Event-driven measurement system comprising a master sensor and associated measurement method

Abstract

Event-driven measurement system comprising a master sensor and associated measurement method. The invention relates to an event-driven measurement system (1) and its associated method, in which a network (10) of sensors performs measurement cycles comprising: measuring a parameter from time t0 until time t when the measurement reaches a threshold or limit time, then sending a read request (22); at least one master sensor (110) is selected from among these sensors; a logic circuit (12) receives two requests (22) from the master sensor (110), determines (31), for each request (22), a measurement time tmes, determines (32) a difference Δtmes between these measurement times tmes, and when the difference Δtmes is greater in absolute value than a value Δtmes-th, triggers (33) a measurement event (4). Figure for the abstract: Fig. 7

Description

Title of the invention: Event-driven measurement system comprising a master sensor, and associated measurement method technical field

[0001] The present invention relates to the field of event-driven measurement systems, that is to say, systems capable of detecting a change in a measured parameter dynamically. It finds a particularly advantageous application in the field of imagers. STATE OF THE ART

[0002] In event-driven measurement systems, a parameter to be measured, representative of a physical quantity, is dynamically measured to detect a change in that parameter. Thus, if the parameter does not change, it is not necessary to repeat the measurement. Conversely, if the parameter changes significantly, its measurement can be updated. Event-driven measurement systems make it possible to limit the number of measurements as well as the volume of data generated. To this end, these systems aim in particular to limit spatial and / or temporal redundancies.

[0003] The power consumption of these systems is reduced. They are therefore particularly advantageous in applications of autonomous objects, for example for video surveillance in the case of imagers.

[0004] In the specific example of imagers, conventional imagers typically comprise a two-dimensional array of sensors, each containing at least one photodiode, for example. For each image, a luminance measurement can be performed by each sensor. The larger the sensor array, the greater the number of measurements required for a given refresh rate. The measured analog data, particularly luminance, is converted into digital data. This conversion step can even be the primary contributor to power consumption.

[0005] In the field of imaging, to eliminate spatial redundancies, there are sensor arrays called TFS (Time to First Spike) arrays. A TFS sensor, while not limited to the field of imaging, is configured to perform measurement cycles. Each cycle typically includes - a measurement of the parameter representing the physical quantity, for example luminance, from a time t0 until a time t where the measurement reaches a threshold value or a time t corresponding to a determined limit time, and - when the measurement reaches the threshold measurement value or the time limit is reached, a request to read the measurement is sent.

[0006] The time between time t0 and time t can then be representative of the physical quantity to be measured.

[0007] In the example of the imagers, the time measured between the start of the operation and the crossing of the threshold represents a measurement of the luminance. TFS sensors can limit spatial redundancies by generating read requests that are processed in a grouped manner for different sensors in the array.

[0008] In the field of imaging, to eliminate temporal redundancies, there are also sensors called "DVS" (from the English "Dynamic Vision Sensor"). These sensors are configured to compare the voltage across two capacitors representing present and past luminance levels. This comparison is made in an analog manner. DVS sensors make it possible to eliminate temporal redundancies in an image, as a static scene does not produce any data.

[0009] The document Akrai, M., Margotat, N., Sicard, G., & Fesquet, L. (2021, November). A hybrid event-based pixel for low-power image sensing. In 2021 28th IEEE International Conference on Electronics, Circuits, and Systems (ICECS) (pp. 1-6). IEEE, describes a hybrid system in which a DVS master module activates several TFS sensors. The DVS module analogically measures the luminance variation measured by the photodiodes of the TFS sensors, so as to trigger or not a measurement event by the TFS. This solution, however, remains improvable, particularly in terms of data management and system architecture.

[0010] An object of the present invention is therefore to provide an improved event-driven measurement system. The measurement system may advantageously allow the elimination of spatial and temporal redundancies. The measurement system may further feature an improved architecture.

[0011] The other objects, features and advantages of the present invention will become apparent from an examination of the following description and accompanying drawings. It is understood that other advantages may be incorporated. SUMMARY

[0012] To achieve this objective, according to a first aspect, an event-based measurement system for a physical quantity is envisaged, the system comprising: - a network of sensors, each configured to perform measurement cycles, each cycle comprising: • a measurement of a parameter representative of the physical quantity, from a time t0 until a time t where the measurement reaches a threshold measurement value or a time t corresponding to a determined limit time, and • when the measurement reaches the threshold measurement value or the time limit is reached, a request to read the measurement is sent, - a module called "master module", configured to trigger, by the measurement system, at least one measurement event of the parameter measured by the sensor network.

[0013] Advantageously, the master module comprises at least one sensor, referred to as the "master sensor," selected from among the sensors in the network, and the system comprises, in addition to a logic circuit configured to: - receive at least two read requests from different measurement cycles by at least one master sensor, and determine, for each of the at least two read requests, a measurement time tmes corresponding to the time elapsed between the corresponding instant t0 and instant t, - determine a difference in Atmes between the said measurement times, and • when said difference Atmes is greater in absolute value than a differential threshold value Atmes triggers a measurement event of the parameter measured by the sensor network, • when said difference Atmes is less in absolute value than the differential threshold value Atmes, do not trigger the measurement event.

[0014] The system thus exploits the operating mechanism of TFS-type sensors to eliminate temporal redundancy. The system measures the time required to reach a threshold value, or a limit time. If two measurement times are equivalent, then the physical quantity has not changed substantially. It is then not necessary to immediately repeat the measurement by the sensor network. Conversely, if two measurement times are significantly different, then the physical quantity has changed.

[0015] The use of TFS-type sensors eliminates spatial redundancy by comparing digital measurement time data with analog data, as done by the DVS master module of existing solutions. The use of digital data (for example, AER, short for Address-Event Representation) reduces the volume of data to be processed and simplifies this processing. Analog-to-digital conversion is reduced. Furthermore, the system is less susceptible to noise because the data processed is digital, not analog.

[0016] Furthermore, the use of a logic circuit for comparing measurement times provides greater flexibility in the system architecture. This circuit does not need to be implemented at the sensor level, thus simplifying system integration. This is therefore a significant advantage compared to existing hybrid systems of TFS sensors activated by a DVS-type master module, which require a substantial implementation area.

[0017] The event-driven measurement system can thus exhibit optimized power consumption while offering an improved architecture. Particularly in the case of imagers, the sensor fill factor can be improved by using a logic circuit located remotely from the sensor array. The sensitive part of the system is therefore increased compared to existing solutions.

[0018] A second aspect concerns a method for event-based measurement of a physical quantity, the method comprising: - a first measurement cycle by at least one sensor in a sensor network, called the "master" sensor, comprising: • a measurement of a parameter representative of the physical quantity, from a time t0 until a time t where the measurement reaches a threshold value or a time t corresponding to a determined limit time, • when the measurement reaches the threshold measurement value or the limit time is reached, the sending of a first reading request of the measurement by at least one master sensor to a logic circuit, - a reception, by the logic circuit, of the first reading request, - following the reception of the first reading request, a determination by the logic circuit of a first measurement time tmes (i) corresponding to a time elapsed between time t0 and time t, - a second measurement cycle, at least partially subsequent to the first measurement cycle, and preferably entirely subsequent, by at least one master sensor, comprising: • a measurement of the parameter representing the physical quantity, from a time t0 until a time t where the measurement reaches a threshold value, • When the measurement reaches the threshold value, a second request to read the measurement is sent to the logic circuit. - a reception, via the logic circuit, of the second read request, - following the receipt of the second read request, a determination by the logic circuit of a second measurement time tmes (2) corresponding to a time elapsed between time t0 and time t, - a determination of a difference Atmes between the first measurement time tmes (i) and the second measurement time tmes (2) by the logic circuit, and: • when said difference Atmes is greater in absolute value than a differential threshold value, a measurement event of the parameter by the sensor network, • when said difference Atmes is less in absolute value than a differential threshold value, a non-triggering of the measurement event.

[0019] The method exhibits the effects and advantages described with respect to the system according to the first aspect. In particular, the method reduces temporal redundancy by using digital data through the comparison of measurement times. Power consumption is reduced. Indeed, the activation of the sensor network, excluding the master sensor, can be limited if no change is detected. The system's sensitivity to noise is reduced, as the system handles digital data. BRIEF DESCRIPTION OF THE FIGURES

[0020] The aims, objects, features and advantages of the invention will become clearer from the detailed description of an embodiment thereof, which is illustrated by the following accompanying drawings in which:

[0021] [Fig.1] Fig.1 represents a schematic view of the event measurement system, according to an example embodiment.

[0022] [Fig.2] Fig.2 represents a schematic of a TFS-type sensor, comprising a photodiode, according to an example of implementation.

[0023] [Fig.3] Fig.3 is a diagram representing two measurement cycles by two TFS type sensors, as an example.

[0024] [Fig.4A] Figures 4A and 4B illustrate the sending of read requests by a TFS type sensor matrix, a verification of the requests and the subsequent reset, according to an example embodiment.

[0025] [Fig.4B]

[0026] [Fig. 5] [Fig. 5] shows a diagram of a master module comprising a sensor master and logic circuit, according to an example implementation.

[0027] [Fig.6] Figure [Fig.6] schematically illustrates the principle of event measurement by comparison of measurement times, according to an example of implementation.

[0028] [Fig.7] Fig.7 schematically illustrates the digital measurement circuit enabling the triggering of the measurement event, according to an example implementation.

[0029] [Fig.8] Fig.8 schematically illustrates an event measurement system comprising a master sensor and another sensor, as well as its operation, according to an example embodiment.

[0030] [Fig.9] Fig.9 represents the chronogram of the steps in the measurement process as illustrated in [Fig.8].

[0031] [Fig. 10] The [Fig. 10] schematically illustrates an event measurement system comprising four subgroups of sensors, each comprising a master sensor among 16 sensors, according to an example embodiment.

[0032] [Fig. 11] The [Fig. 11] illustrates an architecture of the TFS circuit (current source, analog comparator, and current amplifier) ​​for use with a photodiode, according to an example embodiment.

[0033] The drawings are given by way of example and are not limiting of the invention. They constitute schematic representations of principle intended to facilitate understanding of the invention and are not necessarily to scale with practical applications. DETAILED DESCRIPTION

[0034] Before beginning a detailed review of embodiments of the invention, optional features which may possibly be used in association or alternatively are stated below.

[0035] According to one example, the logic circuit is at least partially, and preferably entirely, located away from the sensor network. The digital data is generated at the logic circuit level upon receipt of the read request signals. Thus, as mentioned previously, the layout of the logic circuit within the system architecture is made more flexible. Relocating the logic circuit frees up space within the sensor network, and in particular increases the sensor area. In the case of an imager, this increases the sensor coverage ratio of the network. Furthermore, it provides greater flexibility in the placement of the sensors relative to each other within the sensor network.

[0036] According to one example, the system is configured so that the read request, originating from a sensor in the network, is received by the logic circuit. The system includes an addressing module configured to retrieve address information from the sensor that issued the read request within the sensor network, based on the read request. The sensor network can be configured to issue a read request encoded to understand address information from the emitting sensor. The request. This allows the use of TFS sensors to reduce spatial redundancy, particularly through information encoding. The system thus limits both spatial and temporal redundancy by using TFS sensors, and therefore by manipulating digital rather than analog data. In a specific example, the read request passes through several sensors in the network before being received by the logic circuit. The read request can therefore be processed in a grouped manner across different sensors.

[0037] In one example, the sensor network comprises several sensor subgroups, in each of which at least one sensor is a master sensor. The sensor network and the logic circuit are configured so that the measurement event includes at least one measurement cycle for the sensors in the subgroup other than the master sensor, and, in another example, also for the master sensor, the other sensors being previously inactive. Data management can thus be spatially optimized according to the sensor subgroups. The data volume and power consumption can be further reduced, while simplifying information management, thanks to smaller sensor subgroups.

[0038] According to one example, the sensor network forms a sensor matrix extending in at least two dimensions. When the sensors form a two-dimensional matrix, the issuance of a grouped query by column and by row is facilitated. This simplifies the addressing of read requests and thus reduces spatial redundancies.

[0039] According to one example, the sensors of the array, and preferably of the matrix, each comprise at least one acquisition element chosen from among a photodiode, a microbolometer, a radio frequency antenna, or any other sensor, and defining at least part of a pixel. According to one example, the measured parameter is representative of at least one parameter chosen from among luminance L, temperature, a frequency or frequency variation, or any other physical quantity, received by said acquisition element of said pixel. The system is thus an event-driven imager, whose redundancies, at least temporal ones, are reduced. The system is therefore particularly well-suited to image acquisition for autonomous cameras, for example, for video surveillance applications. For example, for an infrared image, temperatures are measured; a radar antenna can measure intensity and / or a frequency variation (for example, by Doppler effect to measure velocity).

[0040] According to one example, the measurement event by the sensor network includes: - a transmission of the second measurement time tmes (2) by the logic circuit, and / or - an additional measurement cycle, by at least one sensor, and preferably several sensors, of the network, the cycle comprising: • a measurement of the parameter representing the physical quantity, from a time t0 until a time t where the measurement reaches a threshold value or a time t corresponding to a determined limit time, and • when the measurement reaches the threshold measurement value or the time limit is reached, a request to read the measurement is sent.

[0041] According to one example, the measurement event comprises a measurement cycle by at least one sensor in the network, different from at least one master sensor, said sensor having previously been inactive. Thus, it is the indication that the physical quantity has changed, given by the master sensor, that triggers a measurement by the other sensor(s). If the physical quantity has not changed, the other sensors being inactive, they do not generate data. The volume of data generated is reduced, and therefore the power consumption is decreased.

[0042] According to one example, the measurement event includes an additional measurement cycle by at least one master sensor.

[0043] According to one example, the sensor network comprises several subgroups of sensors, in which, for each subgroup, one sensor is a master sensor, and the measurement event includes at least one measurement cycle by the sensors in the subgroup other than the master sensor, said other sensors being previously inactive. The activation of the measurement by the other sensors is thus carried out by subgroups of sensors, advantageously located near each other within a subgroup. Data management can thus be spatially optimized. It is therefore possible to activate the sensors only in a part of the network concerned by the change in the physical quantity. The volume of data, and therefore the power consumption, can be further reduced, while still maintaining satisfactory information.

[0044] According to one example, at least several sensors in the network being suitable to be a master sensor, and preferably each sensor in the network being suitable to be a master sensor, the method further includes, prior to the first measurement cycle by at least one master sensor, a selection of at least one master sensor from among the sensors in the network.

[0045] According to one example, each sensor in the network is suitable as a master sensor. Depending on the desired spatial resolution, for example, to obtain a more detailed sensitivity to changes in the physical quantity for a portion of the sensor network, it is possible to select the desired master sensors. The method may also include selecting at least one subgroup associated with a selected master sensor. The modularity of the event-driven measurement is thus improved.

[0046] According to one example, with all sensors in the network being master sensors, the first measurement cycle and the second measurement cycle are performed by all sensors of the network, the measurement event preferably includes a transmission of the second measurement times tmes (2) by the logic circuit, and preferably to the reading module. The method thus offers better spatial resolution, although it generates a larger volume of data.

[0047] According to one example, the method comprises several successive measurement cycles by at least one master sensor, each cycle being followed by the receipt, by the logic circuit, of a corresponding read request, a determination by the logic circuit of a corresponding measurement time tmes, and a determination of the difference Atmes between two measurement times tmes, said determination being made between the measurement times tmes of two successive measurement cycles. The measurement times are thus compared with each other over successive cycles, and preferably directly successive ones. Time redundancies can be reduced without having to save an entire history of measurement times. The method is therefore simplified.

[0048] According to one example, the method further includes adjusting the differential threshold value Atmes_th to modify the sensitivity. Because a comparison of measurement times is performed by a digital logic circuit, the differential threshold value is digital. Thus, threshold management by a processor is simplified. This value can therefore be adjusted in a simplified manner, and the time resolution of the system can therefore be modified more easily.

[0049] According to one example, the sensor network forming a sensor matrix extending in two dimensions and each sensor comprising at least one acquisition element chosen from a photodiode, a microbolometer and a radio frequency antenna, or any other sensor, and defining at least in part a pixel, the measured parameter is preferably representative of the luminance L, or any other physical quantity, received by said acquisition element.

[0050] A parameter "approximately equal to / greater than / less than" a given value means that this parameter is equal to / greater than / less than the given value, to within ±10% of that value. A parameter "approximately between" two given values ​​means that this parameter is at least equal to the smaller of the given values, to within ±10% of that value, and at most equal to the larger of the given values, to within ±10% of that value.

[0051] In this patent application, when two parts are described as distinct, this means that these parts are separate. They may be: - positioned at a distance from each other, and / or - mobile relative to each other and / or - joined together by being fixed by added elements, this fixing being removable or not.

[0052] In this patent application, the term "fixed" used to describe the connection between two parts means that the two parts are linked / fixed to each other with respect to all degrees of freedom, unless explicitly specified otherwise. For example, if it is stated that two parts are fixed in translation along a direction X, this means that the parts can be movable relative to each other, possibly with respect to several degrees of freedom, excluding freedom in translation along the direction X. In other words, if one part is moved along the direction X, the other part moves in the same direction.

[0053] The expression "A and / or B" means (A), (B), or (A and B). The expression "A, B and / or C" means (A), (B), (C), (A and B), (A and C), (B and C), or (A, B and C).

[0054] The event measurement system 1 and the associated measurement method are now described in more detail with reference to the figures, according to several embodiment examples.

[0055] As illustrated, for example, in [Fig. 1], the measurement system 1 comprises an array 10 consisting of several sensors 100. This array 10 can form a two-dimensional matrix 10, in which the sensors 100 are arranged in several rows and several columns. As described in more detail later, a master sensor 110 is chosen from among these sensors 100. The array 10 of sensors 100, 110, and the sensors 100, 110, are hereafter referred to as such when describing elements that may be common between a master sensor 110 and a slave sensor 100. It may be foreseen that these sensors 100, 110 are arranged in a non-matrixed manner. This may be the case, in particular, for systems comprising an array of electrodes, for example, such as those used for electroencephalograms. These sensors may or may not be connected to each other.

[0056] The measuring system 1 further comprises a logic circuit 12, or equivalently a logic module. As described in more detail later, this logic circuit 12 is configured so as to trigger or not trigger a measurement event based on the measurement taken by the sensors 100, 110, in order to limit the temporal redundancy of the system 1. The measuring system 1 may further comprise a readout module 13, for example communicating with the network 10 and the logic circuit 12.

[0057] Before describing the operation of logic circuit 12, the operation of a TFS-type sensor is reviewed. An example of a TFS sensor comprising a photodiode is described in more detail, by way of example, within the context of an imager application. This operation can be transposed to any other type of TFS sensor. In the following, the term "sensor" refers to a TFS sensor.

[0058] As is known, these sensors 100, 110 are configured to perform measurement cycles. With reference to Figures 2 and 3, a sensor 100 comprises a means or element for acquiring the measured physical quantity, in this case a Photodiode 100a, 110a. Note that other acquisition methods can be considered in an imaging application, for example a microbolometer or an RF antenna. During a measurement cycle 2, the measured value can reach a threshold value.

[0059] During measurement 20, the measurement by the acquisition means can evolve, and for example increment, until it reaches a threshold measurement value. In the case of an increment, the measurement by sensor 100, 110 can therefore be integrated. The measured time increments as measurement 20 progresses until the measurement threshold is detected on sensor 100, 110.

[0060] For example, and as illustrated in Figures 2 and 3, the voltage V(1) or V(2) across the photodiode 100a, 110a changes until it reaches the threshold value V*. V* can be set independently to adjust the sensor's sensitivity. A comparator element 100d, 110d can be configured to compare the measured parameter to the threshold measurement value. A readout element 100e, 110e can be configured to send a readout request.

[0061] For other types of TFS sensors, parameters other than voltage, for example, can be monitored. The TFS sensor may, for example, include a microbolometer for which a resistance value is measured.

[0062] Alternatively, a predetermined time limit may be reached without the measurement value reaching the threshold measurement value. Equivalently, when the measurement value, and in particular the integrated measurement value, does not reach the threshold measurement value at the predetermined time limit, a read request 22 may be sent 21 at time t corresponding to the elapsed time limit. Time t may then be equal to t0 + him, where tiim is predetermined.

[0063] Once the measurement has reached the threshold value at time t, or the time limit has been reached, the sensor 100,110 is configured to send a read request 21 of the measurement. A certain amount of time has therefore elapsed since the start of the measurement at time t0. The system 1 is configured to associate a measurement time corresponding to the time elapsed between time L and time t when the read request 22 is sent. The measurement time is representative of the physical quantity being measured. The measurement time of a TFS sensor is a numerical value. For example, and particularly in the case of a photodiode, the shorter this time, the larger the physical quantity being measured. Note that this is not necessarily the case, as for example with a thermistor.

[0064] Following the sending of this request, the sensor 100, 110 can be reset 24 by an action commonly called "reset". The measured value can thus be reset to zero 24. For this purpose, the sensor 100 110 may include an element 100c, 110c configured to receive the reset signal and reset the sensor 100, 110. Each measurement cycle 2 may include a reset following the sending 21 of the request 22.

[0065] Several measurement cycles 2a, 2b can take place successively. As illustrated in [Fig.3], for the voltage V(1) of a first sensor, two measurement cycles 2a, 2b can correspond to two measurement times tmes(1) and tmes(2) respectively. The same applies to the voltage V(2) of a second sensor, whose measurement cycle 2a corresponds to the measurement time tmes(3).

[0066] The read request 22 can be transmitted via any communication device included in the system, for example, a bus. In the case of a network 10 of sensors 100 forming a matrix 10, the read requests 22 can be transmitted via several sensors 100, 110 of the matrix 10, along a row and / or a column. For this purpose, a common power line can be shared by several sensors 100. These requests can be processed by the system 1 in a grouped manner, for example, within a time interval At. This interval can, for example, be counted from the time tx of reception of the previous read request 22 from the sensor. Alternatively, the system can be provided for as a shared bus or network. In this case, the request 22 may not be transmitted via several sensors 100.

[0067] Each read request 22 can be encoded to include address information from the sensor 100, 110 emitting the request 22. Encoding the address information thus reduces spatial redundancies. Note that this reduction of spatial redundancies is not limited to a matrix configuration.

[0068] As illustrated, for example, by Figures 4A and 4B, the sensors shown in dashed lines reached the threshold value within a time interval At. Each sensor therefore sends two read requests 21, one for its row and the other for its column, at the instant the measurement threshold value is reached. The system may include an addressing module 130 configured to retrieve the address of the sensors that sent the requests 22a in the matrix 10 from the read requests, and more specifically from its encoding. For this purpose, for example, the addressing module 130 can cross-reference the read requests 22a received for each row and for each column within the time interval At and deduce which sensors 100, 110 were activated. However, simply crossing the requests would lead to the conclusion that sensor 100, 110 located on the first column and the second row also issued a request 22a.To resolve this ambiguity, an additional step is therefore preferable. This step is designated as request verification 23. After request verification 23, sensors 100 and 110 that have issued a request 22a can be reset by a reset, for example issued by the addressing module 130, or by a local reset on each sensor 100 and 110. The sensor that has not issued a request 22a continues its measurement phase.

[0069] Figures 4A and 4B illustrate this for a second set of sensors 100, 110 which have been activated in a time interval At from an instant ty, this time interval being equal to or different from the previous one.

[0070] Note that this addressing operation can be transposed to non-matrixed sensors, the addressing module being able to be configured to find the address of sensor 100, 110 in network 10.

[0071] To reduce temporal redundancies, the system includes a master module 11 configured to trigger a measurement event 4 of the parameter via the sensor network 10 of sensors 100, 110, illustrated for example in [Fig. 6]. This module is specifically configured so that the measurement event 4 is triggered only when the physical quantity to be measured has changed over time. The change in the physical quantity is obtained, more specifically, based on two time measurements. The master module 11 is therefore configured to take several time measurements in order to evaluate the change in the physical quantity.

[0072] The master module 11 can for this purpose be associated with a logic circuit 12 configured to use the read requests 22 issued by the network 10 of sensors 100, 110 to perform this comparison.

[0073] As illustrated, for example, by Figures 5 to 7, at least one sensor 110 of the array 10 is a master sensor. In other words, a sensor 100 can be chosen to be used as a master sensor 110. All the sensors 100 of the array 10 can be TFS sensors, with one or more sensors 100 being chosen to be a master sensor 110. Preferably, a master sensor 110 can be identical to a sensor 100, with the master sensor 110 being controlled differently.

[0074] The master module 11 can therefore be considered to include the master sensor 110. This master sensor 110 operates in the same way as the previously described TFS sensors. It is therefore also a TFS-type sensor. The logic circuit 12 receives read requests 22a, 22b from different measurement cycles 2a, 2b 20 performed by the master sensor 110. The logic circuit 12 can receive a timestamp information or timestamp signal from the time clock 120, which is a time measurement device.

[0075] The logic circuit 12 determines a measurement time tmes for each of the read requests 22a, 22b issued by the master sensor 110. To do this, the logic circuit 12 is associated with a time clock 120 configured to associate a measurement start time t0 with a time t for the reception of the read request 22a, 22b by the logic circuit 12. The measurement time tmes can thus be determined. The time provided by the time clock 120 can, for example, correspond to the time that has elapsed between the start of the operation, for example the last reset signal 24, and the arrival of the read request 22.

[0076] The logic circuit 12 is further configured to determine 32 any possible difference Atmes between these two measurement times tmes(i), tmes(2), and preferably between the measurement times of two successive measurement cycles 2a, 2b. These cycles are preferably temporally separated and do not overlap.

[0077] When the difference in Atmes is greater in absolute value than a threshold value, called the differential threshold value Atmes, the physical quantity to be measured has therefore changed. In the case of an imager, for example, the luminance has therefore changed between the first cycle 2a and the subsequent cycle 2b. The logic circuit 12 can therefore trigger a measurement event 4.

[0078] When the difference in Atmes is less than the differential threshold value Atmes_th in absolute value, the measured physical quantity is considered not to have changed significantly. This consideration may depend on the fixed differential threshold value, and therefore on the sensitivity of the master sensor 110 determined by this differential threshold value. The logic circuit 12 does not trigger measurement events, which limits temporal redundancy. In this case, system 1 does not generate measurement data and therefore no data stream to update the measurement of the physical quantity.

[0079] In case of equality at the differential threshold value Atmes, one can choose to be in one or the other of the above cases.

[0080] These steps can then be repeated for several measurement cycles 2 of the master sensor 110. It can be foreseen that a master sensor 110 performs measurement cycles periodically, for example to monitor the potential evolution of the physical quantity to be measured.

[0081] The method may comprise a number n of measurement cycles 2 by at least one master sensor 110, n being a non-zero positive integer greater than or equal to 3, each nth cycle comprising: - a measurement of the parameter from time t0 until time t when the parameter reaches the threshold measurement value, or an instant t corresponding to the limit time, - the sending of a read request at time t by at least one of the master sensors 110 to the logic circuit 12.

[0082] Each nth cycle can be followed by the reception 30, by the logic circuit 12, of an nth read request 22 and the determination 1 by the logic circuit 31 of an nth measurement time corresponding to a time elapsed between time t0 and the time t by the time-stamping device 120. The determination 32 of the difference Atmes of the measurement times can be made between successive measurement cycles n-1 and n.

[0083] As illustrated in [Fig. 6] and in more detail in [Fig. 7], the logic circuit 12 can include a memory module 122 and a comparator module 121. The memory module 122 can be configured to store the measurement time of the previous cycle Atmes (i). The comparator module 121 can be configured to determine 32 the difference Atmes and compare 34 this value to the threshold value Atmes

[0084] During subsequent cycles, the measurement time value tmes of a cycle n can be stored in memory 122 and preferably can replace the value of cycle n-1, for comparison with the measurement time of a subsequent cycle n+1. In other words, the measurement time value can be recorded in a shift register 122. The timestamp signal enters the shift register 122. As an example, the measurement time values ​​tmes of several cycles, preferably successive cycles, can be stored in memory 122, for example, to determine the direction of change of the physical quantity being measured. The difference value Atmes can be stored in memory 122, and several difference values ​​Atmes can be compared.

[0085] The memory module 122 can be configured to store a new measurement time value only after the triggering 33 of a measurement event 4. This further reduces the actions performed by the system 1.

[0086] The measurement event 4 can include various combinable or alternative actions. As illustrated, for example, in [Fig. 6], the measurement event 4 can include a transmission 40 of the measurement time tmes of the subsequent cycle by the logic circuit 12, for example, to a read module 13. Since the measurement time tmes is indeed representative of the physical quantity to be measured, this value can therefore be used to update the measurement of the physical quantity.

[0087] As illustrated, for example, in [Fig. 6], the measurement event 4 can include at least one additional measurement cycle 41 by at least one sensor 100, 110, and preferably several sensors of the network 10. This measurement cycle 41 can include the steps described previously in general terms for TFS sensors. All or some of the sensors 100, 110 of the network can perform this measurement cycle 41. Thus, following the determination that the physical quantity has changed, a measurement can be acquired for its sensors 100, 110. Preferably, the measurement event 4 then includes only a single additional measurement cycle 41. Alternatively, the measurement event can include a plurality of additional measurement cycles 41. For this purpose, and as illustrated in [Fig. 7], the logic circuit 12 can output an "ON" event 33, triggering the measurement event 4.

[0088] The measurement event may include, via the logic circuit 12, the emission 35 of an "OFF" event. An "ON" or "OFF" event is an event. It can play the same role. The difference between an "OFF" and an "ON" event is the direction of change of the measured physical quantity, for example, brightness. The "OFF" event can trigger a reset action 24 of the master sensor 110, to initiate a subsequent cycle, like an "ON" event.

[0089] According to an example, it can be foreseen that the sensor network 10 of 100, 110 will perform measurement cycles 41 until it receives an "OFF" event from the logic circuit 12. Alternatively, each measurement cycle 41 can be conditioned on the emission 33 of an "ON" or "OFF" event.

[0090] Due to the more digital nature of the data handled by the logic circuit 12, the logic circuit 12 can be located outside the sensor network 10 of sensors 100, 110. Equivalently, the logic circuit 12 can be located at least partially outside the sensor network 10 of sensors 100, 110. The logic circuit 12 can be located at least partially spatially separate from the sensor network 10 of sensors 100, 110. This offers greater flexibility in the architecture of system 1, and in particular that of the sensor network 10 of sensors 100, 110. It is possible, in particular, to upgrade the architecture of network 10 with more advanced technological nodes because the analog portion is minimized. The logic circuit 12 can, for example, be entirely outside the network 10, with the exception of the connection elements linking it to the network 10 to transmit the data.

[0091] In the specific example of imagers, this remote location of the logic circuit 12 makes it possible to increase the fill factor of the sensor array 10 of 110, 100. This is particularly advantageous compared to existing hybrid imager solutions implementing a DVS-type master module, which requires a differential comparison area of ​​two successive measurements, implemented analogically, on the sensor array. Due to digital processing and particularly in synergy with the remote logic circuit, the fill factor of the sensor array can be increased compared to existing solutions. The fill factor of the photodiodes 100a, 110a is, for example, greater than that of an imager with a DVS-type pixel integrated into the array, since the comparison is no longer performed at the pixel level.The fill factor is not degraded by the addition of the DVS function, considering a given sensor with its own fill factor. The fill factor is more specifically taken as the ratio of the sensitive area of ​​sensors 100, 110 of the array 10 to the total area of ​​the array 10. For example, this can correspond to the area occupied by photodiodes 100a, 110a relative to the total area of ​​the matrix 10 or the pixel.

[0092] Since the threshold value is numerical, it is also possible to adjust the threshold 6 very simply. This allows for greater or lesser sensitivity to variations in the measured quantity.

[0093] It can be anticipated that the evaluation frequency of the master sensor 110 is adjusted by controlling the zeroing or reset frequency 24 of the master sensor 110. For this purpose, the time limit for performing the measurement cycle can be adapted.

[0094] The threshold measurement value for triggering the sending of read requests can also be adjusted, and this for all or part of the sensors 100, 110. The lower this threshold measurement value, the more activations there are of the sensor(s) 100 slaved to each master sensor 110.

[0095] The dynamics of system 1 can be modified according to the step size and depth of the time clock 120. The threshold, depth, and / or resolution of the time clock 120 can be dynamically adjusted as needed. The temporal resolution can be improved, for example, by choosing a smaller step size for the time clock 120.

[0096] It can be anticipated that these adjustments 6 will be carried out in a differentiated or non-differentiated manner between the master sensors 110.

[0097] In one example, the network 10 includes several sensors among which: - at least one sensor is a master sensor 110, - the other sensors 100 are non-master TFS sensors.

[0098] Preferably, before the triggering 33 of the measurement event 4 by the logic circuit 12, the other non-master sensors 100 are inactive. By inactive, it is understood that they do not perform a measurement cycle 2 whose data is sent to the reading system 13. For example, these other sensors 100 do not perform a measurement cycle before receiving an "ON" or "OFF" event from the logic circuit 12. By "previously inactive", it is more specifically understood that these sensors 100 have been inactive since the start of the operation of the system 1 or since the last "ON" or "OFF" event.

[0099] As illustrated, for example, in [Fig. 8], a sensor can act as the master sensor 110 as previously described. When the logic circuit 12 outputs an "ON" event 33 or an "OFF" event 35, this same sensor and one or more additional sensors can act as sensors 100 and perform an additional measurement cycle 41. In practice, it is not necessary to repeat the measurement on the master sensor 110, which can already provide a value, for example, through the second measurement time. However, it is advantageous to perform a data acquisition with the sensors 100 surrounding it, which can be controlled by the master sensor 110.

[0100] According to an example, several sensors 100 of the network 10 are suitable to be a master sensor 110. The system 1 can be configured so as to be able to select the sensor or sensors, from the network 10, playing the role of master sensor 110.

[0101] The master sensor 110 can receive signals 5, and in particular "start" signals 50 and "stop" signals 51. These signals allow control of the reset signal of the master sensor 110. The "stop" signal 51 allows stopping the measurement cycle of the master sensor 110. These signals can originate from or pass through the logic circuit 12.

[0102] An example of operation is described with reference to [Fig. 9]. Once the master function is activated by the Start signal, the time clock 120 increments until the voltage across the photodiode reaches the value Vth or the time clock 120 reaches its highest value, i.e., the limit time. When one of these two conditions is met, the corresponding measurement time value tmes (here, 19) can be stored in memory module 122 and a new Start signal can be issued, accompanied by a reset 24 of the master sensor 110 in order to then perform a second luminance measurement by the master sensor 110. The new data (here, 12) is then inserted into memory module 122 in a shift register, so that the comparison can be made. If the difference in measurement times Atmes is greater than the threshold value Atmes (here, 3), a measurement event 4 can be triggered 33 by the logic circuit 12 (ON event).For this purpose, a reset signal 24 can be sent to the sensors 100. These are then used as simple TFS sensors and the Stop signal 51 can be kept at 1 so as not to activate the master function, for the duration of the luminance measurement L. A new measurement cycle 41 can thus be carried out by the sensors 100.

[0103] The number of usable TFS pixels can be adjusted according to requirements. Thus, a master sensor 110 can easily control any number of sensors 100 or 110. The master sensor 110 can be used in simple TFS mode (i.e., without using the master function). Therefore, particularly for imaging applications, the tiling of space is complete in the matrix 10. Holes in the image can thus be avoided, unlike most existing solutions combining TFS and DVS or DVS and CIS pixels (CMOS image sensor, short for CMOS Image Sensors, the standard sensors for imagers).

[0104] The network 10 can comprise several subgroups 10a, 10b, 10c, lOd of sensors 100, 110. Each subgroup 10a, 10b, 10c, lOd can define a portion of space within the network 10, these portions being separated from each other. Each subgroup of sensors 100, 110 can therefore be spatially separated from the other groups. The subgroups are, for example, juxtaposed.

[0105] At least one sensor among these subgroups is a master sensor 110. The other sensors may be simple TFS 100 sensors. Preferably, the sensors within the same subgroup are neighbors and preferably directly neighbors at least one to one.

[0106] As illustrated, for example, in [Fig. 10], the array 10 comprises four subgroups 10a, 10b, 10c, 10d of sensors 100, 110, each consisting of sixteen TFS sensors. Within each subgroup, one of these sixteen TFS sensors is used as the master sensor 110. This approach reduces the activity of system 1, as most of the sensors 100 are inactive when the physical quantity does not change significantly, for example, when the scene is stationary in the case of an imager. Furthermore, it allows for differentiated monitoring of the physical quantity's evolution between different areas of the array 10 of sensors 100, 110. Thus, it is possible to considerably reduce the energy consumption of system 1.

[0107] By way of example, each sensor in network 10, and more specifically each sensor within a subgroup, is suitable as a master sensor 110. This sensor 110 can therefore be selected from among the sensors in network 10 or from subgroup 10a, 10b, 10c, 10d. When the master module 11, comprising the master sensor 110 and the logic circuit 12, detects a change in the physical quantity, the logic circuit 12 can trigger the measurement by the sensors 100 of the subgroup, and preferably by itself. As seen previously, the master sensor 110 is indeed a fully functional TFS sensor. By way of example, each sensor 100, 110 in network 10 can be selected as a master sensor 110.

[0108] From the preceding description, it is understood that system 1 can operate in different modes: - a simple TFS mode in which no master sensor is selected, - a fully master mode in which each sensor is a master sensor 110, - a hybrid mode in which the operation of certain TFS type sensors 100 is controlled by the detection of a change by the master sensor(s) 110 and the logic circuit 12.

[0109] The master function can be activated on demand.

[0110] By way of example, [Fig. 11] illustrates in more detail a TFS pixel architecture for use with a photodiode. According to the example shown, it comprises a current source 123, an analog comparator module 121 and a current amplifier 124.

[0111] The invention is not limited to the embodiments described above and extends to all embodiments covered by the invention. The present invention is not limited to the examples described above. Many other embodiments are possible, for example by combining features described above, without departing from the scope of the invention. In the context of an imager requirement, the system 1 can be configured to measure luminance in the visible or infrared range, for example. The TFS sensor can be adapted according to the wavelength of the radiation being measured.

[0112] The examples described above can be transposed to applications other than imagers. In general, any TFS-type sensor network system can exhibit the characteristics described above. This can notably be the case for systems comprising a sensor array. The characteristics described above can also be implemented with these sensor networks organized in a non-matrix fashion, particularly insofar as the system is configured to retrieve the sensor addresses. This could, for example, be an electrode array like those used for electroencephalograms. For instance, a network 10 with a tensor or toroidal structure could be used. The query encoding and addressing can be adapted accordingly. A read request 22 could, for example, include a sensor number 100, 110 in the network 10 via a serial link.

[0113] Furthermore, the features described in relation to one aspect of the invention can be combined with another aspect of the invention.

Claims

Demands

1. System (1) for event-based measurement of a physical quantity, system (1) comprising: • a network (10) of sensors called "TFS type" (100), each configured to perform measurement cycles (2), each cycle (2) comprising: • a measurement (20) of a parameter representative of the physical quantity, from a time t0 until a time t where the measurement (20) reaches a threshold measurement value or a time t corresponding to a determined limit time, and • when the measurement (20) reaches the threshold measurement value or the time limit is reached, the sending (21) of a read request (22) of the measurement (20) by said TFS type sensor (100), • a module called the “master module” (11), configured to trigger, via the measurement system (1), at least one measurement event (4) of the parameter measured by the network (10) of TFS type sensors (100), Characterized in that the master module (11) comprises at least one sensor called the "master sensor" (110) selected from among the TFS type sensors (100) of the network (10) and in that the system (1) further comprises a logic circuit (12) configured such that: • receive (30) at least two read requests (22) from different measurement cycles (2) and issued by at least one master sensor (110), and determine (31), for each of the at least two read requests (22), a measurement time tmes corresponding to the time elapsed between the corresponding instant to and instant t, • determine (32) a difference Atmes between said measurement times t pf lmes? O when said difference Atmes is greater in absolute value than a differential threshold value Atmes trigger (33) a measurement event (4) of the parameter measured by the network (10) of type TFS sensors (100), O when said difference Atmes is less in absolute value than the differential threshold value Atmes do not trigger the measurement event (4).

2. System (1) according to the preceding claim, wherein the logic circuit (12) is at least partly, and preferably entirely, remote from the network (10) of TFS type sensors (100).

3. System (1) according to any one of the preceding claims, configured so that the read request (22), originating from a TFS type sensor (100, 110) of the network (10), is received by the logic circuit (12), the system (1) comprising an addressing module (130) configured to retrieve address information from the TFS type sensor (100, 110) that issued the read request (22) in the network (10) of TFS type sensors (100), from the read request (22).

4. System (1) according to any one of the preceding claims, wherein the network (10) of TFS type sensors (100) comprises several subgroups (10a, 10b, 10c, lOd) of TFS type sensors (100), in each of which at least one TFS type sensor is a master sensor (110), the network (10) of TFS type sensors (100) and the logic circuit (12) being configured such that the measurement event (4) comprises at least one measurement cycle (2) for the TFS type sensors (100) of the subgroup (10a, 10b, 10c, lOd) other than the master sensor (110), said other TFS type sensors (100) being previously inactive.

5. System (1) according to any one of the preceding claims, wherein the network (10) of TFS type sensors (100) forms a matrix (10) of sensors (100) extending in two dimensions.

6. System (1) according to any one of the preceding claims, wherein the TFS type sensors (100, 110) of the array (10) each comprise at least one acquisition element (100a, 110a) selected from a photodiode, a microbolometer, a radio frequency antenna, and defining at least part of a pixel.

7. A method for measuring a physical quantity based on an event, the method comprising: • a first measurement cycle (2a) by at least one sensor, called the "master sensor" (110), of a network (10) of sensors called "TFS type" (100), the first cycle (2a) comprising: • a measurement (20) of a parameter representative of the physical quantity, from a time t0 until a time t where the measurement (20) reaches a threshold measurement value or a time t corresponding to a determined limit time, • when the measurement (20) reaches the threshold measurement value or the time limit is reached, the sending (21) of a first read request (22) of the measurement (20) by at least one master sensor (110) to a logic circuit (12), • a reception (30), by the logic circuit (12), of the first read request (22), • following the reception (30) of the first read request (22), a determination (31) by the logic circuit (12) of a first measurement time tmes (^corresponding to a time elapsed between the instant t0 and the instant t, • a second measurement cycle (2b), at least partly subsequent to the first measurement cycle (2a), by at least one master sensor (110), comprising: • a measurement (20) of the parameter representing the physical quantity, from a time t0 until a time t where the measurement (20) reaches a threshold measurement value or a time t corresponding to a determined limit time, • when the measurement (20) reaches the threshold measurement value or the time limit is reached, a second reading request (21) is sent to the logic circuit (12) to read (22) the measurement. • a reception (30), by the logic circuit (12), of the second read request (22),

8.

9. • following the reception (30) of the second read request (22), a determination (31) by the logic circuit (12) of a second measurement time tmes (2) corresponding to a time elapsed between time t0 and time t, • a determination (32) of a difference Atmes between the first measurement time tmes (i) and the second measurement time tmes (2) by the logic circuit (12), and: • when said difference Atmes is greater in absolute value than a differential threshold value Atmes, a measurement event (4) of the parameter is triggered by the network (10) of sensors (100, 110), • when said difference Atmes is less in absolute value than a differential threshold value Atmes a non-triggering of the measurement event (4). A method according to the preceding claim, wherein the measurement event (4) by the array (10) of TFS-type sensors (100) comprises: • a transmission (40) of the second measurement time tmes (2) by the logic circuit (12), and / or • an additional measurement cycle (41), by at least one TFS type sensor (100, 110) of the network (10), the cycle comprising: • a measurement (20) of the parameter representing the physical quantity, from a time t0 until a time t where the measurement (20) reaches a threshold measurement value or a time t corresponding to a determined limit time, and • when the measurement (20) reaches the threshold measurement value or the time limit is reached, sending (21) a read request (22) for the measurement. Method according to the preceding claim, wherein the measurement event (4) comprises a measurement cycle (41) by at least one TFS type sensor (100) of the network (10), different from at least one master sensor (110), said sensor (100) being previously inactive.

10. A method according to the preceding claim, wherein the network (10) of TFS type sensors (100) comprises several subgroups (10a, 10b, 10c, lOd) of TFS type sensors (100), in which, for each subgroup (10a, 10b, 10c, lOd), one TFS type sensor is a master sensor (110), and the measurement event (4) comprises at least one measurement cycle (41) by the TFS type sensors (100) of the subgroup (10a, 10b, 10c, lOd) other than the master sensor (110), said other sensors (100) being previously inactive.

11. A method according to any one of the four preceding claims, wherein, at least several TFS type sensors (100) of the network (10) being suitable to be a master sensor (110), the method further comprises, prior to the first measurement cycle (2a) by the at least one master sensor (110), a selection (5) of the at least one master sensor (110) from among the TFS type sensors (100) of the network (10).

12. A method according to any one of claims 7 and 8, wherein, all TFS type sensors (100) of the network (10) being master sensors (110), the first measurement cycle (2a) and the second measurement cycle (2b) are performed by all TFS type sensors (110) of the network (10), the measurement event (4) comprising a transmission (40) of the second measurement times (2) by the logic circuit (12).

13. A method according to any one of the six preceding claims, the method comprising several successive measurement cycles (2) by at least one master sensor (110), each cycle (2) being followed by the reception (30), by the logic circuit (12), of a corresponding read request (22), a determination (31) by the logic circuit (12) of a corresponding measurement time tmes, and a determination (32) of the difference Atmes between two measurement times tmes, said determination (32) is made between the measurement times tmes of two successive measurement cycles (2).

14. A method according to any one of the seven preceding claims, further comprising an adjustment (6) of the differential threshold value Atmes lhdc so as to modify the sensitivity.

15. A method according to any one of the eight preceding claims, wherein the network (10) of sensors (100, 110) forming a matrix (10) of TFS-type sensors (100, 110) extending according at least two dimensions and each TFS type sensor (100, 110) comprising at least one acquisition element (100a, 110a) selected from a photodiode, a microbolometer and a radio frequency antenna, and defining at least part of a pixel, the measured parameter is representative of the luminance L received by said acquisition element (100a, 110a).

Images