WO2023118951A1

WO2023118951A1 - Quality estimation of an amcw lidar range estimation

Info

Publication number: WO2023118951A1
Application number: PCT/IB2021/062332
Authority: WO
Inventors: Pedro Nelson SAMPAIO BARBOSA; Pranjali PARASHRAM SHINDE; Miguel Vidal Drummond; Daniel António MACEDO BASTOS; Abel LORENCES RIESGO; Paulo Miguel NEPOMUCENO PEREIRA MONTEIRO
Original assignee: Bosch Car Multimedia Portugal, S.A.
Priority date: 2021-12-22
Filing date: 2021-12-27
Publication date: 2023-06-29

Abstract

The present invention describes a method for estimating the quality of a range estimation obtained by an amplitude modulated continuous wave (AMCW) light detection and ranging (LIDAR). The proposed method for estimating the quality of an AMCW LIDAR range estimation comprises the steps of obtaining a first range estimation from a first frequency bin; obtaining a second range estimation from a second frequency bin; comparing the first range estimation of the first frequency bin with the second range estimation from the second frequency bin; producing the quality estimation based on the compared value.

Description

Quality estimation of an AMCW LIDAR range estimation

Technical Field

The present invention describes a method for estimating the quality of a range estimation obtained by an amplitude modulated continuous wave (AMCW) light detection and ranging (LIDAR).

Background art

Interference is a classic problem of wireless communications, in which various transponders in the same scene share the same medium, and thus each must be able to operate reliably while having interference from others.

The main solution for avoiding general interference is to guarantee that the signals of one LIDAR are orthogonal or at least partially uncorrelated to others, in at least one dimension: frequency, time, polarization, or space. Multiple interference avoidance methods and inventions that build on such a guideline have been proposed.

US patent application US20170082737A1 describes a universal clock signal shared amongst all LIDAR systems that guarantees that only one LIDAR "fires" at a time, thus providing time orthogonality, i.e., orthogonal time slots.

US patent US9575164B2 discloses a LIDAR system based on a pair of emitting units, wherein a first unit transmits an optical beam that is orthogonally polarized to the optical beam of the second unit. A drawback of such a configuration is that there are only two orthogonal polarization states, which is likely insufficient if many LIDARs are considered. The article T. Fersch, R. Weigel and A. Koelpin, "A CDMA Modulation Technique for Automotive Time-of-Flight LiDAR Systems," in IEEE Sensors Journal, vol. 17, no. 11, pp. 3507- 3516, June 1, 2017, with doi: 10.1109/JSEN.2017.2688126 proposes the use of optical code division multiple access (OCDMA) for LIDAR. As a result, beams from other LIDAR systems have orthogonal codes, which minimizes interference. However, it should be noted that considering an increasing number of LIDAR systems sharing the same scene requires longer codes, which in turn results in higher latency. OCDMA has likewise been proposed in US patent application US20170329010A1, mentioning nonetheless that different beams from the same LIDAR system can be modulated with different (orthogonal) codes in order to minimize interference.

US patent US9989629B1 proposes a LIDAR system that changes its wavelength over time. Consequently, the probability of another LIDAR using the same wavelength is minimized.

Finally, US patent application US20170090031A1 proposes a LIDAR transmitter that emits a beam at a given time slot, with a given wavelength, towards a given direction. As a result, the probability of an interfering beam arriving at the same time slot, with the same wavelength and from the same direction is significantly minimized.

The previously mentioned inventions are either transparent to signalling (i.e., work for pulsed or AMCW signals), or are applicable only to pulsed LIDARs. As expected, the bulk of LIDAR interference avoidance methods belongs to such a category, and very few inventions are applicable or directed to AMCW LIDARs. Known state-of-the-art with regard to interference avoidance in AMCW LIDARs is mentioned in US patent application US20170329011A1 which proposes intensitymodulating different beams with signals confined to different frequency bands. For instance, modulating a beam with a unique AMCW tone is a valid particular case. As one frequency band is assigned to one beam alone, beams are orthogonal between themselves. As in US patent application US20170329010A1, the goal is to minimize interference between beams emitted by the same LIDAR system through orthogonal frequencies.

The same US patent application US20170329011A1 also proposes an AMCW signal with several tones multiplexed in time (one tone at a time), such that the LIDAR "hops" in frequency over time. Frequency hopping is a method originally devised for wireless communications for improving robustness against interference. If an AMCW LIDAR operates always at the same RF frequency, and if it happens that another AMCW LIDAR is also operating at the same (or a very close) RF frequency, this will cause permanent disruption. If, on the contrary, LIDARs use frequency hopping, the probability of two LIDARs using the same frequency at the same time is much lower, and so is the probability of permanent or even temporary disruption .

US patent US7405812B1 basically proposes that an AMCW LIDAR should have a reference clock signal with bad quality on purpose, such that the waveforms of two AMCW LIDARs, even when using the same tone frequency, become poorly correlated over time. Such a poor cross-correlation is beneficial, as it causes interference to be mitigated. The inventions above teach how to avoid interference.

However, there is also the opposite paradigm: not to avoid interference, but to accept it; this, however, requires that interference is detected. From this particular point of view, the problem to be solved by the present invention is how to detect AMCW interference.

Summary

The present invention describes a method for estimating the quality of an AMCW LIDAR range estimation comprising the steps of obtaining a first range estimation from a first frequency bin; obtaining a second range estimation from a second frequency bin; comparing the first range estimation of the first frequency bin with the second range estimation from the second frequency bin; producing the quality estimation based on the compared value.

In a proposed embodiment of present invention, the first frequency bin is not used to obtain the second range estimation.

Yet in another proposed embodiment of present invention, comparing both range estimations comprises a difference between the first range estimation and the second range estimation.

Yet in another proposed embodiment of present invention, the quality estimation comprises removing out-of-band interference from the first frequency bin and from the second frequency bin through filtering. Yet in another proposed embodiment of present invention, the estimation based on the compared value is defined by a predetermined threshold value, which defines the quality of the estimation when above said value.

The present invention further describes a computer program configured to carry out every step of the described method.

The present invention further describes a (non-transitory) machine-readable storage device, on which the computer program configured to carry every step of the described method is stored.

The present invention further describes a data processing system, comprising the necessary physical means for the execution of the computer program configured to carry every step of the described method.

The present invention further describes an electronic control unit, configured to carry out every step of the method herein disclosed.

General Description

Since the quality of a LIDAR range estimation depends on noise and on interference, and since existing interference detectors are unable to distinguish interference from noise, the present invention aims to provide an accurate AMCW interference detection method for significant amounts of noise.

Therefore, the generalized problem to be solved is how to estimate the quality of an AMCW range estimation. If such an estimation is affected by interference and/or significant amounts of noise, the estimated quality should be poor, allowing the likely very erroneous range estimation to be discarded .

The main and obvious advantage of the new disclosed method for estimating the quality of an AMCW range estimation is that only high-quality estimations can be picked, whereas others are discarded. If the estimation is impaired by interference and/or significant amounts of noise, the estimated quality should be poor such that the range estimation can be discarded.

Another important advantage is that the disclosed method is complementary to interference avoidance methods, further increasing the robustness of an AMCW LIDAR to interference.

The main limitation of the present invention is that it applies only to non-continuous ("bursty") AMCW signals. Such a limitation is not relevant as a LIDAR always has a limited amount of time to emit and process a beam.

The concept of the proposed method aims to obtain two estimations of the Time of Flight (ToF), τ_ToF and T_ToF,alt, via two different methods. If the received signal is not corrupted by (in-band) interference nor noise, both ToF estimations, τ_ToF and T_ToF,alt, should be identical. This means that the estimated quality is high. Therefore, the estimated quality is inversely proportional to the difference between both of channel estimations.

Brief description of the drawings

For better understanding of the present application, figures representing preferred embodiments are herein attached which, however, are not intended to limit the technique disclosed herein.

Figure 1 - illustrates a non-continuous multi-tone AMCW signal, emitted by a LiDAR for example, wherein the references are related to:

A - time in ps;

B - burst amplitude.

Figure 2 - illustrates the power spectral density (PSD) of the received non-continuous multi-tone AMCW signal, received by a LiDAR for example, wherein the references are related to:

C - frequency in MHz;

D - PSD in dB.

Figure 3 - illustrates the comparison between the original method for estimating the ToF (first ToF estimation),τ_ToF , with the alternative method (second ToF estimation), T_ToF,alt, without interference. The references are related to:

E - Channel delay in ps;

F - Estimatedτ_ToF in ps;

100 - original methoτd_ToF ;

200 - alternative method T_ToF,alt . Figure 4 - illustrates the comparison between the original method for estimating the ToF (first ToF estimation),τ_ToF , with the alternative method (second ToF estimation), T_ToF,alt, with interference. The references are related to:

E - Channel delay in ps;

F - Estimatτed τ_ToF in ps;

100 - original methodτ_ToF ;

200 - alternative method T_ToF,alt .

Figure 5 - illustrates the data dispersion of the comparison between both estimations |τ_ToF — T_ToF,alt| for multiple realizations, each performed for a given distance, in the particular of the illustration, only detection and amplification noise is considered. The references are related to:

G - distance in m;

H - |τ_ToF — T_ToF,alt| in ns•

Figure 6 - illustrates the data dispersion of the comparison between both estimations |τ_ToF — T_ToF,alt| for multiple realizations, each performed for a given distance, in the particular of the illustration, besides detection and amplification noise there is also AMCW interference present. The references are related to: G - distance in m;

H - |τ_ToF — T_ToF,alt| in ns• Description of Embodiments

With reference to the figures, some embodiments are now described in more detail, which are however not intended to limit the scope of the present application.

The first ToF estimatiτo_n,τ_ToF , can be obtained from any known or state of the art method that processes the tones of received an AMCW signal.

Figure 1 illustrates an example of an emitted non-continuous multi-tone AMCW signal with 4 ps long, based on a 20 MHz square wave, and phase modulated at 2 Mbaud with the sequence 0, π, 0, π, 0, π, 0, π. The purpose of such a modulation is to generate AMCW tones at 19 MHz and 21 MHz. The receiver is open during 6 ps, allowing for a maximum delay of 2 ps, which corresponds to a maximum range of 300 m.

Therefore, and through the analysis of Figure 2, which illustrates the power spectral density (PSD) of the received non-continuous multi-tone AMCW signal received by the LiDAR, such proposed method uses the identified frequency bins at 19 MHz and 21 MHz. For the proposed emitted non-continuous multi-tone AMCW signal, the Discrete Fourier Transform (DFT) of the burst signal acquired by the LiDAR will always comprise peak maximum values located in the 19 MHz and 21 MHz range, and in 1/TRx, where TRx represents the waveform duration sampled by the receptor (6 ps as previously mentioned) . Therefore, it is possible to obtain three coarse ToF estimations, which are located in the peak of spectrogram calculated for 19 MHz, in the peak of spectrogram calculated for 21 MHz, and from the phase of the first bin frequency (not being DC), which has a frequency of 1/TRx. In a general approach, if there is no interference or excessive noise, i.e., if the received waveform has quality, the three estimations should match.

To understand the second (or alternative) ToF estimation, T_ToF,alt further analysis of Figure 2 is required, and particularly, the spectrum illustrated therein. A range estimation in an AMCW LIDAR should focus on the frequency bins with the highest power. Excluding DC components, the frequency bins that correspond to generated AMCW tones on the emitted signal, i.e., of 19 and 21 MHz, have the highest power, nearly 40dB. Further a first frequency bin right after DC also has a very high power (over 40dB), and thus should be taken into account as well. In said figure, such a frequency bin has a frequency of = 166.67kHz. Given that

such a frequency bin has an inherently low frequency, an unambiguous but coarse AMCW estimation is obtained from it, herein referred to as T_ToF,alt. As the first ToF estimation, τ_ToF, cannot be any coarser than T_ToF,alt, both values,τ_ToF and T_ToF,alt are suitable for being compared.

Figures 3 and 4 illustrate the comparison between the results of the original method for estimating the ToF (first ToF estimation),τ_ToF , and the alternative method (second ToF estimation), T_ToF,alt, the received signal without and with interference respectively. No noise is present in any of the disclosed results, and it is easily noticeable that interference in the signal create significant visual differences between ToF estimations.

Figures 5 and 6 illustrate the comparison between the estimations |τ_ToF — T_ToF,alt|, that despite considering the detection and amplification noise in both cases, the effect of the presence of interference in the AMCW signal is particularly evident. Through this comparison it is possible to confirm that the metric |τ_ToF — T_ToF,alt|tends to increase with noise, through an increasing dispersion of values, which significantly increases along with the inclusion of AMCW interference .

The proposed method for estimating the quality of an AMCW estimation may thus be summarized, in of the possible embodiments, through the following steps:

1. Obtain a first ToF estimation from at least one frequency bin, τ_ToF;

2. Obtain a second ToF estimation from at least one frequency bin, T_ToF,alt, wherein the frequency bins used for obtaining the first ToF estimation,τ_ToF , are not used for obtaining the second ToF estimation;

3. Compare both estimations, |τ_ToF — T_ToF,alt|, and from such a comparison obtain a quality estimation;

4. (Optional) From such a quality estimation one may decide whether to classify the obtained estimation (e.g., bad, ok, good), or even to discard it.

The herein disclosed method for estimating the quality of an AMCW estimation inherently includes removing out-of-band interference, as only selected frequency bins are used, which in turn implies filtering.

However, if out-of-band interference is bursty, the envelope of the interfering burst affects the first frequency bin right after DC. Hence, if the second ToF estimation is based on such a frequency bin, the resulting quality estimation becomes (slightly) conservative. Nevertheless, it is typically better to have a conservative quality estimation than an optimistic one.

The metric obtained from comparing both estimations can be made more or less conservative, and can be set to a predetermined threshold value, considering therefore that if the estimation is below said threshold value, then the estimation does have quality. If this premise is not achieved, the quality may be compromised. In generic terms, it is possible to conclude that the determined quality = constant x |τ_ToF — T_ToF,alt|

For instance, the highest conservative degree is obtained when good quality is assigned only if |τ_ToF — T_ToF,alt|= 0.

The proposed quality estimator cannot distinguish noise from interference as, just like interference, noise affects one frequency bin independently from another (i.e., the impact of noise on one frequency bin has no correlation with the impact of noise on another frequency bin).

The proposed method can be generalized to any given number of ToF estimations, such that all are compared to obtain a quality estimation.

The proposed method applies to any kind of AMCW signal, with any given number of periods (except for an infinite number of periods), and with any given number of tones. This means that the proposed method also applies to a short pulse (which is a degenerate case of an AMCW signal with 1 tone and 1 period).

Claims

1. Method for estimating the quality of an AMCW LIDAR range estimation comprising the steps of obtaining a first range estimation from a first frequency bin; obtaining a second range estimation from a second frequency bin; comparing the first range estimation of the first frequency bin with the second range estimation from the second frequency bin; producing the quality estimation based on the compared value.

2. Method for estimating the quality of an AMCW LIDAR range estimation, according to the previous claim wherein the first frequency bin is not used to obtain the second range estimation.

3. Method for estimating the quality of an AMCW LIDAR range estimation, according to any of the previous claims wherein comparing both range estimations comprises a difference between the first range estimation and the second range estimation.

4. Method for estimating the quality of an AMCW LIDAR range estimation, according to any of the previous claims wherein the quality estimation comprises removing out-of-band interference from the first frequency bin and from the second frequency bin through filtering.

5. Method for estimating the quality of an AMCW LIDAR range estimation, according to any of the previous claims wherein the quality estimation based on the compared value is defined by a predefined threshold value, which defines the quality of the estimation when above said value.

6. Computer program configured to carry out every step of one of the methods described in claims 1 to 5.

7. (Non-transitory) Machine-readable storage device in which the computer program of claim 6 is stored.

8. Data processing system comprising the necessary physical means for the execution of the computer program of claim 6.

9. Electronic control unit, configured to carry out every step of one of the methods of claims 1 to 5.