amplitude, e.g., larger than a predefined threshold. For instance, the predefined threshold may correspond to a count of not less than 100 or 1 ,000 photons associated with the early photon event 516. The large signal amplitude of the early photon event 516 can be easily detected, because it is well above the noise level given by interference and/or background photons. It is expected that secondary light being reflected at a nearby object - e.g., at distances of up to 30 meters or up to 60 meters - will have such a comparably large signal amplitude. Hence, from the presence of the early photon event 516, it can be deduced that the object has already been detected and accurate ranging can be performed based on the early photon event 516 alone - without the need to perform a correlation or the like. I.e., it is not required to harvest the coherence of multiple time-separate photon events induced by the modulation of the pulse train, because the object is comparably close and the signal is strong. In other words, the distance to the object can be determined without considering the pulse timing associated with the respective laser pulse of primary light within the pulse train. Hence, the modulation scheme varying the inter-pulse distance may be ignored when detecting such an early photon event 516.

In some examples, in response to detecting the early photon event 516, it becomes possible to prematurely abort the transmit phase 201 , at a time duration 202A. Then, laser pulse(s) of the pulse train after the time duration 202A are not transmitted. This is advantageous, because cooldown of a corresponding laser light source is facilitated. Also, eye safety restrictions can be better met.

Such an approach may, in particular, be helpful when using beam steering (cf. FIG. 1 ). Specifically, where beam steering is used, typically, the time duration during which the beam steering mechanism is static is comparably limited (e.g., the steering mirror, when resonantly actuated, moves at a certain fast scanning frequency, e.g., at 100 Hz to 500 Hz). Hence, it is not easily possible to, first, transmit an isolated test pulse to find any nearby object; and, then, second, conditionally transmit the modulated pulse train. Thus, and integrated approach where the early photon event is detected from the primary light of a laser pulse of the pulse train is helpful.

FIG. 4 schematically illustrates aspects with respect to a laser pulse 400. The laser pulse 400 can be part of the transmit phase 201. The laser pulse 400 can be transmitted as primary light.

The laser pulse 400 has a certain duration 401 and a certain amplitude 402. The time position 403 of the laser pulse 400 is also illustrated, e.g., arbitrarily defined with respect to the maximum position.

The time position 403 of multiple laser pulses 400 of a pulse train can be set in accordance with a modulation reference code. Details with respect to the modulation reference code are illustrated in connection with FIG. 5.

FIG. 5 schematically illustrates aspects with respect to a modulation reference code 300. As illustrated in FIG. 5, the modulation reference code 300 has a certain length corresponding to the measurement phase duration 202 (cf. FIG. 3) and includes multiple pulse timings 301 - 305. At each pulse timings 301 - 305, a laser pulse 400 (cf. FIG. 4) is transmitted, i.e., primary light is transmitted. Also illustrated in FIG. 5 is an example inter-pulse time gap 310. The inter- pulse time gap 310 illustrated in FIG. 5 is the minimum inter-pulse time gap of the modulation reference code 300; other inter-pulse time gaps have longer durations, e.g., between pulse timing 304 and pulse timing 305.

The function of the modulation is explained next. The photon events corresponding to the secondary light reflected at the object will have the same or at least associated pulse timings as the pulse timings 301 - 305 of the modulation reference code 300. Thus, there is a-priori knowledge available on what signal pattern to expect in time domain. On the other hand, background or interference photon events are unlikely to have the same pulse timings. Hence, a filter may be applied that is specifically sensitive to the pulse timings 301 - 305 of the modulation reference code 300, to filter out background or interference photon events. An example implementation of such a filter corresponds to performing a time-domain correlation between the measured photon events and a correlation code which is associated with the modulation reference code 300: This correlation detects matching patterns in the measured photon events and the modulation reference code 300. Such matching patterns (identified by a maximum in the correlation) occur at certain times during the measurement phase 205. It can be expected that the best matching pattern - having the highest correlation level, i.e., providing a maximum of the correlation - corresponds to the RTT of the light to and back from the object. Hence, based on the maximum of the correlation, it is possible to determine the distance of the object.

As a general rule, it would be possible that the correlation code is the same as the modulation reference code 300. In other examples, it would be possible to determine the correlation code based on the modulation reference code 300. For instance, it would be possible to consider a physical model of distortion, e.g., to operating characteristics of the laser and/or detector. For example, a time characteristic of the laser and/or detector can be considered that leads to delay or jitter, etc.. This would have the advantage of an even more accurate ranging.

FIG. 6 schematically illustrates aspects with respect to the measured photon events 51 1 - 515. Specifically, FIG. 6 illustrates the time-sequence of measurement signals output by the array detector 102. The detector 102 is characterized by a sample time interval 520.

As a general rule, the detector 102 can be designed so that it comprises sufficient dynamic range to detect photons at all times. Due to the shape of the amplified detector response which has a length between 100 ps to a few nanoseconds, it is sometimes not possible to distinguish two photons with a spacing of less than the response length. Therefore, the length of the amplified response limits the sample time interval time 520. .

The photon count per sample time interval 520 is illustrated in FIG. 6 over the course of time. As illustrated, most of the time, the photon count is zero. However, as illustrated in FIG. 6, there is a total of five photon events 51 1 - 515 having a non-zero photon count. It is generally possible to determine the signal time of the photon events 51 1 - 515, e.g., to provide a timestamp for each photon event 51 1 - 515; and/or the signal amplitude of the photon events 51 1 - 515.

According to various examples, it is possible to determine, for at least some of the photon events 51 1 - 515, a respective signal amplitude - i.e., the photon count - by comparing the respective measurements signal obtained from the array detector 102 with at least three baselines 501 - 503 (dashed lines in FIG. 6) that are offset from each other. For example, the baselines 501 - 503 can be generated using a digital signal generator in combination with a digital-to-analog converter. The amplitude of the baselines 501 - 503 can be set such that they correspond to different signal amplitudes of the measurement signal. Specifically, as illustrated in FIG. 6, the baseline 501 corresponds to a single-photon events: for example, the photon events 512 - 514 cross the baseline 501 , but do not cross the baseline 502; hence, they are identified as single-photon events 512 - 514. Then, it is possible to perform a threshold comparison between each one of baselines 501 - 503 and the measurement signal obtained from the array detector 102. Each time one of the baselines 501 - 503 is crossed, it is possible to conclude on the signal amplitude and the corresponding signal can be fed to a TDC implemented by the compute unit 90. Using one or more TDCs, it is possible to determine the signal time for each one of the photon events 51 1 -515.

As a general rule, it would be possible that there is one TDC per baseline 501 - 503, to facilitate parallel processing. Thus, for at least some of the photon events and for each baseline, a respective signal time 531 (for sake of simplicity, only illustrated in FIG. 6 for the photon event 51 1 ) can be determined by feeding an intersection point of the measurement signal with the respective baseline 501 -503 to the respective TDC.

Typically, the TDCs are associated with a dead time 521. The dead time 521 is larger than the sample time interval 520. For the reason of the dead time 521 , the photon event 513 cannot be post-processed, because the TDC associated with the baseline 501 is still blocked due to processing of the photon event 512.

As will be appreciated, this makes it necessary that the dead time 521 of the TDCs is less than an average background photon rate at the array detector 102 under normal conditions such as ambient daylight light intensity (otherwise the system would be“blinded” practically all the time). For example, ambient daylight light intensity may be defined as follows: a light intensity of 1 kW/m². Using a coaxial optical setup with spatial filtering and a narrow bandpass filter centered at the wavelength of the primary light, the light intensity at the array detector 102 - at an ambient light intensity of 1 kW/m² - can be as low as 15 W/m².

Various techniques described herein are based on the finding that such a low background photon rate at the array detector 102 can normally only be achieved when using a co-linear or coaxial optical setup, e.g., as illustrated in connection with FIG. 1. Specifically, in other scenarios where the receive path is not via the steering mirror 150, it is typically required to pick up light from a comparably large solid angle, to cover for the entire scan range. Then, the average background photon rate at the array detector 102 is exceedingly large. This can prevent an implementation of a baseline 501 that corresponds to a single-photon event. And this in turn can strongly limit the measurement range achievable.

Generally, the number of sample time intervals 520 and photon events 51 1 - 515 over the entire measurement phase duration 206 can be very large. For example, if the measurement phase duration 206 is 20 microseconds long and if the sample time interval 520 is 20 ps long, then there would be 100,000 sample time intervals. Then, performing a correlation between the correlation code and the sequence of photon 51 1 - 515 at a time resolution corresponding to the sample time interval 520 or the dead time 521 can require significant computational resources sample time interval. Hereinafter, techniques will be described which facilitate determining the distance in a less computationally expensive manner and, thereby, quicker. This facilitates real-time processing of the measurement signals obtained from the array detector 102. A respective approach is explained in connection with FIG. 7.

FIG. 7 is a flowchart of a method according to various examples. Optional blocks are marked using dashed lines.

Initially, in optional block 1010A, a detection for an early photon event is performed. Here, the signal output by the detector can be compared directly with a predefined threshold, i.e., each measured photon event 51 1 -515 in isolation. If one of these photon events is larger than the predefined threshold, this can correspond to an early photon event - specifically, if the photon event occurs early, i.e., while the transmit phase is still active and before transmission of the pulse train has been completed. Then, this early photon event can be used for determining the distance to the object. Consequently, in block 1010B it is checked whether an early photon event has been detected. Blocks 101 1 and 1012 are only executed if an early photon event has not been detected (cf. FIG. 3, early photon event 516). Blocks 101 1 and 1012 correspond to a two-step approach for ranging an object, i.e., determining its distance. Blocks 101 1 and 1012 rely on a modulation scheme being applied, e.g., using a CDMA-modulated pulse train. The two-step approach uses a first coarse search performed at a low time resolution to identify one or more candidate distances, see block 101 1. The two-step approach also uses a second fine search performed at the higher resolution at and around the one or more candidate distances, see block 1012.

By such a two-step approach, it becomes possible to restrict the amount of computational operations, because the high temporal resolution associated with the fine search at the higher resolution can be restricted to the one or more candidate distances.

An example implementation of the two-step approach is illustrated in connection with FIG. 8 and FIG. 9.

FIG. 8 schematically illustrates aspects with respect to binning. FIG. 8 illustrates a sequence 570 of bins 571 - 582. Each bin has a binning time duration 590. The binning time duration 590 is much larger than the sample time interval 520 and the dead time 521 , e.g., by at least a factor of 10 or at least a factor of 100.

Example binning time durations 590 are in the range of 1 ns to 5 ns, e.g., around 3 ns.

The practical implementation, it would be possible that the binning time duration 590 corresponds to a clock frequency of the compute unit 90. Then, the various bins 571 - 582 can be populated in real-time, from clock cycle to clock cycle of the compute unit 90. This facilitates a computationally inexpensive binning.

As illustrated in FIG. 8, each photon event 51 1 - 517 is assigned to one of the bins 571 - 582 (except those photon events 51 1 - 517 for which a respective signal time cannot be determined due to the dead time 521 of the TDCs; but they are not illustrated in FIG. 8). When binning, it would be possible to discard information on the signal amplitude of the respective photon events 51 1 - 517. Furthermore, as part of the binning operation, information of the particular fine signal time of each photon event 51 1 - 517 within the respective bin 571 - 582 is discarded - i.e., the time position of each photon event 51 1 -517 within a bin may be neglected when binning.

Optionally, as illustrated in FIG. 8, it is also possible to discard information on the count of photon events per bin 571 - 582, beyond a binary discrimination. Specifically, as illustrated in FIG. 8, each bin 571 - 582 indicates whether there are no photon events within the respective bin or whether they are one or more photon events 51 1 - 517 within the respective bin 571 - 582. In other words, each bin 571 - 582 can be in two states, populated or un-populated (binary bins). For example, while there are in fact two photon events 51 1 , 512 within the bin 571 , still, the bin 571 simply indicates that there are one or more photon events.

Such a binary design of the bins 571 - 582 can be specifically applicable in a scenario in which the minimum inter-pulse distance 310 (cf. FIG. 5) is larger or equal to the binning time duration 590. Namely, in this scenario, it is only possible that one photon event 51 1 -517 per bin 571 - 582 stems from secondary light; excess photon events 51 1 -517 must necessarily originate from background photons.

In other scenarios, a multi-level filling state of each bin 571 - 582 may be considered. For example, the signal amplitude of each photon event within a given bin 571 - 582 may be summed and each bin 581 - 582 may then have a respective filling state corresponding to the sum of signal amplitudes of all photon events within that bin 571 - 582. This avoids false positives.

In any case, it is possible to perform a first correlation between the sequence of bins and the expected signal pattern, i.e., between photon count of each bin 571 - 582 and the correlation code (in FIG. 7, the associated modulation reference code 300 is reproduced as a guide to the eye). Since the bins 571 - 582 are binary, it is sufficient to perform a single-level comparison between the correlation code and the binary filling state of the bins 571 - 573. The single-level comparison can pertain no distinguishing between different amplitudes of the transmitted pulses 400 or the detected photon events.

The first correlation can be intuitively described as moving the correlation code against the binning sequence 510 in time domain and, for each time increment, check for a match between the binary filling state of the bins 571 - 582 and the pulse timing of the correlation code. From this comparison, one or more coarse maximums 591 - 593 result (cf. FIG. 9) that are at well- defined temporal positions within the measurement phase duration 206. The coarse maximums 591 - 593 correspond to the best matches. The coarse maximums 591 - 593, in other words, provide for candidate RTTs. Due to the use of a single-level comparison, there may be ambiguity - it may not be known at high certainty which of the candidate RTTs corresponds to the distance of the object. Also, the candidate RTTs may only be known at the coarse timing corresponding to the binning time duration 590: i.e., the error margin of the coarse maximums 591 - 593 is not smaller than the binning time duration. Next, it is possible to perform for each temporal positions of the coarse maximums 591 - 593, a second correlation between respective sub-sequences of the photon events 51 1 - 515, cropped from the overall sequence of photon events. In other words, it is possible to restrict the high-resolution correlations to the relevant time durations, to resolve the time position of the correlation maximums 591 - 593 on sample time interval 520. Thus, the number of computational steps is greatly reduced, because only a small sub-fraction of all photon events has to be considered.

The process of cropping and of performing the second correlations is explained in greater detail with respect to FIG. 9, specifically for the coarse correlation maximum 591 . This process allows to determine, at lower error margins, the height and the time-position of the coarse correlation maximum 591 , at limited computational resources.

In detail, in the example of FIG. 9, the binning sequence 570 includes a number of four bins 571 -574 which are expected to include the photon events associated with respective laser pulses of the reference code 300 (also illustrated in FIG. 9, as a guide to the eye) resulting in the coarse correlation maximum 591 . It is noted that there are gaps 580 between the bins 571 - 574 where no photon events are expected (since there have been no laser pulses fired). In view of this finding, it is possible to crop (/^') all photon events outside the duration 202 of the reference code 300 (or, more specifically, the correlation code) and to crop (/^'/^') all photon events included in bins within the gaps 580. Thus, in other words and as illustrated in the inset of FIG. 9 (dashed line; magnified view), sub-sequences 701 -703 of photon events are selected from each one of the four bins 571 -574 that are expected to include the photon events of the secondary light associated with the coarse correlation maximum. Then, the correlation can be executed to search for a fine correlation maximum 595. The fine correlation maximum 595 is within the time span defined by the coarse correlation maximum 591 of the first correlation and the binning time duration 590: for instance, in the example of FIG. 9, the fine correlation maximum 595 identified by the second correlation based on the coarse correlation maximum 591 is offset to longer RTTs with respect to the coarse correlation maximum 591 by a time duration 590A. The error margin with which the time position of the fine correlation maximum 595 is known is in the order of the sample time interval 520. This is much less than the error margin with which the time position of the coarse correlation maximum 591 is known, which is in the order of the binning time duration 590.

Ambiguity between the multiple coarse maximums 591 - 593 output by the first correlation - thus, serving as candidates - can be resolved by the one or more second correlations that take into account the signal time of each photon event resolved at sample time interval 520. For example, the particular fine correlation maximum 595 having the largest amplitude may be selected for determining the distance to the object. In some examples, where there are multiple correlation maximums, they may stem from multiple reflections at, e.g., different objects. Then, instead of discarding distance values associated with less prominent coarse or fine correlation maximums, it would be possible to output respective distances forfurther processing by higher- layer applications.

Further, it would be possible to take into account the signal amplitudes of the photon events obtained from the multiple baselines 501 - 503. For example, the signal amplitudes of the photon events corresponding to the secondary light can be compared with expected signal amplitudes due to divergence and absorption along the line of travel of the primary and secondary light. At the other hand, the signal amplitudes of background light will tend to deviate from these expected signal amplitudes which can be considered in the second correlation. Thus, generally, the second correlations can each implement a multi-level comparison.

Further, in some examples, it would be possible to determine the amplitude of of the photon events associated with the fine correlation maximum(s) 595 and derive a reflectivity of the object thereform (taking into account the baseline of the reduced signal level for more remote objects). Alternatively or additionally, it would be possible to determine the amplitude of the fine correlation maximum(s) 595 and derive the reflectivity therefrom.

FIG. 10 schematically illustrates a method according to various examples. For example, the method of FIG. 10 can be implemented by the system 100 according to FIG. 1 . Specifically, it would be possible that the method of FIG. 1 is implemented by the compute unit 90.

Initially, at block 1021 , a pulse train of laser pulses is transmitted. As such, block 1021 corresponds to the transmit phase 201 . The pulse train is modulated in accordance with a modulation reference code.

Next, at block 1022, a sequence of photon events is measured, using an array detector, for a certain measurement phase duration. As such, block 1022 corresponds to the measurement phase 205.

It would then be possible to determine, for at least some of the photon events, signal times and signal amplitudes. This can be achieved using one or more TDCs. For example, multiple baselines can be determined in analog domain and compared against the measurement signal obtained from the array detector. Then, an intersection between the measurement signal and the baselines can be processed by a respective TDC, wherein each baseline is associated with a respective TDC. Thereby, the signal times can be determined and each photon event can be assigned with a time stamp. By discriminating between multiple, offset baselines, it is possible to determine the signal amplitudes. Corresponding techniques are described above in connection with FIG. 5.

Then, the photon events can be binned, based on the respective signal timings, block 1023. Binning can correspond to the act of assigning, to each of the photon event, based on the respective signal time, a corresponding bin of a binning sequence. The binning sequence can thus correspond to a downsampling of the time resolution. For instance, where time stamps are used to track the signal times, a number of most-significant bits of the time stamp can be used to decide on the bin.

Then, at block 1024, a first correlation can be performed on bin level. In other words, it would be possible to perform a first correlation between the photon count (possibly defined binary) of each bin of the sequence of bins and the modulation reference code used to modulate the pulse train at block 1021 .

Then, one or more coarse correlation maximums are identified in block 1024. For one or more of these coarse correlation maximums, a second, high-resolution correlation can be performed, at block 1025. Here, the one or more second correlations are performed on the photon-event level, i.e., based on the full information on the signal timings available from the TDC operation, as explained above. At the same time, only a reduced number of photon events may be considered per second correlation, based on cropping around the respective correlation maximum of the first correlation. The second correlation yields one or more fine correlation maximums.

If block 1025 yields multiple fine correlation maximums, it is possible to filter, at block 1026, the fine correlation maximums. This can be done in order to obtain a reliable result for the distance to the object. For example, sometimes a situation can occur where - e.g., due to non- orthogonality of the used pulse sequence - there are multiple fine correlation maximums per measured distance, i.e., a main fine correlation maximum and one or more side-lobe fine correlation maximums arranged with respect to the main fine correlation maximum. Thus, based on a degree of non-orthogonality - e.g., derivable from the modulation reference code - it would be possible to filter the one or more side-lobe fine correlation maximums. At block 1027, the distance to the object is determined, e.g., based on one or more fine correlation maximums that have not been filtered in block 1026. Optionally, block 1027, it would also be possible to determine a reflectivity of the object.

Although the invention has been shown and described with respect to certain preferred embodiments, equivalents and modifications will occur to others skilled in the art upon the reading and understanding of the specification. The present invention includes all such equivalents and modifications and is limited only by the scope of the appended claims.

For illustration, various techniques have been described which employ an optical system using spatial filtering. Yet, such techniques may be readily applied to other kinds and types of systems, e.g., FLASH systems where multiple directions are illuminated contemporaneously using primary light.

For further illustration, by using a CDMA modulation reference code for obtaining a CDMA- modulated pulse train, it is also possible to transmit multiple pulse trains of laser pulses, during the same time instance or during overlapping transmit time phases in two or more different directions. Then, the spatial resolution can be increased by using one and the same detector to resolve between the respective orthogonal codes.

Claims

1. A method of time-of-flight ranging, comprising:

- transmitting a pulse train of laser pulses, the pulse train being modulated in accordance with a modulation reference code (300),

- using a detector, measuring a sequence of photon events (51 1-515) over a measurement time duration,

- binning at least some of the photon events (511-515) of the sequence of photon events (51 1-515) into a sequence (570) of bins (571-582),

- performing a first correlation between a photon count of each bin of the sequence (570) of bins (571-582) and a correlation code associated with the modulation reference code (300),

- performing one or more second correlations between one or more sub-sequences of photon events (51 1-515) cropped from the sequence of photon events (51 1-515) based on one or more correlation maximums (591-593) of the first correlation and the correlation code, and

- determining at least one distance of an object based on the one or more second correlations.

2. The method of claim 1 , further comprising:

- for at least some photon events (511 -515) of the sequence of photon events (511- 515): determining a respective signal amplitude by comparing a measurement signal obtained from the array detector with at least three baselines (501-503) offset from each other.

3. The method of claim 2,

wherein a first baseline (501 ) of the at least three baselines (501-503) corresponds to a single-photon event (512-514).

4. The method of claim 2 or 3, further comprising:

- for the at least some photon events of the sequence of photon events and for each baseline (501-503): determining a respective signal time (531 ) by feeding an intersection point of the measurement signal with the respective baseline (501-503) to a respective time- to-digital converter.

5. The method of claim 4, wherein a dead time of the time-to-digital converters is less than an average background photon rate at the array detector at ambient daylight light intensity.

6. The method of any one of the preceding claims,

wherein the first correlation operates based on a single-level comparison between the correlation code and a binary filling state of the bins (571-582)).

7. The method of any one of the preceding claims,

wherein a minimum inter-pulse distance (310) of the pulse train is equal to or larger than a binning time duration (590) of each bin.

8. The method of any one of the preceding claims,

wherein a binning time duration (590) of each bin (571-582) corresponds to a clock frequency of a compute unit (90) performing the first correlation and the second correlation.

9. The method of any one of the preceding claims,

wherein the pulse train of laser pulses is emitted along a transmit beam,

wherein the array detector detects light along a receive beam,

wherein the transmit beam and the receive beam are coaxially steered by a steering mirror.

10. The method of any one of the preceding claims, further comprising:

- determining the correlation code based on the modulation reference code (300) based on a model of pulse-time distortion of at least one of a laser source used for transmitting the pulse train of laser pulses and the detector.

1 1. The method of any one of the preceding claims,

wherein said cropping comprises, for each one of the one or more correlation maximums (591-593) of the first correlation: selecting the one or more sub-sequences (701 - 703) of the photon events (51 1 -515) from bins of the sequence (570) of bins (571 -582) expected to include photon events (511-515) of reflected laser pulses having a round-trip- time associated with the respective correlation maximum (591-593) of the first correlation.

12. The method of any one of the preceding claims,

wherein each one of the one or more second correlations yields a respective further correlation maximum (595),

wherein an error margin of the further correlation maximums (595) of the one or more second correlations is on the order of a sample time interval (520) of the detector, wherein an error margin of the one or more correlation maximums (591-593) of the first correlation is on the order of a binning time duration (590) of each bin.

13. The method of any one of the preceding claims, further comprising:

- while transmitting the pulse train of laser pulses: detecting a photon event having an amplitude that exceeds a predefined threshold, and

- selectively aborting said transmitting of the pulse train of laser pulses in response to detecting the photon event having the amplitude that exceeds the predefined threshold.

14. The method of any one of the preceding claims:

wherein a reflectivity of the object is determined based on the second correlation.