Optical sensing system

Abstract

The invention relates to an optical sensing system (1) for optical sensing. The system (1) comprises at least one optical sensor (30), said optical sensor (30) comprising a plurality of sensing units (2 '), preferably in a matrix configuration, where each said sensing unit (2') is preferably a pixel sensor (2 '), where each sensing unit comprises a photodetector, where each photodetector is adapted to output a signal upon detection of photons. The system (1) further comprises optics (3) capable of generating an image of the scene (4) on the optical sensor (30). The system (2) further comprises a plurality of processing devices (5 '), wherein each said processing device (5') is connected to at least one sensing unit (2 ') corresponding thereto. Each of said processing means (5 ') is adapted to receive at least one input corresponding to at least one output (7') of a corresponding sensing unit (2 '). Each of the processing means (5 ') is adapted to integrate an output (7', 7 ", 7" ') of a corresponding sensing unit (2', 2 ", 2" ') to obtain a first integrated output (10), where the processing means are adapted to generate a first output signal (8) when the first integrated output (10) reaches at least a first predetermined value (9) within a first predetermined time span ([Delta] t).

Description

Technical Field

[0001] The present invention relates to optical sensing systems. In particular, the present invention relates to optical sensing systems for compact optical sensing. Background Art

[0002] Optical sensing systems are used in a variety of applications. Many of these systems are synchronous systems. Such systems can only detect light signals at certain points in time, which means that any signals arriving outside of these points in time will not be detected. The solution to this problem is to check the light signal frequently, which is energy intensive. In addition, many of these systems are large and complex.

[0003] Therefore, there is a need for an optical sensing system that can detect light signals whenever they arrive, such as asynchronous detection, without consuming a lot of energy, while being compact and simple.

[0004] The present invention is intended to partially solve the above-mentioned problems. Summary of the invention

[0005] It is an object of embodiments of the present invention to provide a compact and reliable optical sensing system.The above object is achieved by a system and method according to the present invention.

[0006] In a first aspect, the present invention relates to an optical sensing system for optical sensing, comprising:

[0007] - at least one optical sensor, said optical sensor comprising a plurality of sensing cells, said plurality of sensing cells preferably being arranged in a matrix, wherein each of said sensing cells is preferably a pixel sensor, wherein each sensing cell comprises a photodetector, wherein each photodetector is adapted to output a signal upon detection of a photon,

[0008] - an optical device capable of producing an image of the scene on said optical sensor,

[0009] - a plurality of processing devices, wherein each of said processing devices is connected to at least one sensing unit corresponding thereto,

[0010] wherein each of the processing means is adapted to receive at least one input corresponding to at least one output of a corresponding sensing unit,

[0011] Each of the processing devices is adapted to integrate the output of the corresponding sensing unit to obtain a first integrated output, wherein the processing device is adapted to generate a first output signal when the first integrated output reaches at least a first predetermined value within a first predetermined time span.

[0012] For example, the processing device is adapted to generate the output signal when the total number of photons detected by the sensing unit reaches at least a predetermined number within the first predetermined time span.

[0013] An advantage of embodiments of the present invention is that false detections are filtered. An advantage of embodiments of the present invention is that noise such as ambient light is reduced or eliminated. An advantage of embodiments of the present invention is that reliable detection is obtained.

[0014] An advantage of embodiments of the present invention is that asynchronous detection is achieved.

[0015] An advantage of embodiments of the present invention is that a low power consumption and simple system is obtained.

[0016] An advantage of embodiments of the present invention is that processing of sensed information is accomplished by a processing structure having massively parallel input connections to each pixel sensor, such as implemented using a neural network, rather than using a traditional processor with serial data and instructions.

[0017] Preferred embodiments of the first aspect of the present invention include one or a suitable combination of more than one of the following features.

[0018] The system preferably comprises at least one adjacent sensing unit adjacent to the sensing unit, wherein the system comprises at least one adjacent processing unit adjacent to the processing unit, wherein each sensing unit is connected to the at least one adjacent processing unit corresponding to the at least one adjacent sensing unit. Preferably, each of the processing units is adapted to integrate the outputs of the corresponding sensing unit and the at least one adjacent sensing unit to obtain the first integrated output. An advantage of embodiments of the present invention is that processing is performed in a parallel manner, thereby obtaining fast and efficient processing. An advantage of embodiments of the present invention is that a compact system is obtained due to the parallel connection.

[0019] An advantage of embodiments of the present invention is that pulses corresponding to different neighboring optical sensors are taken into account. An advantage of embodiments of the present invention is that a more reliable detection is obtained due to the detection confirmation of said neighboring optical sensors, thus filtering out false detections and ambient light.

[0020] The first integral output preferably has a decay rate, wherein the decay rate is preferably adjustable. An advantage of embodiments of the present invention is that the decay rate determines a first predetermined time span within which a certain total number of pulses need to be received in order to generate the output signal. An advantage of embodiments of the present invention is that adjusting the decay rate allows controlling and reducing power consumption in the system.

[0021] Each treatment means preferably comprises (or consists of) a first treatment means connected in series to a second treatment means, wherein the decay rate of the first treatment means is faster than the decay rate of the second treatment means.

[0022] The second processing means is preferably adapted to integrate the first output signal of the first processing means to obtain a second integrated output, wherein the second processing means is adapted to generate the second output signal when the second integrated output is at least a second predetermined value.

[0023] An advantage of embodiments of the present invention is that the first processing means and the second processing means are two different filter layers with two different filtering requirements.

[0024] Each processing means is preferably adapted to at least partially reset after generating its output signal. An advantage of embodiments of the invention is that the processing means can be used to receive and process long trains of pulses. An advantage of embodiments of the invention is that asynchronous detection is obtained because the processing means is reset after generating said output signal so that it is ready to receive a new set of pulses.

[0025] The system preferably comprises a switch element for each processing means, wherein the switch element is adapted to allow bypassing of said processing means.An advantage of embodiments of the invention is that a selection can be made between using the first processing means or the second processing means depending on detected conditions such as ambient light conditions.

[0026] Each processing means is preferably a neuron, preferably a leaky integrate and fire (LIF) neuron. An advantage of embodiments of the present invention is that a compact system is obtained. An advantage of embodiments of the present invention is that a simple and uncomplicated system is obtained. An advantage of embodiments of the present invention is that the decay rate of the neuron can be adjusted.

[0027] The system preferably further comprises at least one light source and scanning means, wherein the scanning means is adapted to scan a light beam from the light source along a trajectory over the scene, preferably continuously. An advantage of embodiments of the invention is that the light source allows triangulation and determination of a depth profile of the field of view between the optical sensor and an object in the scene, wherein the light source and the optical sensor are asynchronous, i.e. the generation and detection of light are not synchronized.

[0028] The system preferably comprises at least two optical sensors.An advantage of embodiments of the invention is that triangulation is possible.

[0029] In a second aspect, the invention relates to a method for optical sensing, comprising the following steps:

[0030] a- receiving a plurality of optical signals,

[0031] b- for each optical signal, integrating the signal in parallel to obtain a first integrated output,

[0032] c- When the first integrated output is at least a first predetermined value within a first predetermined time span, a first output signal is generated.

[0033] Preferred embodiments of the second aspect of the present invention include one or a suitable combination of more than one of the following features:

[0034] - the method preferably further comprises the step of decaying the integrated output over time, wherein the rate of decay is preferably adjustable,

[0035] - The method further comprises the following steps:

[0036] d- for each first output signal, integrating the signal in parallel to obtain a second integrated output, and

[0037] e- generating a second output signal when the second integrated output is at least a second predetermined value,

[0038] The decay rate in steps a to c is faster than that in steps d to e,

[0039] - resetting the first integrated output after generating the first output signal, and resetting the second integrated output after generating the second output signal.

[0040] The above and other characteristics, features and advantages of the present invention will become apparent from the following detailed description taken in conjunction with the accompanying drawings, which illustrate the principles of the present invention by way of example. This description is given for illustrative purposes only and does not limit the scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present disclosure will be further illustrated by the following description and accompanying drawings.

[0042] Figure 1 An optical sensing system (1) is shown according to an embodiment of the present invention, comprising a plurality of sensing units (2', 2", 2'") and a plurality of processing devices (5', 5", 5'", 5'"), wherein each sensing unit (2', 2", 2'") is connected to a corresponding processing device (5', 5"), and wherein a first processing device (5', 5") and a second processing device (5', 5") are connected in series.

[0043] Figure 2An optical sensing system (1) according to an embodiment of the present invention is shown, comprising a plurality of sensing units (2', 2", 2'") and a plurality of processing devices (5', 5"), wherein each sensing unit (2', 2", 2'") is connected to a corresponding processing device (5', 5"), wherein the processing device (5', 5") comprises a first neuron (12) and a second neuron (13) connected in series.

[0044] Figure 3 A processing device (5) according to an embodiment of the invention is shown, wherein the processing device (5) is adapted to generate a pulse (21) when the total number of input pulses (7) to the processing device (5) within a predetermined time span (Δt) is at least a predetermined number (11).

[0045] Figure 4 and Figure 5 A processing device (5) including a first neuron (12) and a second neuron (13) according to an embodiment of the present invention is shown, wherein an example output (7") of the sensing unit (2") is shown in (a), a corresponding integrated output (10) through the first neuron (12) is shown in (b), a corresponding output (17) of the first neuron (12) is shown in (c), a corresponding integrated output (23) of the first neuron (12) through the second neuron (13) is shown in (d), and a corresponding output (18) of the second neuron (13) is shown in (e).

[0046] Figure 6 A processing device (5) according to an embodiment of the present invention is shown, wherein in (a) a first neuron (12) and a second neuron (13) are connected in series, and wherein in (b) a first switching device (22') and a second switching device (22") are located before each of the first neuron (12) and the second neuron (13).

[0047] Figure 7 An example of synchronous detection according to the prior art is shown, wherein an input of a train of pulses (24) is shown in (a), an observation window (25) is shown in (b), and an output (26) is shown in (c).

[0048] Figure 8 Detection under strong ambient light conditions (a) and weak ambient light conditions (b) according to an embodiment of the present invention is shown.

[0049] Fig. 9 An example of asynchronous detection according to an embodiment of the present invention is shown.

[0050] Any reference signs in the claims should not be construed as limiting the scope.The same reference signs in different drawings refer to the same or similar elements. DETAILED DESCRIPTION

[0051] The present invention relates to an optical sensing system for optical sensing.

[0052] The present invention will be described with respect to specific embodiments and with reference to certain drawings, but the invention is not limited thereto but only by the claims. The drawings described are only schematic and not restrictive. In the drawings, for the purpose of illustration, the size of some elements may be exaggerated and not drawn to scale. The sizes and relative sizes do not correspond to actual reductions in the practice of the invention.

[0053] The terms first, second, etc. in the specification and claims are used to distinguish between similar elements and are not necessarily used to describe a sequence in time, space, order, or any other manner. It should be understood that the terms so used are interchangeable under appropriate circumstances, and that the embodiments of the invention described herein are capable of operating in other sequences than described or shown herein.

[0054] Furthermore, the terms top, bottom, etc. in the specification and claims are used for descriptive purposes and not necessarily for describing relative positions. It is to be understood that the terms so used are interchangeable under appropriate circumstances, and that the embodiments of the invention described herein are capable of operation in other orientations than described or illustrated herein.

[0055] In the description provided herein, many specific details are set forth. However, it should be understood that embodiments of the present invention may be practiced without these specific details. In other cases, well-known methods, structures, and techniques are not shown in detail in order not to obscure the understanding of this description.

[0056] References throughout this specification to "one embodiment" or "an embodiment" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment is included in at least one embodiment of the present invention. Thus, the phrases "in one embodiment" or "in an embodiment" appearing in various places throughout this specification do not necessarily all refer to the same embodiment, but may also refer to the same embodiment. Furthermore, in one or more embodiments, the particular features, structures, or characteristics may be combined in any suitable manner as would be understood by one of ordinary skill in the art from this disclosure.

[0057] Similarly, it should be appreciated that in the description of exemplary embodiments of the present invention, various features of the present invention are sometimes grouped together in a single embodiment, figure, or description thereof in order to streamline the present disclosure and aid in understanding one or more of the various innovative aspects. However, this method of disclosure should not be interpreted as reflecting an intention that the claimed invention requires more features than those expressly recited in each claim. On the contrary, as reflected in the following claims, innovative aspects lie in less than all the features of a single preceding disclosed embodiment. Therefore, the claims following the Detailed Description section are hereby expressly incorporated into this Detailed Description section, with each claim existing independently as a separate embodiment of the present invention.

[0058] In addition, although some embodiments described herein include some features included in other embodiments, and do not include other features included in other embodiments, the combination of features of different embodiments should be within the scope of the present invention and form different embodiments, as understood by those skilled in the art. For example, in the following claims, any of the claimed embodiments can be used in any combination.

[0059] Unless otherwise defined, all terms, including technical and scientific terms, used to disclose the present invention have the meanings commonly understood by ordinary technicians in the field to which the present invention belongs. By further guidance, term definitions are included to better understand the teachings of the present invention.

[0060] As used herein, the following terms have the following meanings: "a", "an", and "the" as used herein refer to both singular and plural referents unless the context clearly indicates otherwise. For example, "a pollutant" refers to one or more than one pollutant.

[0061] The recitation of numerical ranges by endpoints includes all numbers and fractions subsumed within that range, as well as the recited endpoints.

[0062] In a first aspect, the present invention relates to an optical sensing system for optical sensing, preferably for compact optical sensing. The system comprises at least one optical sensor, the optical sensor comprising a plurality of sensing units, each of the sensing units preferably being a pixel sensor. Each sensing unit comprises a photodetector. Preferably, the photodetector is a single photon detector, preferably a single photon avalanche detector. Alternatively, the photodetector is an avalanche photodetector. Each of the photodetectors is adapted to output a signal upon detection of a photon. For example, upon detection of one photon or a predetermined number of photons, the photodetector outputs a signal (e.g., a pulse). The system further comprises an optical device capable of producing an image of a scene on the optical sensor. The system further comprises a plurality of processing devices, each of which is connected to at least one sensing unit corresponding thereto. For example, each sensing unit has a processing device corresponding thereto, wherein the sensing units are connected in parallel to the processing device corresponding thereto. Such a parallel connection is advantageous for several reasons, for example allowing a fast, efficient and compact system.

[0063] Each of the processing devices is adapted to receive at least one input corresponding to at least one output of a corresponding sensing unit. For example, the processing device is connected to the sensing unit and receives input from the sensing unit. The output of a sensing unit is, for example, a pulse or a pulse train corresponding to whether the sensing unit receives a photon.

[0064] For example, the output signal is generated, for example based on the speed at which pulses (e.g. corresponding to the photons) arrive at the processing means, i.e. whether they arrive within a first predetermined time span. For example, the processing means is adapted to generate the output signal when the total number of photons detected by the sensing unit is at least a predetermined number within the first predetermined time span. For example, when the number of pulses is at least 2 pulses, preferably at least 5 pulses, more preferably at least 10 pulses within at most 10 nanoseconds, preferably at most 5 nanoseconds, more preferably at most 2 nanoseconds. In other words, if many pulses arrive within a short period of time, it is likely that the detection is positive, while if only a few pulses arrive within the short period of time, it is likely that the detection is false. This is conducive to obtaining reliable detection and filtering out false detections. For example, in the presence of ambient light, the probability of receiving 5 pulses within 2 nanoseconds is low. This also allows asynchronous detection, because the processing means receives pulses and generates an output signal at any time, as long as a certain number of photons are detected within the first predetermined time span.

[0065] Each processing device is suitable for integrating the output of the corresponding sensing unit to obtain a first integrated output, wherein the processing device is suitable for generating a first output signal when the first integrated output is at least (or reaches) a first predetermined value. This means that the first integrated output changes dynamically based on the output of the corresponding sensing unit until it reaches the first predetermined value. As described above, the input signal and the output signal to and from the processing device are digital signals, but the signal within the processing device may include an analog signal. For example, in this case, the first integrated output signal is an analog signal, but is only used within the processing device and is not output by the processing device. In practice, integrating the output from one sensing unit will not allow the integrated output to reach the first predetermined value within a first predetermined time span, because the first predetermined time span is usually very short. Thus, in practice, the outputs of more than one sensing unit (e.g., adjacent sensing units) are connected to each processing device to allow the first integrated output to reach the first predetermined value, as shown below. Ideally, the first integrated output decays over time, or is reset at certain time points (i.e., it is only allowed to reach the first predetermined value when it is within the first predetermined time span), as shown below. Having said that, the detection of the sensing unit is not limited to single photon detection. Furthermore, those skilled in the art understand that the number of photons can be calculated even when using a sensing unit comprising a typical photodetector that is not a single photon detector. For example, the total number of photons can also be the total number of detections, such as when a certain number of detections of at least the predetermined number are detected within the first predetermined time span.

[0066] Preferably, each sensing unit is adapted to generate a stream of pulses or digital signals (e.g. signals within at least 2 values ​​(e.g. '1' and '0')), e.g. whereby each pulse is the result of detecting an impinging photon. The processing device is also preferably adapted to receive the stream of pulses or digital signals. For example, input signals to the processing device and output signals from the processing device are digital signals. This advantageously eliminates the need for an analog to digital converter. However, processing the signal within the processing device may involve analog signals, such as where the pulses are integrated in the processing device (as described below).

[0067] Preferably, each photodetector is arranged in a reverse bias configuration. Preferably, the detector is capable of detecting a single photon impinging thereon. Preferably, the detector is adapted to output a logic signal, such as an electrical detection signal, upon detection of a photon. For example, a detected signal may be represented by a signal comprising a logic '1' (e.g., detected), while a non-detected signal may be represented by a signal comprising a logic '0' (e.g., non-detected). Alternatively, a detected signal may be represented by or result in a pulse signal, such as a transition from a logic '0' to a logic '1' and then a transition from a logic '1' back to a logic '0', while a non-detected signal may be represented by (or result in the absence of) such a pulse signal.

[0068] Preferably, the system comprises at least one adjacent sensing unit adjacent to the sensing unit, wherein the system comprises at least one adjacent processing device adjacent to the processing device. Each sensing unit is connected to the at least one adjacent processing device corresponding to the at least one adjacent sensing unit. In other words, each processing device receives outputs from the sensing unit corresponding thereto and at least one, or preferably at least two or three adjacent sensing units. Thereafter, the detection is based on more than one sensing unit, for example, a group of adjacent sensing units. This is conducive to making the detection less likely to be erroneous, i.e., originating from a temporally and spatially uncorrelated photon stream, such as ambient light, but a true detection, i.e., originating from light actively projected onto the scene that generates a temporally and temporally correlated photon stream in the vicinity of the pixel. For example, if three adjacent sensing units each detect a photon within 2 nanoseconds, such a detection may be a true detection.

[0069] Preferably, each of the processing means is adapted to integrate the output of the at least one sensing unit and the at least one adjacent sensing unit to obtain the first integrated output, wherein the processing means is adapted to generate the first output signal, preferably a digital output signal, when the combined integrated output is at least the first predetermined value within the first predetermined time span. Having pulses from multiple sensing units (e.g. adjacent or closely located sensing units) within a short period of time is unlikely to be a false detection.

[0070] Preferably, each first integral output has a decay rate, wherein each said decay rate is preferably adjustable. In other words, in the presence of a plurality of first and second processing means, each said processing means can be adjusted independently. The decay rate determines the first predetermined time span and how fast the pulses must arrive in succession in order to generate the first output signal. For example, the first integral output will remain decaying until enough pulses are received so that the first integral output eventually reaches the first predetermined value. Thus, it is continuously monitored whether enough pulses arrive to reach the first predetermined value in a short time. The skilled person will appreciate that determining that so many pulses are received in such a long time can be achieved by other means other than having a decay rate, for example by resetting the first integral output after a predetermined time span.

[0071] Adjusting the decay rate allows defining the length of the time window considered by the processing means and controlling the behavior of the system and adapting the system, for example, according to the available optical power. For example, we assume that the optical system has a light source that is scanned along different parts of the scene and detected by the optical sensor (as described below). Having a slow decay rate (i.e. increasing the observation window) reduces the filtering capacity of the processing means, however it allows reducing the power consumption of the light source, since a reduction in the optical power will reduce the chances of detecting a photon within a fixed observation window. Therefore, setting a longer observation window (i.e. a slower decay rate) allows reducing the optical power at the expense of reducing the filtering of irrelevant photon detections.

[0072] Preferably, each processing device is a first processing device connected in series to a second processing device. For example, the output of the first processing device is input to the second processing device. In principle, the first processing device and the second processing device are similar, except that the decay rate in the first processing device is faster than the decay rate in the second processing device. This means that the first predetermined time span in which a certain total number of pulses need to be received in the first processing device is different from the second predetermined time span in the second processing device. This allows the two processing devices to have different roles. For example, in the case where the first processing device has multiple inputs, the first processing device ensures consistency conditions (spatial conditions). For example, by projecting points on the scene using the light source and applying a certain shape on the projected points, we can constrain the pixels to mark valid detections only when an applied kernel corresponding to an expected point shape in the pixel array is detected within a specified time. For example, a 2x1 kernel can be applied, which means that the projected point must span at least 2x1 pixels. This is generally applicable to elliptical point shapes. When a light pulse of length tp is emitted, the constraint imposed on the pixel will be that a cluster of 2x1 pixels must be emitted within tp in order to regard the detected photons as coming from active projection. Preferably, tp is on the order of 1 nanosecond.

[0073] In other words, the condition requires that the input to more than one sensing unit be positive within a predetermined time span before the processing device generates an output. In this case, for example, if only one input is positive, the integrated output will not reach the predetermined value because it requires at least one or two adjacent sensing units to have a positive input to the processing device.

[0074] The first predetermined value in the first processing device may be different from the predetermined value in the second processing device (hereinafter referred to as the second predetermined value) and may be independently adjustable.

[0075] The second processing means has a different role, namely to ensure persistence conditions (spatiotemporal conditions). For example, by using a light source to project a dot, we can enforce that the trajectory of the dot is continuous, as long as the pulse repetition frequency is higher than the displacement speed of the dot. Let's assume that a 1 nanosecond pulse is repeated every 40 nanoseconds, and a time window of 41 nanoseconds will be able to capture 2 pulses. By enforcing that a pixel (or cluster of pixels) must see at least 2 pulses in a 41 nanosecond window, we can filter out a large portion of the detected ambient photons, because the ambient light photon statistics have a low chance of generating 2 consecutive detections in a 42 nanosecond window.

[0076] For example, the second processing means is adapted to integrate the first output signal of the first processing means to obtain a second integrated output, wherein the second processing means is adapted to generate a second output signal when the second integrated output is at least the second predetermined value. For example, wherein the time required to reach the second predetermined value is a second predetermined time span that is longer than the first predetermined time span. For example, the first predetermined time span is at least half as short as the second predetermined time span.

[0077] In other words, the condition (i.e., persistence) requires that the input to the second processing device includes many pulses (continuously or nearly continuously) over a period of time, which means that the input is persistent, which means that the input is likely to be true. For example, if 9 out of 10 consecutive detections by the sensing unit are positive, then the detection is likely to be positive. The decay rate of the second processing device is longer because the second processing device needs to check or monitor the input over a longer period of time, while the decay rate of the first processing device is shorter because the first processing device needs to check or monitor the input (i.e., the output of more than one sensing unit) over a short period of time. By applying a combination of both consistency and persistence, a powerful filtering strategy is created that is very efficient in hardware implementation.

[0078] Preferably, the processing means is adapted to associate weights with the pulses of the sensing cell itself and each connected neighbouring sensing cell. This can be achieved, for example, by providing a unique pulse length to the pulses of each neighbouring sensing cell, the pulse length modulating the contribution of each pulse to the integrated signal. Another implementation may be to associate a unique charge with each pulse signal, which is added to or subtracted from the integrated signal when each pulse is present. In this way, by assigning different weights, we can prioritise contributions from different types of neighbours. For example, diagonal neighbours may be given a lower weight than directly adjacent neighbours, with the aim of minimising the chance that ambient or thermal detections contribute to real detections, as these detections originate from active light.

[0079] Preferably, the processing device is adapted to at least partially reset after generating its output signal. For example, the processing device returns to its initial state after generating the output signal. For example, the value of the integral output signal is reset. Preferably, the two processing devices connected in series are reset to different values. For example, the integral output signal in each processing device is reset to a different value. For example, the first processing device is completely reset, for example, the first integral output signal is completely reset, for example, reset to zero, and the second processing device is partially reset, for example, the second integral output signal is partially reset, for example, reset to half of its value. This is useful in the case where a pulse arrives while the integral output signal in the processing device is above a threshold and has not yet decayed. For example, in the case of a long string of pulses detected, where the integral output quickly reaches a threshold or a predetermined value. In this case, the integral output should be reset to zero so that more pulses can be processed.

[0080] In this case, the first processing means is completely reset, since it requires a plurality of sensing units to detect pulses within a short period of time, in which case the first predetermined value is reached quickly. However, the second processing means is partially reset, since it requires a sensing unit to detect pulses within a longer period of time, in which case the second predetermined value is reached slowly. A complete reset means that in case of a pulse train which is repeated faster than the decay rate of the processing means, then only 1 / 2 of the pulses will result in a pulse at the output of the processing means. A partial reset avoids this situation and provides a continuous pulse train at the output of the processing means, in which case only the first pulse will not result in a pulse at the output (see Figure 4 , 5 and 9).

[0081] Preferably, the system comprises a switching element for each processing device, wherein the switching element is adapted to allow bypassing of the processing device. For example, the system is programmable such that the switching means bypasses one of the first processing device or the second processing device. This allows selecting the most appropriate processing device depending on the conditions, for example depending on the ambient light conditions or depending on the effective signal strength. For example, if the signal strength is too low, using two processing devices in series may not be preferred. For example, in the case of 2 sequential processing devices, there are three situations: (1) the first processing device is on and the second processing device is off, (2) the first processing device is off and the second processing device is on, or (3) the first processing device is on and the second processing device is on.

[0082] For example, in cases where ambient light is negligible, it is advantageous to use only the second processing means (case 2), thus reducing power consumption (i.e., power consumption of the first processing means) and increasing processing speed. However, in cases where ambient light is significant, it is advantageous to use the first processing means (case 1) to additionally filter out noisy detections due to ambient light. Finally, in cases where very reliable detection is required, it is advantageous to use both the first and second processing means (case 3), which allows two layers of filtering and high detection reliability.

[0083] In another example, this may also depend on the bandwidth of the received signal, for example, in case of very high bandwidth, the first processing device (i.e., with a shorter time window) may be suitable for use, while for low bandwidth, the second processing device (i.e., with a longer time window) may be suitable for use.

[0084] Preferably, each processing device is a neuron, preferably a leaky integrate and fire (LIF) neuron, but the skilled person will appreciate that other types of neural networks can be used, such as convolutional networks, spike networks, etc. In the case where the processing device is a neuron, all implementations discussed above with respect to the processing device are applicable. For example, the system preferably includes a first neuron and a second neuron connected in series. The first neuron has at least two or at least three inputs, such as the output of one sensing unit and one or two adjacent sensing units. The second neuron has one input, wherein the input is the output of the first neuron. Also similar to the above discussion, the first neuron has inputs from the sensing unit and the adjacent sensing unit, which allows consistent detection within a first time window - detection of photons in adjacent pixels within a certain time span, the first window is usually short and the same size as the light pulse to be detected, while the second neuron allows persistent detection within a second time window - repeated detection of photons or neuron discharges in the same pixel during a certain second time span, the second time window is usually longer, preferably equal to or slightly longer than the repetition time of the light pulse. Depending on the light conditions, a switching device can be used to select one or both neurons for operation, so that a programmable filter behavior can be achieved by operating the switch. The advantage of using neurons is that this architecture achieves very low power consumption and compact implementation. Also similar to the above, the neuron is suitable for integrating at least one of its inputs, wherein each neuron is suitable for generating an output signal, preferably a digital output signal or a digital pulse, when the value of the integration is at least the predetermined value. However, in practice, for the first neuron (i.e., the neuron performing consistency detection), if the pulse comes from only one sensing unit, the pulse is integrated in the first neuron, but it is not enough to reach the predetermined value. Thus, in practice, in order to reach the predetermined value, it is necessary to integrate the outputs from multiple sensing units. Similar to the above, each neuron is suitable for at least partially resetting after generating the output signal. In one implementation, the neuron resets itself after the discharge. In another implementation, the neuron is partially reset (e.g., reset midway). This avoids the situation where the pulse arrives when the neuron is still above the predetermined value or threshold. The choice of leaky integrate and fire neurons is suitable for the present invention because the neuron can integrate the output of the sensing unit to obtain an integrated output, wherein the integrated output leaks (i.e., decays) over time, and wherein the neuron can discharge when the integrated output reaches a predetermined value.

[0085] In a configuration in which two neurons are connected in series, the first neuron is preferably completely reset because the first neuron has a faster decay rate, i.e., a short time frame, such as at most 10 nanoseconds, preferably at most 5 nanoseconds, more preferably at most 2 nanoseconds or 1 nanosecond. In other words, if a predetermined number of pulses arrive within said short time frame, the detection of the first neuron is almost instantaneous. For example, in the case where the sensing unit and one or two of its neighbors detect the impinging photon within a short time. On the other hand, the second neuron is preferably partially reset because the second neuron has a slower decay rate, i.e., a long time frame, such as at least 10 nanoseconds, preferably at least 20 nanoseconds or at least 50 nanoseconds. For example, the decay rate of the first neuron (or in general, the first processing device) is at least twice, or at least three or four times, or generally much faster than the decay rate of the second neuron (or in general, the second processing device). The second neuron is not affected by ambient light because the ambient light has been filtered by the first neuron. The second neuron then has a long time frame, since it is responsible for finding out whether the detection is persistent, i.e. whether the detection is positive or '1' in at least M preferably consecutive observation windows within the past N observation windows, where M is greater than 1 and at most equal to N. For example, whether in 10 consecutive time windows the detection is positive in most or all of the time windows, e.g. in 9 or 8 time windows. In addition to the first level of filtering of the first neuron, the second neuron also provides a second level of filtering.

[0086] References to one or two processing devices (or neurons) in the present invention should not be construed as limiting in this regard. The skilled person will appreciate that more layers of neurons may be used, for example to achieve further filtering. For example, a third neuron may be able to check whether a neighbor has fired within a certain time span in the past.

[0087] Preferably, the system further comprises at least one light source and a scanning device, wherein the scanning device is adapted to scan a light beam from the light source at least partially over the scene along a trajectory. This is advantageous in allowing triangulation and determination of the depth profile of the field of view. For example, the depth profile can be estimated by triangulating the points of the at least one object in the scene detected by the optical sensor using the data of the emitted light of the at least one light source. This is similar to triangulating the outputs of two optical sensors, since the data of the emitted light of the light source is known, i.e. it is known at which part of the scene the light will be illuminated at a certain time. An advantage of the present invention is asynchronous detection, since the light source and the optical sensor are asynchronous, i.e. the generation and detection of light are asynchronous. However, the light source and the optical sensor must be calibrated, for example so that the pattern and the projection angle at each timestamp are known.

[0088] Preferably, the system comprises at least one optical sensor, wherein the distance is estimated by a displacement on the sensor of an optical signal corresponding to a point of the at least one object in the scene detected by the optical sensor, the displacement being an expected position based on a priori knowledge of the light source and / or the displacement being a displacement of an optical signal corresponding to a point of the at least one object in the scene detected by at least one other optical sensor. This is advantageous because it is simple to obtain the distance between each object and the sensor.

[0089] Scanning the light beam from the light source is advantageous in allowing triangulation. For example, by using a system comprising a light source illuminating a light beam onto a scene (e.g. an environment) and two optical sensors oriented differently from each other, for example, wherein the two sensors (i.e., their sensing units) have a shared field of view of the scene, the xy-time data of the two sensors (i.e., their sensing units) can be converted into xyz-time data by triangulation. For example, by making the light source suitable for illuminating a light beam onto the scene in an illumination trajectory, wherein the light source comprises means suitable for scanning the light beam (preferably continuously) over the scene, wherein the optical sensor monitors the light spot generated by the light beam and outputs the position of a point of at least one object in the scene (e.g., a point of the surface of the object) along the trajectory at a plurality of time points, wherein the xy-time data of the two optical sensors (i.e., their sensing units) can be converted into xyz-time data using triangulation. The light source can, for example, act as a reference point so that the position of the point of the object of the first optical sensor can be synchronized with the position of the point of the object of the second optical sensor to create a depth or z dimension. The light source may illuminate and scan the light beam over the scene to be imaged in a Lissajous manner or pattern or raster scan etc. This illumination trajectory advantageously allows efficient and fast image detection since after a few illumination cycles a large portion of the image has been illuminated. Other illumination patterns are contemplated.

[0090] Preferably, the scanning is continuous, so that objects in the scene are continuously scanned and identified. For example, a light source generates a light beam, which generates a light spot on an object, wherein the light beam continuously scans the scene along the trajectory, wherein, for example, at intervals, the light beam will scan all or almost all of the scene. Then, the reflected signal is received by at least one sensing unit. The scanning can be, for example, a Lissajous pattern or a grating scan. The scanning device can be, for example, a MEMS scanner, a reflector, an optical phased array, or a metasurface close to the beam scanning.

[0091] For example, the light source is adapted to be at a wavelength detectable by the sensing unit, for example between 100 nanometers and 10 micrometers, preferably between 100 nanometers and 1 micrometer. For example, the optical sensor comprises a photodetector or a matrix of photodetectors capable of detecting photons impinging on each detector within a wavelength detection window falling within the range of 100 nanometers and 10 micrometers, preferably between 100 nanometers and 1 micrometer.

[0092] Preferably, the scanning is achieved by discrete steps in the scanning angle, wherein the sensing sampling frequency of the sensor is typically faster than the stepping frequency of the scanner.

[0093] Preferably, the light energy emitted along the scanning trajectory is pulsed, meaning that the scanning structure scans the light beam across the scene in a discrete or continuous manner, and the beam energy distribution is pulsed in time, for example, the light beam is a pulse train of 1 nanosecond pulses that repeat every 40 nanoseconds. In this case, the preferred time constant of the first processing device considering consistency will be 1 nanosecond to 2 nanoseconds, while the time constant associated with the second processing device will be 41-42 nanoseconds or slightly longer.

[0094] Preferably, each optical sensor comprises a matrix of pixel sensors connected to a matrix of processing means. For example, the system comprises more than 100 pixel sensors, preferably more than 1,000 pixel sensors, more preferably more than 10,000 pixel sensors, even more preferably more than 100,000 pixel sensors, most preferably more than 1,000,000 pixel sensors. For example, the system may be arranged in a matrix, wherein the optical sensors comprise 1,000 pixel sensor rows and 1,000 pixel sensor columns.

[0095] Preferably, the system comprises image representation means, such as a screen or other image representation device, for reproducing the positions of points of said objects in said scene.

[0096] Preferably, the optical system is used for 3D vision applications. For example, the system can be used to visualize objects in three dimensions. Alternatively, the sensor can allow analysis of a scene, for example by extracting features of objects in the scene, without necessarily generating an image of the scene.

[0097] Preferably, the system also includes multiple optical sensors (e.g., at least two optical sensors) and / or multiple light sources. This facilitates creating 3D vision and allows triangulation (e.g., between a sensing unit of a first optical sensor and a sensing unit of a second optical sensor) to obtain depth data. For example, each optical sensor can be oriented differently so that a 3D perception of the scene to be imaged is captured, for example, by triangulating the outputs of two such optical sensors.

[0098] Preferably, the processing device, such as the first processing device, generates an output signal, such as the first output signal, when multiple inputs thereof (such as inputs of more than one sensing unit) are positive within a period of time (such as the first predetermined time span). Alternatively, the processing device (such as the second processing device) generates an output signal, such as the second output signal, when its input (such as input from the first processing device) includes continuous pulses within a period of time (such as the second predetermined time span). This can be summarized as follows: the processing device generates an output when multiple inputs thereof are positive within a period of time, or the processing device generates an output when one of its inputs contains continuous (or semi-continuous) pulses within a period of time.

[0099] Preferably, each processing device is a first processing device connected in series to a second processing device, wherein the second processing device is adapted to integrate the first output signal of the first processing device to obtain a second integrated output, wherein the second processing device is adapted to generate the second output signal when the second integrated output is at least a second predetermined value for a second predetermined time span. This is similar to that described above. However, rather than relying on the decay rate of each processing device, the first processing device is reset after a first predetermined time span and the second processing device is reset after a second predetermined time span, wherein the first predetermined time span is half as short as the second predetermined time span.

[0100] In a second aspect, the invention relates to a method for optical sensing. The method comprises the step (a) of receiving a plurality of optical signals, for example using a plurality of sensing units according to the first aspect.

[0101] The method further comprises a step (b) of integrating the signals in parallel for each optical signal to obtain a first integrated output. For example, a processing unit is associated with each sensing unit, wherein each processing unit integrates the optical signal input thereto to obtain the first integrated output. The method further comprises a step (c) of generating a first output signal when the first integrated output is at least a first predetermined value within a first predetermined time span.

[0102] Preferably, the method further comprises the step of adapting the plurality of sensing units to output a signal upon detection of a photon. Preferably, the method further comprises the step of decaying the integrated output over time, wherein the rate of decay is preferably adjustable.

[0103] Preferably, the method also includes a step (d) of integrating the signals in parallel for each first output signal to obtain a second integrated output, and a step (e) of generating a second output signal when the second integrated output is at least a second predetermined value, wherein the attenuation rate in steps (a) to (c) is faster than that in steps (d) to (e).

[0104] Preferably, the method further comprises the step of resetting the first integrated output after generating the first output signal, and the step of resetting the second integrated output after generating the second output signal.

[0105] Any features of the second aspect (method) may be as correspondingly described in the first aspect (system).

[0106] In a third aspect, the invention relates to the use of the system according to the first aspect and / or the method according to the second aspect for optical sensing.

[0107] Further characteristics and advantages of embodiments of the invention will be described with reference to the accompanying drawings.It should be noted that the invention is not limited to the specific embodiments shown in the drawings or described in the examples, but only by the claims.

[0108] Figure 1 An optical sensing system (1) according to an embodiment of the present invention is shown, the optical sensing system (1) comprising an optical sensor (30), the optical sensor (30) comprising a plurality of sensing units (2', 2", 2'"), the system (1) further comprising an optical device (3) capable of generating an image of a scene (4) on the optical sensor (2', 2", 2'") . The optical sensing system (1) further comprises a light source (14), the light source (14) irradiating a light beam (16) on the scene (4) to generate a light spot or light point. The system (1) further comprises a scanning device (15) suitable for scanning the light beam (16) on the scene (4). For example, the scanning device (15) may be a reflector or a MEMS mirror, etc., which are capable of scanning the light beam (16) on the scene (4), for example, on different parts of the scene (4). The light beam (16) is scanned along the irradiation trajectory, for example, in a Lissajous manner. Then, the reflection signal from the scene (4) is captured by the sensing units (2', 2", 2'") .

[0109] The system (1) also includes a plurality of processing devices (5', 5", 5"', 5""), namely, a first processing device (5', 5") adapted to receive the output (7', 7", 7"') of a sensing unit (2', 2", 2"'). The system (1) includes at least one adjacent sensing unit (2", 2"') adjacent to the sensing unit (2'). Similarly, the system (1) includes at least one adjacent processing device (5") adjacent to the processing device (5'). Each sensing unit (2', 2", 2"') is connected to a processing device (5', 5") corresponding thereto, and is connected to an adjacent processing device (5") corresponding to an adjacent sensing unit (2", 2"').

[0110] The output (7', 7", 7'") of the sensing unit (2', 2", 2'") is input to the first processing device (5', 5"). The first processing device (5', 5") is connected in series to the second processing device (5', 5"), wherein the output end of the first processing device (5', 5") is connected to the input end of the second processing device (5', 5")

[0111] The outputs (7', 7", 7'" of the sensing units (2', 2", 2'"'), the outputs (8) of the first processing device (5', 5") and the outputs of the second processing device (5', 5") are digital outputs, such as pulse trains. As shown in the figure, the first processing device (5', 5") is suitable for receiving multiple inputs from multiple sensing units (2', 2", 2'"'), such as a sensing unit (2') and at least one adjacent sensing unit (2", 2'"'). The second processing device (5', 5") is suitable for receiving only one input, namely the output of the first processing device (5', 5").

[0112] The processing device (5') Figure 1 2 '. However, the processing means (5') may also be part of the sensing unit (2'). In the case where the processing means (5') is a neuron, the neuron is preferably directly connected to the sensing unit (2').

[0113] Figure 2 An optical sensing system (1) according to an embodiment of the present invention is shown, comprising a plurality of sensing units (2', 2", 2'") and a plurality of processing devices (5', 5"), wherein each sensing unit (2', 2", 2'") is connected in parallel to a corresponding processing device (5', 5") . The processing device (5', 5") comprises a first neuron (12) and a second neuron (13) connected in series. Figure 1 , the output (6) of the first neuron (12) is input to the second neuron (13). The input (7', 7", 7"') and output (6, 8) of each neuron (12, 13) are digital, such as pulses. For example, a beam of photons is incident on the sensing unit (2', 2", 2"'), and the sensing unit outputs a series of pulses accordingly. Figure 2 In the example, the first neuron (12) acts as Figure 1 The first processing device (5', 5") in the embodiment of the present invention, while the second neuron (13) acts as Figure 1 The second processing device (5"', 5"").

[0114] Figure 3A processing device (5) according to an embodiment of the invention is shown, wherein the input (7) of the processing device (5) is a stream of pulses. The processing device (5) is adapted to generate a pulse (21) at an output (8) of the processing device (5) when the total number of pulses (11) is at least a predetermined number (11) within a predetermined time span (Δt). In the figure, the total number of pulses (11) is shown as coming from only one input (7). However, in practice, the total number of pulses (11) preferably takes into account a plurality of inputs. This allows filtering out ambient light or false detections, since it is difficult to obtain a total number of X false detections from a sensing cell and its neighbours within a time span of Y. For example, in the case where there are three inputs to the processing device (5) corresponding to three sensing cells (2', 2", 2"') adjacent to each other, and in the case where a total of 9 pulses need to arrive within 2 nanoseconds in order to allow the processing device to generate an output pulse (21), then approximately three pulses need to arrive within 2 microseconds per sensing cell (2', 2", 2"'). This is a very effective way to reduce false detections and obtain reliable detections.

[0115] Figure 4 An example of a processing device (5) comprising a first neuron (12) and a second neuron (13) according to an embodiment of the present invention is shown. An example of an output (7") of a sensing unit (2") is shown in (a), wherein the output (7") is a pulse train (19) corresponding to an event of the sensing unit (2"). These pulses (19) are input to the first neuron (12). Similar pulses of adjacent sensing units (2', 2'") are also output (7', 7'") by the adjacent sensing units (2', 2'") and input into the first neuron (12). As shown in (b), the first neuron (12) is adapted to integrate its inputs (7', 7", 7'") . For example, at a first time point (t1), a pulse is received, whereby the value of the integrated output (10) jumps. The integrated output (10) in the first neuron (12) decays at a decay rate. At a second time point (t2), another pulse is received, whereby the integrated output (10) again jumps to a value higher than the threshold value (9) or the first predetermined value. Since the integrated output (10) (within the first predetermined time span (Δt)) crosses the limit of the threshold value (9), the first neuron (12) generates an output pulse (17) at a third time point (t3), as shown in (c). This occurs again at a fourth time point (t4) and a fifth time point (t5), whereby another output pulse is generated at a sixth time point (t6).

[0116] The integrated output (10) is the result of integrating not only the output (7") of one sensing unit (2"), but also the output (7', 7'") of adjacent sensing units (2', 2'"). Thus, in the event that a pulse comes from only one sensing unit (2"), which may thus be the result of an erroneous detection, the integrated output (10) of the one sensing unit (2") will not be high enough to reach the threshold value (9). Thus, in order to reach the threshold value (9), the outputs (7', 7", 7'") of one sensing unit (2") and its adjacent sensing units (2', 2'") are needed to ensure reliable detection and filter out erroneous detections.

[0117] The output (6) of the first neuron (12) comprising a pulse (17) is input to the second neuron (13). Unlike the first neuron (12), the second neuron (13) has only one input (6). The second neuron (13) is also adapted to integrate its input (6). However, the integrated output (23) decays at a different decay rate than in the first neuron (12). This is shown in (d), where the decay rate of the integrated output (23) in the second neuron (13) is slower than the decay rate of the first integrated output (10) in the first neuron (12). Having a slower decay rate allows knowing whether a detection is a persistent detection or not. For example, the first neuron (12) allows detecting reliable detections and filtering out false detections by depending on the detections of neighboring sensing units, or what we call consistency detections. On the other hand, the second neuron (12) allows detecting persistent detections, such as detections that occur continuously over a period of time, here, a second predetermined time span (Δt2), or what we call persistence detections. Since the pulse (17) at the input end of the second neuron (13) is continuous, and since the second integrated output (23) in the second neuron (13) reaches the threshold (9), a pulse (18) is generated by the second neuron at the seventh time point (t7) as shown in (e). For better understanding, Figure 5 (a-e) show the vertical view of Figure 4 (a~e) Similar situations.

[0118] Figure 6A processing device (5) is shown, wherein in (a), a first neuron (12) and a second neuron (13) are connected in series. Another advantageous implementation is shown in (b), wherein a switching device (22', 22") is located before each of the first neuron (12) and the second neuron (13). For example, the first switching device (22') is located before the first neuron (12), and the second switching device (22") is located before the second neuron (13). Each switching device (22', 22") allows the neuron (12, 13) corresponding thereto to be bypassed. For example, the first switching device (22') allows the first neuron (12) to be bypassed, wherein the output (7') of the sensing unit (2') is immediately input to the second neuron (13). Similarly, the second switching device (22") allows the second neuron (13) to be bypassed, wherein the output of the first neuron (12) becomes the output of the processing device (5). Only one (or none) switching device (22', 22") may be operational at a time, otherwise the processing device (5) will be unable to process any input and produce any useful output.

[0119] The switching arrangement (22', 22") is advantageous in situations where only one neuron is required for operation. For example, in situations where ambient light or noise is very low, the first neuron (12) can be bypassed without compromising the reliability of the system. Similarly, bypassing the second neuron (13) is advantageous in situations where the output of the first neuron does not require further filtering. Bypassing neurons (12, 13) may also be advantageous, for example in terms of reducing power consumption and increasing processing speed.

[0120] Figure 7 An example of synchronous detection commonly seen in the prior art is shown. This example assumes that a pulse train (24) is input to the system, as shown in (a), and assumes that observation windows (25) are separated by constant time periods (t), as shown in (b). Since the observation window (25) only covers certain time points, no pulses outside the observation window (25) will be detected. For example, since only two pulses (24) coincide with two observation windows (25), the output (26) includes only the two pulses, while the other two pulses are not detected. However, in the present invention, asynchronous detection is obtained, which overcomes the disadvantages of synchronous detection.

[0121] Figure 8The detection situation under different ambient light conditions according to an embodiment of the present invention is shown. In (a), the detection is performed under strong ambient light conditions. Therefore, a sensing unit (2') and two adjacent sensing units (2", 2"') are required. The outputs of the sensing unit (2') and the two adjacent sensing units (2", 2"') are fed to the first neuron (12) or the first processing system (5'). However, in (b), the detection is performed under weak ambient light conditions. Therefore, one sensing unit (2') and one adjacent sensing unit (2") are sufficient. The outputs of the one sensing unit (2') and the one adjacent sensing unit (2") are input into the first neuron (12) or the first processing device (5'). An implementation that allows switching between (a) and (b) can be obtained by using switching elements (22', 22"), wherein under strong ambient light conditions, the first switching element (22') is not activated and the first neuron (12) processes its input. However, under weak ambient light conditions, the first switching element (22') is activated, the first neuron (12) is bypassed, and the input is fed directly to the second neuron (13). Other configurations can be envisioned based on other conditions, such as effective light intensity, bandwidth, etc.

[0122] Fig. 9 An example of asynchronous detection according to an embodiment of the present invention is shown. Each time a pulse (29) of a different sensing unit is received in a first time window (T1, for example 1 nanosecond) and the integrated output exceeds a predetermined value of the processing device (i.e., the number of pulses is greater than a predetermined number), the first processing device generates a pulse (27). After this, there is a second time window (T2, for example 20 nanoseconds) associated with the second processing device, wherein the second time window (T2) is much longer. The first processing device (5', 5") is connected to the second processing device (5"', 5""). Each time a pulse of the first processing device is received in the second time window (T2) and the integrated output exceeds a predetermined value of the processing device (i.e., the number of pulses is greater than a predetermined number), the second processing device generates a pulse (28).

[0123] Other arrangements for achieving the purposes of the methods and apparatus embodying the present invention will be clear to those skilled in the art. The foregoing description gives details of certain embodiments of the present invention. However, it is apparent that no matter how detailed the foregoing is in the text, the present invention can be applied in a variety of ways. It should be noted that the use of certain terms in describing certain features or aspects of the present invention should not be interpreted as implying that the terms herein are again defined to be limited to the specific features or aspects of the present invention in conjunction with the terms.

[0124] Reference numerals list

[0125] 1Optical sensing system

[0126] 2', 2", 2"' sensing unit and adjacent sensing unit

[0127] 3 Optics

[0128] 4 Scenario

[0129] 5', 5" processing device / first processing device 5"', 5"" second processing device

[0130] 6 Output signal of the (first) processing device / output signal of the first neuron

[0131] 7Output signal of optical sensor

[0132] 8 Output signal of (first) processing device / output signal of second neuron

[0133] 9First predetermined value

[0134] 10 First integral output of optical sensor

[0135] 11 Number of pulses of the optical sensor within a predetermined time

[0136] 12 First neuron 13 Second neuron

[0137] 14 Light Source

[0138] 15 Scanning device

[0139] 16 Beam

[0140] 17 Digital output signal of the first neuron

[0141] 18 Digital output signal of the second neuron

[0142] 19 Digital input signal of processing device / digital output signal of optical sensor

[0143] 20 Decay rate of neurons

[0144] 21 Processing device digital output signal

[0145] 22 Switching elements

[0146] 23 The second integral output of the first neuron

[0147] 24 Input pulse train of synchronous detection system

[0148] 25 Observation Window

[0149] 26 Synchronous detection system output

[0150] 27 Output of the first neuron or processing device

[0151] 28 Output of the second neuron or processing device

[0152] 29 Pulses of different sensing units 30 Optical sensors

[0153] t1 first time point t2 second time point t3 third time point

[0154] t4 fourth time point t5 fifth time point t6 sixth time point

[0155] t7 seventh time point

[0156] 31 Output signal of the second processing device / second output signal

[0157] 32 Second preset value

Claims

1. An optical sensing system (1) for optical sensing, comprising: - at least one optical sensor (30), the optical sensor (30) comprising a plurality of sensing units (2'), the plurality of sensing units (2') preferably being arranged in a matrix, wherein each of the sensing units (2') is preferably a pixel sensor (2'), wherein each sensing unit (2') comprises a photodetector, wherein each photodetector is adapted to output a signal upon detecting a photon, - an optical device (3) capable of producing an image of a scene (4) on said optical sensor (30), - a plurality of processing devices (5), wherein each of said processing devices (5') is connected to at least one sensing unit (2') corresponding thereto, wherein each of the processing means (5') is adapted to receive at least one input corresponding to at least one output (7') of a corresponding sensing unit (2'), Each of the processing means (5', 5") is adapted to integrate the output (7', 7", 7'") of the corresponding sensing unit (2', 2", 2'") to obtain a first integrated output (10), wherein the processing means (5', 5") is adapted to generate a first output signal (8) when the first integrated output (10) reaches at least a first predetermined value (9) within a first predetermined time span (Δt), Features The system (1) further comprises at least one light source (14) and a scanning device (15), wherein the scanning device (15) is adapted to scan a light beam (16) from the light source (14) along a trajectory on the scene (4), preferably continuously.

2. An optical sensing system (1) according to claim 1, wherein the system (1) includes at least one adjacent sensing unit (2", 2"') adjacent to the sensing unit (2'), wherein the system (1) includes at least one adjacent processing device (5") adjacent to the processing device (5'), wherein each sensing unit (2') is connected to the at least one adjacent processing device (5") corresponding to the at least one adjacent sensing unit (2", 2"').

3. An optical sensing system (1) according to claim 2, wherein each of the processing devices (5', 5") is suitable for integrating the output (7', 7", 7'") of the corresponding sensing unit (2', 2", 2'") and at least one adjacent sensing unit (2", 2'") to obtain the first integrated output (10).

4. The optical sensing system (1) according to any one of the preceding claims, wherein each first integrated output (10) has a decay rate (20), wherein each said decay rate (20) is adjustable.

5. An optical sensing system (1) according to claim 4, wherein each processing device (5', 5") is a first processing device (5', 5") connected in series to a second processing device (5'", 5"), wherein the decay rate of the first processing device (5', 5") is at least twice the decay rate of the second processing device (5'", 5") 6. An optical sensing system (1) according to any one of claims 5, wherein the second processing device (5'", 5'") is suitable for integrating the first output signal (8) of the first processing device (5', 5") to obtain a second integrated output (23), wherein the second processing device (5'", 5'") is suitable for generating a second output signal (31) when the second integrated output (23) is at least a second predetermined value (32) within a second predetermined time span.

7. An optical sensing system (1) according to any one of claims 5 or 6, wherein each processing device (5', 5", 5"', 5"") is suitable for being at least partially reset after generating its output signal (8, 31), wherein the first processing device (5', 5") is fully reset and wherein the second processing device (5"', 5"") is partially reset.

8. An optical sensing system (1) according to any one of claims 1 to 4, wherein each processing device (5', 5") is a first processing device (5', 5") connected in series to a second processing device (5'", 5"), wherein the second processing device (5'", 5") is suitable for integrating the first output signal (8) of the first processing device (5', 5") to obtain a second integrated output (23), wherein the second processing device (5'", 5") is suitable for generating a second output signal (31) when the second integrated output (23) is at least a second predetermined value (32) within a second predetermined time span, wherein the first processing device is reset after the first predetermined time span, and wherein the second processing device is reset after the second predetermined time span, wherein the first predetermined time span is half of the second predetermined time span.

9. An optical sensing system (1) according to any of the preceding claims, wherein the system (1) comprises a switching element (22', 22") for each processing device (5', 5", 5"', 5""), wherein the switching element (22', 22") is suitable for allowing the processing device (5', 5", 5"', 5"') to be bypassed.

10. The optical sensing system (1) according to any of the preceding claims, wherein each processing means (5', 5", 5"', 5"") is an artificial neuron (12, 13), preferably a leaky integrate and fire (LIF) neuron.

11. An optical sensing system (1) according to any one of the preceding claims, wherein the system (1) comprises at least two optical sensors (30), wherein the system (1) is suitable for triangulating data of a sensing unit (2) of one optical sensor (30) with data of a sensing unit (30) of another optical sensor (30).

12. An optical sensing system (1) according to any one of the preceding claims or any one of claims 6 to 12, wherein the processing means (5', 5") generates the first output signal (8) when more than one of its inputs is positive within the first predetermined time span, or The processing means (5'", 5"") generates the second output signal (31) when one of its inputs comprises consecutive pulses within the second predetermined time span.

13. A method for optical sensing, comprising the steps of: a- using an optical device (3) to generate an image of a scene (4) on an optical sensor (30), the sensor (30) comprising a plurality of sensing units (2), b- receiving a plurality of optical signals (7') from the plurality of sensing units (2), c- for each optical signal, integrating said signal (7') in parallel to obtain a first integrated output (10), and d- generating a first output signal (8) when the first integrated output (10) is at least a first predetermined value (9) within a first predetermined time span (Δt), Characterized in that the method comprises the following steps: e- With the aid of at least one light source (14) and a scanning device (15), a light beam (16) from the light source (14) is scanned along a trajectory over the scene (4).

14. The method of claim 13, wherein the method further comprises the step of decaying the first integrated output (10) over time, wherein the rate of decay is adjustable.

Images