Facing direction detection

Abstract

A method of facing direction and fall detection using a lighting system includes receiving, by a first luminaire and a second luminaire, a sound produced by a person in an area. The method further includes determining spatial directivity signatures of the sound based on sound levels of the sound as received by at least the first luminaire and the second luminaire, where the sound levels of the sound are at at least two frequencies of the sound. The method also includes estimating a facing direction of the person based on the spatial directivity signatures of the sound by executing an artificial intelligence (AI) model trained using spatial directivity signature data labelled with facing directions.

Description

[0001] Facing direction detection

[0002] TECHNICAL FIELD

[0003] The present disclosure relates generally to lighting system integrated facing direction detection, and more particularly to estimating facing directions of a person based on sounds produced by the person.

[0004] BACKGROUND

[0005] The facing direction of a person is an important ingredient for lighting control as well as other applications including smart home systems. In addition, facing direction information can help in understanding interactions between people. In retail applications, facing direction information can indicate which shelf and / or product a person is facing. Facing direction detection is also relevant in surveillance systems, computer games, car driver alertness monitoring, assistive technologies, elderly care, and suspicious behavior detection. Optical cameras are often used to detect the facing directions of people. Cameras may also be used to detect whether a person has fallen down. However, privacy concerns may limit the use of cameras in some applications. In some cases, wearable devices are used for fall detection. However, compliance, for example, by the elderly with use of wearable devices may pause challenges to the use of wearable devices to detect falls. Thus, a solution for estimating facing directions of a person with minimal or no privacy and compliance concerns and using the facing direction information for purposes such as fall detection may be desirable.

[0006] SUMMARY

[0007] The present disclosure relates generally to lighting system integrated facing direction detection, and more particularly to estimating facing directions of a person based on sounds produced by the person. In an example embodiment, a method of facing direction detection using a lighting system includes receiving, by a first luminaire and a second luminaire, a sound produced by a person in an area. The method further includes determining spatial directivity signatures of the sound based on sound levels of the sound as received by at least the first luminaire and the second luminaire, where the sound levels of the sound are at at least two frequencies of the sound. The method also includes estimating a facing direction of the person based on the spatial directivity signatures of the sound by executing an artificial intelligence (Al) model trained using spatial directivity signature data labelled with facing directions.

[0008] In another example embodiment, a lighting system for facing direction detection includes a first luminaire that includes a first processor and a first microphone, where the first luminaire is configured to receive a sound produced by a person in an area via the first microphone. The lighting system is configured to determine a first spatial directivity signature of the sound based on first sound levels of the sound at at least two frequencies of the sound as received by the first luminaire. The system further includes a second luminaire that includes a second processor and a second microphone, where the second luminaire is configured to receive the sound produced by the person via the second microphone. The lighting system is configured to determine a second spatial directivity signature of the sound based on second sound levels of the sound at the at least two frequencies of the sound as received by the second luminaire. The lighting system is configured to execute an Al model to estimate a facing direction of the person based on at least the first spatial directivity signature and the second spatial directivity signature.

[0009] These and other aspects, objects, features, and embodiments will be apparent from the following description and the appended claims.

[0010] BRIEF DESCRIPTION OF THE FIGURES

[0011] Reference will now be made to the accompanying drawings, which are not necessarily drawn to scale, and wherein:

[0012] Fig. 1 illustrates a facing direction detection lighting system according to an example embodiment;

[0013] Fig. 2 illustrates a graph of horizontal plane sound levels of frequency components of a sound emitted by the person of FIG. 1 according to an example embodiment;

[0014] Fig. 3 illustrates a graph of vertical plane sound levels of frequency components of a sound emitted by the person of FIG. 1 according to an example embodiment;

[0015] Fig. 4 illustrates a top view of facing direction regions of the person of FIG. 1 in a horizontal plane according to an example embodiment; Fig. 5 illustrates a side view of facing direction regions of the person of FIG. 1 in a vertical plane according to an example embodiment;

[0016] Fig. 6 illustrates facing directions of the person of FIG. 1 during a falling incident in the area of FIG. 1 according to an example embodiment;

[0017] Fig. 7 illustrates a block diagram of a luminaire corresponding to the luminaires of the system of FIG. 1 according to an example embodiment; and

[0018] Fig. 8 illustrates a method of facing direction detection according to an example embodiment.

[0019] The drawings illustrate only example embodiments and are therefore not to be considered limiting in scope. The elements and features shown in the drawings are not necessarily to scale, emphasis instead being placed upon clearly illustrating the principles of the example embodiments. Additionally, certain dimensions or placements may be exaggerated to help visually convey such principles. In the drawings, the same reference numerals used in different drawings may designate like or corresponding but not necessarily identical elements.

[0020] DETAILED DESCRIPTION OF THE EXAMPLE EMBODIMENTS

[0021] In the following paragraphs, example embodiments will be described in further detail with reference to the figures. In the description, well known components, methods, and / or processing techniques are omitted or briefly described. Furthermore, reference to various feature(s) of the embodiments is not to suggest that all embodiments must include the referenced feature(s).

[0022] FIG. 1 illustrates a facing direction detection lighting system 100 according to an example embodiment. In some example embodiments, the system 100 includes luminaire 102, 104, 106, 108, 110. The luminaires 102-110 may be in an area 114, such as a room. For example, the area 114 may be a room in an elderly care facility, a store, etc. The luminaires 102-108 may be, for example, recessed luminaires that are installed in a ceiling 116 of the area 114. The luminaire 110 may be a table lamp or another type of luminaire that is located, for example, on a piece of furniture 118. For example, the luminaire 110 may be a floor luminaire that is on a floor 124 of the area 114 instead of on the furniture 118.

[0023] In some example embodiments, each one of the luminaires 102-110 may include a processor, microphones, and an occupancy sensor. The light module of each one of the luminaires 102-110 may emit the light provided by the particular luminaire. Each one of the luminaires 102-110 may receive sounds via respective microphones, and the processor of each one of the luminaires 102-110 may process the sound received via the respective microphones.

[0024] In some example embodiments, the system 100 may detect one or more persons such as a person 120 in the area 114. For example, one or more of the luminaires 102-110 may include an occupancy sensor. To illustrate, the luminaires 102-110 may each include a passive infrared (PIR) sensor and / or a thermopile (e.g., a single pixel thermopile) that can be used to detect the person 120 in the area 114 as can be readily understood by those of ordinary skill in the art with the benefit of this disclosure. For example, the processor of the luminaire 102 may detect the person 120 based on detection signals from an occupancy sensor of the luminaire 102. The luminaires 102-110 may also estimate the location ofthe person 120. For example, the luminaires 102-110 may estimate the location of the person in the area 114 using a respective thermopile, where the sensed temperature of the person 120 depends on the location of the person 120 relative to the luminaires 102-110. For example, because the locations of the luminaires 102-110 in the area 114 can be known, the different sensed temperature values of the person 120 as sensed by the respective thermopile of the luminaires 102-110 may be used to estimate the location of the person 120 in the area 114. To illustrate, one of the luminaires 102-110 or a server 112 of the system 100 may receive information from the other luminaires and estimate the location of the person 120. Alternatively or in addition, a server 112 of the system 100 may receive information from the luminaires 102-110 and estimate the location of the person 120. After estimating the location of the person 120, Hidden Markov Model (HMM) can be applied to frack the person 120 as the person 120 moves around in the area 114. For example, the processor of one of the luminaires 102-110 or the server 112 may execute an HMM module to frack the person 120 in the area 114. The server 112 may operate as an edge device that interacts with a cloud server.

[0025] In some example embodiments, the system 100 may determine the facing direction of the person 120. The facing direction of the person 120 as used herein refers to the face of the person 120. For example, the system 100 may determine the facing direction of the person 120 based on one or more sounds emitted by the person 120. To illustrate, the luminaires 102-110 may determine spatial directivity signatures of a sound 130 emitted by the person 120, and the facing direction of the person 120 may be determined by one or more of the luminaires 102-110 and / or by the server 112 based on the spatial directivity signatures determined by two or more of the luminaires 102-110. The facing direction of the person 120 may account for both horizontal and vertical directions. For example, the person 120 may have a facing direction of front-up, front-forward, front-down, right-up, etc., where, for example, front, right, left, and back are designated horizontal direction regions that together form 360 degrees around the person 120 and where up, forward, and down are designated vertical direction regions relative to the face of the person 120.

[0026] In general, facing directions of the person 120 may be user defined. For example, the facing direction of the person 120 may be defined with respect to structures of a generic space (e.g., a room) that may be representative of other spaces of interest. To illustrate, the facing direction of the person 120 may be a front-wall 126 of the area 114. As another example, the facing direction of the person 120 may be the front-wall 126 and the ceiling 116 when the front of the person 120 is facing the front-wall 126 and the face of the person 120 is facing the ceiling 116. As yet another example, the facing direction of the person 120 may be the front-wall 126 and the floor 124 of the area 114 when the front of the person 120 is facing the front-wall 126 and the face of the person 120 is facing the floor 124. As yet another example, the person 120 may face the back-wall 128 and the floor 124 such that the facing direction of the person 120 is the back-wall 128 and the floor 124. In some cases, the facing directions of the person 120 may be defined with respect to objects such as a shelf 132, where, for example, the possible facing directions of the person 120 may be the top shelf level, the middle shelf level, and the bottom shelf level of the shelf 132.

[0027] In some example embodiments, facing directions of the person 120 may be defined more generically using azimuth angles and vertical angles (i.e., elevation and depression angles) or ranges of azimuth and vertical angles. In general, the complete azimuth angle ranges from 0 to 360 degrees with, for example, 0 degree corresponding to north, 90 degrees corresponding to east, 180 degrees corresponding to south, and 270 degrees corresponding to west. Vertical angles range between 90 degrees and -90 degrees from a horizontal plane.

[0028] In some example embodiments, the luminaires 102-110 may determine the spatial directivity signatures of sounds such as the sound 130 emitted by the person 120. A spatial directivity signature of the sound 130 determined by a particular one of the luminaires 102-110 may be defined by the relationship between, for example, sound levels (i.e., audio signal strengths or signal power) of the sound 130 at two or more frequency components of the sound 130 as received by the particular luminaire. For example, the spatial directivity signature of the sound 130 determined by the luminaire 102 may be defined by the relationship between the sound levels of the sound 130 at two frequency components of the sound 130 as received by the luminaire 102. The relationship between the sound levels may be expressed, for example, as a difference between the two sound levels, a ratio of the smaller sound level to the larger sound level, or another expression as can be readily understood by those of ordinary skill in the art with the benefit of this disclosure. As another example, the spatial directivity signature of the sound 130 determined by the luminaire 102 may be defined by the relationship between the sound levels of the sound 130 at a first frequency component and a second frequency component of the sound 130 as received by the luminaire 102 and also by the relationship between the sound levels of the sound 130, for example, at the first frequency component or the second frequency component and a third frequency component of the sound 130 as the sound 130 is received at the luminaire 102.

[0029] FIG. 2 illustrates a graph 200 of horizontal plane sound levels of frequency components of a sound, such as the sound 130, emitted by the person 120 of FIG. 1 according to an example embodiment. Referring to FIGS. 1 and 2, in some example embodiments, different frequency components of the sound 130 have different horizontal spatial directivity. For example, the 1 KHz and the 8 KHz frequency components of the sound 130 have different horizontal spatial directivity that depend on the horizontal facing direction of the person 120. To illustrate, with the person 120 facing toward the horizontal plane direction labelled Front in FIG. 2, the sound level at the 8 KHz frequency component of the sound 130 at a location directly in front of the person 120 is significantly higher than at a location at the back side of the person 120. That is, a microphone located in front of the person 120 may readily receive the 8 KHz frequency component of the sound 130 while the 8 KHz frequency component may be too faint to be received by a microphone located at the back side of the person 120, for example, in the horizontal plane direction labelled Back in FIG. 2. In contrast to the 8 KHz frequency component, microphones both in front and behind the person 120 may readily receive the 1 KHz frequency component of the sound 130 although at possibly different sound levels.

[0030] In some example embodiments, the similarities and differences in such horizontal spatial directivities of the 1 KHz and 8 KHz frequency components of the sound 130 relative to microphones at different locations in the area 114 may define the spatial directivity signature of the sound 130 that can be used to estimate or otherwise determine the horizontal facing direction of the person 120. That is, the spatial directivity signature of the sound 130 defined based on similarities and differences in the horizontal spatial directivity of the 1 KHz frequency component and the 8 KHz frequency component of the sound 130 may be utilized to determine the horizontal facing direction of the person 120. As another example, a spatial directivity signature of the sound 130 that is defined based on similarities and differences in the horizontal spatial directivities of the 1 KHz frequency component and the 4 KHz frequency component of the sound 130 may be utilized to determine the horizontal facing direction of the person 120. As yet another example, a spatial directivity signature of the sound 130 defined based on similarities and differences in the horizontal spatial directivities of the 1 KHz frequency component and the 4 KHz frequency component of the sound 130 as well as based on similarities and differences in the horizontal spatial directivities of the 1 KHz frequency component and the 8 KHz frequency component of the sound 130 may be utilized to estimate or otherwise determine the horizontal facing direction of the person 120. In general, similarities and differences in the horizontal spatial directivities of different frequency components of the sound 130 may be utilized to estimate or otherwise determine the horizontal facing direction of the person 120.

[0031] FIG. 3 illustrates a graph 300 of vertical plane sound levels of frequency components of a sound, such as the sound 130, emitted by the person of FIG. 1 according to an example embodiment. Referring to FIGS. 1 and 3, in some example embodiments, different frequency components of the sound 130 have different vertical spatial directivity. For example, the 1 KHz and the 8 KHz frequency components of the sound 130 have different vertical spatial directivity that depend on the vertical facing direction of the person 120. To illustrate, with the person 120 facing toward zero degree in the vertical plane direction shown in FIG. 3, the sound level at the 8 KHz frequency component of the sound 130 at a location in front of the person 120 (e.g., at 0 degree in the vertical plane) is significantly higher than at a location at the back side of the person 120 (e.g., at 180 degrees in the vertical plane). That is, a microphone located in front of the person 120 may readily receive the 8 KHz frequency component of the sound 130 while the 8 KHz frequency component may be too faint to be adequately received by a microphone located at the back side of the person 120 (e.g., at 180 degrees in the vertical plane). In contrast to the 8 KHz frequency component, microphones both in front and behind the person 120 may readily receive the 1 KHz frequency component of the sound 130 although at possibly different sound levels.

[0032] In some example embodiments, the similarities and differences in such vertical spatial directivities of the 1 KHz and 8 KHz frequency components of the sound 130 relative to microphones at different locations in the area 114 may define the spatial directivity signature of the sound 130 that can be used to estimate or otherwise determine the facing direction of the person 120. That is, the spatial directivity signature of the sound 130 defined based on similarities and differences in the vertical spatial directivity of the 1 KHz frequency component and the 8 KHz frequency component of the sound 130 may be utilized to determine the vertical facing direction of the person 120. As another example, a spatial directivity signature of the sound 130 that is defined based on similarities and differences in the vertical spatial directivities of the 1 KHz frequency component and the 4 KHz frequency component of the sound 130 may be utilized to determine the vertical facing direction of the person 120. As yet another example, a spatial directivity signature of the sound 130 defined based on similarities and differences in the vertical spatial directivities of the 1 KHz frequency component and the 4 KHz frequency component of the sound 130 as well as based on similarities and differences in the vertical spatial directivities of the 1 KHz frequency component and the 8 KHz frequency component of the sound 130 may be utilized to estimate or otherwise determine the facing direction of the person 120. In general, similarities and differences in the vertical spatial directivities of different frequency components of the sound 130 may be utilized to estimate or otherwise determine the vertical facing direction of the person 120.

[0033] In some example embodiments, a frequency domain analysis of the sound 130 may be performed to determine the spatial directivity signatures of the sound 130. To illustrate, a time domain representation of the sound 130 is shown in Equation (1). -( = / ifc (tMtH eit)) * n*(t) Equation (1)

[0034] In Equation (1), Prefers to a particular microphone andjAt) is the audio signal strength (e.g., amplitude) of the sound 130 as received by a microphone k, such as a microphone of one of the luminaires 102-110 shown in FIG. 1. hk is the channel state information and includes audio fading due to the distance, m is noise at the microphone k. 0(t) and <|>(t) are arrival angles of the sound 130 at the microphone k and may correspond to the azimuth (horizontal) angle and elevation or depression (vertical) angle of the facing direction of the person 120, respectively.

[0035] Equation (2) provides the frequency domain representation of the sound 130 that is a Fast Fourier Transform (FFT) of the time domain representation of the sound 130 given in Equation (1), where / m represents individual frequencies of the sound 130. ( / ;„) = FFT(yfc(t))fm = 0,1, 2, ... Equation (2) For example, / / ) (i.e., m=0) may be a base or reference frequency of 100 Hz. Each luminaire of the luminaires 102-110 shown in FIG. 1 may include one or more microphones and a processor and may determine the audio signal strength at frequencies of interest such as 100 Hz, 1 KHz, 2 KHz, 4 KHz, and 8 KHz that each correspond to a respective value of m in Equation (2). Each luminaire may also determine one or more spatial directivity signatures of the sound 130 based on the audio signal strengths at two or more of frequencies of interest as described above.

[0036] In some example embodiments, to minimize the effect of sound level variations of sounds emitted by different people or even by the same person in the area 114, the audio signal strengths of the sounds, as received by two or more of the luminaires 102- 110, may be ‘normalized’ based on the respective audio signal strength of the sounds at reference frequency 0. To illustrate, the audio signal strengths of the sound 130, as received by two or more of the luminaires 102-110, at frequencies of interest, such as 1 KHz, 2 KHz, 4 KHz, and / or 8 KHz, may be ‘normalized’ using the audio signal strength of the sound 130 at the reference frequency 0 such as, for example, 100 Hz. Alternatively, an audio signal strength at another frequency of the sound 130 instead of 100 Hz may be used. As used herein, ‘normalized’ refers to the audio signal strengths of the sound 130 (or other sounds) represented by Equation (2) at particular frequencies such as, for example, 1 KHz, 2 KHz, 4 KHz, and 8 KHz, being divided by the audio signal strength of the sound 130 (or the other respective sounds) at the reference frequency 0 (e.g., 100 Hz) as shown in Equation (3). t — yfcCfe) tsW ~ T Equation (3) k77WT

[0037] In Equation (3), represents the ‘normalized’ audio signal strength at a frequency of the sound 130 labelled as fin, where different values of m refer to different frequencies. For example, fm may be 1 KHz, 2 KHz, 4 KHz, 8 KHz where each frequency corresponds to a different value of m in Equation (3). Each luminaire of the luminaires 102- 110 that receives the sound 130 may determine one or more spatial directivity signatures of the sound 130 based on the ‘normalized’ audio signal strengths of the sound 130, for example, at 1 KHz and 4 KHz, at 1 KHz and 8 KHz, and / or at other combinations of frequencies as described above.

[0038] In some example embodiments, one or more of the luminaires 102-110 and / or the server 112 may execute an artificial intelligence (Al) model to estimate the facing direction of the person 120 based on the sound 130. In particular, one or more of the luminaires 102-110 and / or the server 112 may execute an Al model to estimate the facing direction of the person 120 based on spatial directivity signatures from two or more of the luminaires 102-110 that receive the sound 130. To illustrate, an Al model may be trained using audio signal strength data. For example, the training data may include audio signal strengths of sounds as received by microphones (i.e., microphones of luminaires) in an area (e.g., a room), where the audio signal strengths are with respect to a number of frequencies. The number of frequencies may be M, where M is 2 or more. For example, the frequencies may be 1 KHz and 4 KHz; 1 KHz and 8 KHz; 1 KHz, 4 KHz, and 8 KHz; or 1 KHz, 2 KHz, 4 KHz, and 8 KHz.

[0039] To illustrate, audio signal strengths at two or more frequencies of a sound emitted by a person and as received by a respective microphone of multiple luminaires (e.g., 2, or, 4, or 5 luminaires) in an area may be labelled with the facing direction of the particular person. Alternatively or in addition, the facing direction labels may be applied to the spatial directivity signatures derived by multiple luminaires from the audio signal strengths of sounds received by a respective microphone of the multiple luminaires (e.g., 2, 3, 4, or 5 luminaires). For example, a spatial directivity signature derived by each luminaire of multiple luminaires in an area may be labelled with the facing direction of the person that emitted a sound. As another example, spatial directivity signatures derived by multiple luminaires in an area may be labelled, as a group, with the facing direction of the person that emitted a sound. The facing direction of a person for labelling purposes may be obtained from a camera, a multi-pixel thermopile sensor, in-person observation, etc. Multiple such labelled signal strength data and / or labelled spatial directivity signatures may be included in the training data obtained based on sounds emitted by one or more persons at different locations in an area and at different facing directions can be used to train the Al model.

[0040] In general, a labelled data set of audio signal strengths in the training data may correspond to shown in Equation (2), where m ranges, for example, from 1 to M and where M is 2 or more and where k represents the number of microphones in luminaires of an area. For example, audio signal strengths of a sound represented by and y&C g), where, for example, (is 1 KHz and where2is 8 KHz may be labelled with a facing direction of the person that emitted the sound. That is, the individual audio signal strengths at two or more frequencies or a set of the audio signal strengths at the two or more frequencies may be considered a spatial directivity signature and may be labelled with a facing direction. Alternatively or in addition, at each luminaire that includes a microphone that receives sounds used to generate training data, a spatial directivity signature that includes or is derived from may be labelled with the facing direction of the person that emitted the sound. Such labelled audio signal strength data or labelled spatial directivity signatures from multiple luminaires may be used to train the Al model. For example, spatial directivity signatures from multiple luminaires derived based on a sound may be labelled as a unit. In general, training data may include multiple such labelled audio signal strengths and / or spatial directivity signatures of multiple sounds emitted by one or more persons and captured by microphones of multiple luminaires.

[0041] In some alternative embodiments, the training data may include or may be obtained from ‘normalized’ audio signal strengths (and / or spatial directivity signatures derived therefrom) of sounds at multiple frequencies of each sound, where the ‘normalized’ audio signal strengths of each sound at multiple frequencies is labelled with the facing direction of the person that emitted the sound. To illustrate, the training data may be based on normalized’ audio signal strengths corresponding to in Equation (3), where m ranges from, for example, 1 to M, where M is 2 or more.

[0042] In some alternative embodiments, the arrival angles, 0(t) and <|>(t), of the sound 130 as received by a particular luminaire may be considered as the spatial directivity signature of the sound 130. For example, the arrival angles, 0(t) and <|>(t), may be determined or otherwise closely estimated by assuming initial values of the arrival angles, 0(t) and <|>(t), in Equation (1) and iteratively updating the values by comparing the calculated audio signal strengths at multiple frequencies of the sound 130 to the actual audio signal strengths as determined from the sound 130 as received by the luminaire (e.g., the luminaire 102) that received the sound 130. The spatial directivity signatures determined in such manner may be used to train an Al model and to estimate the facing directions of the person 120 in the manner as described here with respect to spatial directivity signatures determined based on the differences in the directivities of some frequency components (e.g., 1 KHz and 8 KHz) of the sound 130 as shown in FIGS. 2 and 3.

[0043] FIG. 4 illustrates a top view of facing direction regions of the person 120 of FIG. 1 in a horizontal plane according to an example embodiment. Referring to FIGS. 1-4, the horizontal facing directions of a person may be indicated based on designated regions such as, for example, Front, Right, Back, and Left as shown in FIG. 4. That is, the person 120 may face toward the Front region, the Right region, the Back region, or the Left region regardless of the location of the person 120. For example, the Front region may be between boundary lines 402 and 408 that may form, for example, a right angle. The Right region may be between boundary lines 402 and 404 that may form, for example, a right angle. The Back region may be between boundary lines 404 and 406 that may form, for example, a right angle. The Left region may be between boundary lines 406 and 408 that may form, for example, a right angle. The horizontal facing direction regions Front, Right, Back, and Left may be arranged such that the person 120 facing directly toward the north has a light of sight (shown as a dotted arrow) that extends through the center of the Front region.

[0044] FIG. 5 illustrates a side view of facing direction regions of the person 120 of FIG. 1 in a vertical plane according to an example embodiment. Referring to FIGS. 1-5, the vertical facing directions of a person may be indicated based on designated regions such as, for example, Up, Forward, and Down as shown in FIG. 5. That is, the person 120 may face toward the Up region, the Forward region, or the Down region regardless of the location of the person 120. For example, the Up region may be between a boundary line 502 and a vertical line V that may form 60-degrees therebetween, where the vertical line V extends through the person 120. The Forward region may be between boundary lines 502 and 504 that may form 60-degrees therebetween. The Down region may be between boundary line 502 and a vertical line V that may form 60-degrees therebetween. The vertical facing direction regions may extend 360 degrees around the person 120.

[0045] In some example embodiments, the horizontal facing direction regions Front, Right, Back, and Left along with the vertical facing direction regions Up, Forward, and Down may designate the facing directions of the person 120. For example, the facing direction of the person 120 may be Front-Forward, Front-Up, Front-Down, Right-Forward, Right-Up, Right-Down, Back-Forward, Back-Up, etc. Such facing direction labels may be used in the training data used to train the Al model that is executed to estimate the facing direction of the person 120. To illustrate, one or more of the luminaires 102-110 and / or the server 112 may execute the trained Al model to estimate that the person 120, as shown in FIG. 1, has a facing direction of Left-Forward.

[0046] In some alternative embodiments, facing directions may be defined differently than shown in FIGS. 4 and 5 without departing from the scope of this disclosure. For example, combinations of azimuth angles and elevation or depression angles may be used as labels for training data to indicate the facing directions of the person emitting sounds. As such, the execution of the trained Al model may provide facing directions in terms of azimuth and elevation / depression angles. In some example embodiments, training data for an Al model may be obtained based on sounds emitted by a person in the space (e.g., a senior housing room) that has the same layout of luminaires as one or more rooms where Al model will be used. For example, the area 114 shown in FIG. 1 may be used to obtain training data, the luminaires 102-110 may be subsequently used to estimate the facing direction of a person such as the person 120 using the trained Al model. In such cases, facing direction labels such as different wall labels (e.g., front wall, right wall, etc.) may be used as training data labels. As another example, different object labels such as shelf #1, etc. may be used as facing direction labels and thus as facing directions estimated by executing the Al model. In some alternative embodiments, facing directions may be labelled I and J regions, where I refers to horizontal planes and where J refers to vertical planes. In some alternative embodiments, the Al model may be trained such that the Al model is penalized for relying on latent space representations in estimating facing directions based on sound.

[0047] In some alternative embodiments, a self-learning may be used to train an Al model. For example, a self-learning approach can also be applied, where sound directivity signature based on at least two different audio frequencies is recorded at two or more luminaires. The self-supervised approach then clusters the audio events based on their sound directivity signature. An Autoregressive Predictive Code (APC) algorithm which uses the signals from k-1 microphones (i.e., k-1 luminaires) to predict the signals at the kth luminaire. Since the facing direction of the human speaker plays a key role in determining the signal strength distribution among the k microphones in the room, the APC encoder will learn a latent representation of the spatial distribution of the time-series audio signal, which is predominantly based upon the facing direction and audio arrival angles. Subsequently, a small amount of ground truth data is used to tune the Al model into facing directions such as facing direction regions.

[0048] Referring back to FIG. 1, in some example embodiments, two or more of the luminaires 102-110 may receive the sound 130 and determine spatial directivity signatures of the sound 130, for example, based on audio signal strengths 1 KHz and 8 KHz of the sound 130 as received at the particular two or more luminaires. Two or more of the luminaires 102- 110 may each determine a respective spatial directivity signature of the sound 130 in the manner described above, for example, with respect to FIGS. 1-3. The server 112 may receive the spatial directivity signatures from the particular two or more luminaires and execute the trained Al model to estimate the facing direction of the person 120 based on the spatial directivity signatures. For example, the server 112 may estimate the facing direction of the person 120, for example, as Left-Forward following the facing direction designations described with respect to FIGS. 4 and 5. Alternatively, if the Al model is trained using other facing direction designations such as azimuth and elevation / depression angles, the server 112 may estimate the facing direction of the person 120 accordingly. In some alternative embodiments, one or more of the luminaires 102-110 may receive spatial directivity signatures from one or more of the other luminaires of the system 100 and estimate the facing direction of the person 120 in the manner described with respect to the server 112.

[0049] FIG. 6 illustrates facing directions of the person 120 during a falling incident in the area 114 of FIG. 1 according to an example embodiment. Referring to FIGS. 1 and 6, in some example embodiments, the person 120 may emit sounds 602, 604, 606, 608, 610 right before and during a falling incident. In FIG. 6, the arrows extending through the sounds 602-610 are intended to illustrate the facing directions of the person 120 at the different positions Pl, P2, P3, P4, and P5 of the person 120. At position Pl, the system 100 of FIG. 1 may detect the person 120, estimate the location of the person 120, and start tracking the person 120. At position P 1, the person 120 may emit the sound 602 right before starting to slip on the floor 124. The system 100 may, for example, detect the person 120 in the area 114, estimate the location of the person 120, and track the person 120. The system 100 may also estimate, using the Al model described above, the facing direction of the person 120 based on spatial directivity signatures of the sound 602 determined by two or more of the luminaires 102-110 in the manner described above.

[0050] In some example embodiments, as the person 120 may emit the sound 604 as the person starts slipping at the position P2. The facing direction of the person 120 at the position P2 may also be different from the facing direction of the person 120 at the position Pl. The system 100 may determine, using the Al model described above, the facing direction of the person 120 at the position P2 based on the sound 604 as the sound 604 is received by two or more of the luminaires 102-110. As the person 120 falls down more to the position P3, the person 120 may yet have a different facing direction and may emit the sound 606. The system 100 may determine the facing direction of the person 120 at the position P3 based on the sound 606 as the sound 606 is received by two or more of the luminaires 102-110. At the position P4, the person 120 may be mostly on the floor 124 and may emit the sound 608. The system 100 may determine the facing direction of the person 120 at the position P4 based on the sound 608 as the sound 608 is received by two or more of the luminaires 102-110. At the position P5, the person 120 may be attempting to sit up or get back up and may emit the sound 610. The system 100 may determine the facing direction of the person 120 at the position P5 based on the sound 610 as the sound 610 is received by two or more of the luminaires 102-110.

[0051] In some example embodiments, the system 100 (e.g., the server 112) may process the facing directions determined based on the sounds 602-610 to perform a fall detection by determining whether the sequence of facing directions matches a falling event or incident. That is, the system 100 may determine the sequence of facing directions based on time series of spatial directivity signatures determined from sounds 602-610 and may detect whether the person 120 experienced a falling event / incident based on the sequence of facing directions. In general, the system 100 may have or may access a database of multiple sets of sequences of facing directions that correspond to a falling event of a person, and the system 100 may compare the sequence of facing directions determined based on sounds emitted by the person 120 against the multiple sequences of falling directions to perform a fall detection. By detecting and tracking the person 120, the system 100 can ensure that the sounds 602-610 are emitted by the same person.

[0052] In some alternative embodiments, the sounds 602-610 may be sections of a single sound made by the person 120. To be clear, the sequence of facing directions shown in FIG. 6 is an example and the person 120 may have different facing directions during other falling events / incidents.

[0053] In some example embodiments, the system 100 (e.g., the server 112 of the system 100) may process the sounds 602-610 captured by the microphones of one or more of the luminaires 102-110 to determine whether one or more of the sounds 602-610 indicates or is otherwise associated with a falling event / incident. For example, the server 112 may compare each one of the sounds 602-610 against a database of sounds (e.g., panic sound, surprise sound, etc.) that are made by people right before, during, and right after a falling event / incident. By comparing the sounds 602-610 against the database of falling related sounds, the server 112 may confirm whether a falling event / incident detected by the system 100 based on the facing directions of the person 120. The database of falling related sounds may be stored in the server 112 or may otherwise be accessible by the server 112, for example, from a cloud server.

[0054] In some example embodiments, the system 100 (e.g., the server 112 of the system 100) may use information from one or more radar devices of the luminaires 102-110 to disaffirm the result of the fall detection of (e.g., slipping or tripping to the ground) of the person 120 performed based on the facing directions of the person 120, for example, shown in FIG. 6. For example, because a person typically experiences a relatively fast movement during a falling event / incident, if no change in the speed of the person 120 is detected based on radar signals from the radar device of one or more of the luminaires 102-110, the server 112 may disaffirm / invalidate the result of the fall detection performed, for example, by the server 112 based on the facing directions of the person 120 shown in FIG. 6. To illustrate, if no change in the speed of the person 120 is detected or if the speed change is inconsistent (e.g., too slow) with a falling event / incident during a time frame overlapping the detection of a falling event / incident, the server 112 may disaffirm / invalidate a detection of the falling event / incident determined based on the facing directions of the person 120. If the speed of the person 120 is consistent with the falling event or incident, the processor 702 may confirm the result of the fall detection performed that is determined based on facing directions.

[0055] Although FIG. 6 shows sequence / time series of facing directions during a falling event / incident, the system 100 may use a sequence / time series of facing directions of the person 120 for other purposes such as, for example, to detect an activity of the person 120, etc. For example, the system 100 may be used to estimate the level of interest the person 120 has in one or more products on a shelf. As another example, the system 100 may be used to detect the alertness of the person 120 based on changes in the facing directions of the person 120. As another example, the system 100 may be used for theft deterrence in stores by detecting fidgety behavior in stores.

[0056] In general, the system 100 may distinguish between the person 120 experiencing a falling incident and the person 120 engaging in other activities. For example, the system 100 may distinguish between a sequence / time series of facing directions of the person 120 corresponding to walking and talking followed by lying down and talking and to walking and a sequence / time series of facing directions of the person 120 corresponding to walking and talking followed by falling down and screaming for help. As another example, the system 100 may distinguish between a sequence / time series of facing directions of the person 120 corresponding to singing while changing body posture and a sequence / time series of facing directions of the person 120 corresponding to walking and talking followed by falling down and screaming for help. As described above, the system 100 may compare the sequence / time series of facing directions of the person 120 against a database of sequences of facing directions that correspond to falling events and other activities.

[0057] FIG. 7 illustrates a block diagram of a luminaire 700 corresponding to the luminaires 102-110 of the system 100 of FIG. 1 according to an example embodiment. In general, the luminaire 700 may represent each one of the luminaires 102-110 and other luminaires that may be included in the system 100 of FIG. 1. Referring to FIGS. 1-7, in some example embodiments, the luminaire 700 may include a processor 702 (e.g., a microprocessor), a light module 704, one or more microphones 706 (e.g., an array of microphones), a radar device 708, a memory device 710 (e.g., a nonvolatile memory device such as a flash memory device), a communication interface unit 712, and an occupancy sensor 716. The processor 702 may execute software code stored in the memory device 710 to perform operations described herein with respect to the luminaires 102-110. For example, Al models may be stored in the memory device 710. The communication interface unit 712 may transmit and receive signals complaint with one or more wireless communication standards. For example, the communication interface unit 712 may transmit and receive Bluetooth Low Energy (BLE) signals, Wi-Fi signals, ZigBee signals, and / or other wireless signals. The processor 702 may use the communication interface unit 712 to transmit information, for example, to other luminaires and / or to a server such as the server 112. The processor 702 may also receive information through the communication interface unit 712, for example, from other luminaires.

[0058] In some example embodiments, the light module 704 may provide a light provided by the luminaire 700. For example, the luminaire 700 may be a recessed luminaire, a suspended luminaire, a floor luminaire, a desk luminaire, etc. The processor 702 may control the operation of the light module 704 that may, for example, include one or more light sources such as light emitting diode light sources.

[0059] In some example embodiments, the one or more microphones 706 can capture sounds emitted by a person such as the person 120. The processor 702 may process audio data derived from the sounds captured by the microphones 706. For example, the processor 702 may filter out noise from the captured sound (e.g., the sound 130) and may determine audio signal strengths (i.e., the spectral power distribution) of the sound at multiple frequencies (e.g., 1 KHz, 2 KHz, 4 KHz, and / or 8 KHz) of the sound. The processor 702 may determine the spatial directivity signature of a sound captured by the microphones 706 from the audio signal strengths at least at two or more of the frequencies of the sound. The processor 702 may transmit, via the communication interface unit 712, the audio signal strengths and / or the spatial directivity signature, for example, to the server 112 or to another luminaire of the system 100. Alternatively or in addition, the processor 702 may execute one or more Al models 714 to perform operations such as estimating the facing direction of a person based on spatial directivity signatures determined by two or more of the luminaires of the system 100. In some example embodiments, the processor 702 may perform a fall detection based on a sequence of facing directions of a person as described, for example, with respect to FIG. 6. For example, the processor 702 may estimate a sequence of facing directions of the person 120 from a time series of spatial directivity signatures and may subsequently determine from the sequence of the facing directions whether the person 120 experienced a falling event.

[0060] In some example embodiments, the radar device 708 may be used to confirm the falling down (e.g., slipping or tripping to the ground) of a person, such as the person 120, detected based on the facing directions determined from sounds emitted by the person. For example, if the changes in the speed of the person, as determined based on radar signals from the radar device 708, are consistent with or are not inconsistent with the falling event / incident, the processor 702 may confirm the result of the fall detection performed based on facing directions determined using spatial directivity signatures from multiple luminaires. If the changes in the speed of the person are inconsistent with the falling event / incident, the processor 702 may reject the result of the fall detection performed based on facing directions determined using spatial directivity signatures from multiple luminaires.

[0061] In some example embodiments, the processor 702 may process the sounds captured by the microphones 706 to determine the types of the sounds emitted by a person. For example, a panic sound or a surprise sound may suggest a falling event. The processor 702 may compare the sound received via the microphones 706 against a database of sounds that are made by people right before, during, and after a falling event / incident. As such, based on the types of sounds used to determine facing directions of a person, the processor 702 may confirm the result of a fall detection performed based on a sequence of facing directions. As described above, the sequence of facing directions may be determined using spatial directivity signatures that are, for example, determined based on sounds captured by microphones of multiple luminaires.

[0062] In some example embodiments, the occupancy sensor 716 may include a PIR sensor, a thermopile sensor, and / or another type of sensor that can be used to detect a person and even estimate the location of a person. For example, a multi-pixel thermopile can be used to estimate the location of a person, for example, in the area 114. The occupancy sensor 716 may operate in conjunction with the processor 702 to detect and estimate the location of a person. The processor 702 may also use information from the occupancy sensor 716 to frack the person. For example, the processor 702 may execute an HMM software module to frack the person 120 in the area 114 based on information from the occupancy sensor 716. In some alternative embodiments, the server 112 may perform some of the operations described herein with respect to the processor 702. For example, the server 112 may receive audio data derived from sounds captured by the microphones 706 and may determine the audio signal strengths (i.e., spectral power) of the sounds at two or more frequencies. As another example, the server 112 may determine a spatial directivity signature of a sound captured by each one of the luminaires 102-110 of the system 100 of FIG. 1. As another example, the server 112 may determine execute one or more Al models to estimate the facing direction of a person (e.g., the person 120) based on spatial directivity signatures from one or more of the luminaires 102-110 of the system 100 of FIG. 1 or based on spatial directivity signatures determined by the server 112 based on frequency data.

[0063] In some example embodiments, the luminaire 700 may include a driver that may receive AC power, for example, from a municipality power line, and provide DC power to the components of the luminaire 700 as can be readily understood by those of ordinary skill in the art with the benefit of this disclosure. In some alternative embodiments, the luminaire 700 may include more or fewer components than shown in FIG. 7 without departing from the scope of this disclosure. In some alternative embodiments, some components of the luminaire 700 may be integrated into a single component without departing from the scope of this disclosure. In some alternative embodiments, the radiofrequency (RF) signals may be used instead of or in addition to radar signals without departing from the scope of this disclosure.

[0064] FIG. 8 illustrates a method 800 of facing direction detection according to an example embodiment. Referring to FIGS. 1-8, in some example embodiments, the method 800 includes, at step 802, detecting the person 120 in the area 114 and estimating the location of the person 120. For example, one or more of the luminaires 102-110 may detect and estimate the location of the person 120. Alternatively, the server 112 may detect and estimate the location of the person 120 based on occupancy sensor information from one or more of the luminaires 102-110. At step 804, the method 800 may include tracking the person 120 in the area 114. For example, the server 112 may execute an HMM software module to track the person 120 in the area 114 based on occupancy information from the luminaires 102-110.

[0065] In some example embodiments, at step 806, the method 800 may include receiving the sound 130 (and / or one or more of the sounds 602-610) produced by the person 120 in the area 114. For example, two or more of the luminaires 102-110 may receive the sound 130 or one or more of the sounds 602-610. In some example embodiments, at step 808, the method 800 may include filtering the sound, for example, to remove noise from the noise. The filtering may also be performed to remove some frequency components of the sound by implementing a bandpass filter and / or a low pass filter. Each luminaire of the luminaires 102-110 that receives the sound may also filter the sound as received by one or more microphones of the particular luminaire. In some alternative embodiments, each luminaire that receives the sound may send audio data generated from the sound to the server 112, and the server 112 may process (e.g., filter) the audio data representing the sound as received by the particular luminaire.

[0066] In some example embodiments, at step 810, the method 800 may include determining spatial directivity signatures of the sound 130 and / or one or more of the sounds 602-610. For example, the spatial directivity signature of the sound 130 determined by the luminaire 102 may be defined by the relationship between the sound levels (i.e., the spectral power or audio signal strength) of the sound 130 at two frequency components (e.g., 1 KHz and 8 KHz) of the sound 130 as received by the luminaire 102. For example, the spatial directivity signature of the sound 130 as received by the luminaire 102 may be the audio signal level (i.e., audio signal strength) at a first frequency (e.g., 1 KHz) and the audio signal level at a second frequency (e.g., 4 KHz or 8 KHz). As another example, the spatial directivity signature of the sound 130 as received by the luminaire 102 may be the difference in the sound levels at a first frequency (e.g., 1 KHz) and the audio signal strength at a second frequency (e.g., 4 KHz or 8 KHz). As described above, the sound levels at different frequencies of the sound 130 may be ‘normalized’ by the sound level at a reference frequency (e.g., 100 Hz) of the sound 130, and the ‘normalized’ sound levels may be used in determining the spatial directivity signature of the sound 130. Because the spectral power distribution (i.e., the signal power distribution at different frequencies) of the sound 130 as received by a particular luminaire of the system 100 can depend on the location of the person 120 relative to the particular luminaire, in some example embodiments, the location of person 120 may be used to determine the spatial directivity signature by the particular luminaire. In some alternative embodiments, the server 112 may determine the spatial directivity signature of the sound 130 with respect to each one of the luminaires 102-110 that receives the sound 130 and provides audio data of the sound 130 or spectral power distribution information of the sound 130 to the server 112.

[0067] In some example embodiments, at step 812, the method 800 may include estimating a facing direction of the person 120 based on the spatial directivity signatures of the sound 130. For example, if the luminaire 102 in FIG. 1 is in front of the person 120, the sound 130 as captured by the luminaire 102 may have similar sound levels at 1 KHz and 8 KHz frequencies. In contrast, the sound 130 as captured by the luminaire 104 that is at the back side of the person 120 may have significantly different sound levels at 1 KHz and 8 KHz frequencies as can be determined based on the graphs 200 and 300 shown in FIG. 2 and 3. For example, the server 112 may execute an Al model described above to estimate the facing direction of the person 120 based on the spatial directivity signatures of the sound 130, where the spatial directivity signatures of the sound 130 are determined based on the sound 130 as received by the luminaires 102 and 104. The server 112 may also use one or more spatial directivity signatures from one or more other luminaires (e.g., the luminaire 106) of the system 100 such as the luminaires 106-110. As described above, the spatial directivity signatures of the sound 130 with respect to multiple luminaires may be determined by the luminaires. Alternatively, the spatial directivity signatures of the sound 130 with respect to multiple luminaires may be determined by the server 112 based on audio data from the luminaires generated from the sound 130 as received by the luminaires or based on spectral power distribution information of the sound 130 received from the luminaires.

[0068] In some example embodiments, at step 814, the method 800 may include performing a fall detection of the person 120 based on a sequence of facing directions of the person 120 estimated based on a time series of spatial directivity signatures of sounds, such as the sounds 602-610, made by the person 120. For example, the server 112 may estimate facing directions of the person 120 based on the sounds 602-610 shown in FIG. 6 in the manner described above, for example, with respect to step 812. The facing directions of the person 120 may be estimated based on time series data of spatial directivity signatures obtained or otherwise derived from the sounds 602-610. The server 112 may determine whether the sequence of facing directions determined based on the sounds 602-610 corresponds to or otherwise indicates a falling incident or event, for example, by comparing the sequence of facing directions against sequences of facing directions known to correspond to a falling event or incident. If the server 112 determines that the person 120 has fallen down, the server 112 may send a notification (e.g., a text message or an email) to an entity such as an elderly monitoring service operator, a nurse, a family member, etc.

[0069] By using sounds emitted by a person, the system 100 may estimate the facing direction of the person. Estimating the facing directions of the person using sounds emitted by the person instead of using a camera may be desirable, for example, because of better privacy afforded by the use of sounds. By using a sequence of facing directions estimated using sounds, the system 100 may detect falling events / incidents. Fall detection that is performed while providing improved privacy may be desirable for use in elderly homes, stores, etc.

[0070] In some alternative embodiments, the system 100 of FIG. 1 may include more or fewer luminaires than shown without departing from the scope of this disclosure. In some alternative embodiments, the luminaires 102-110 may be installed in a different configuration than shown without departing from the scope of this disclosure. In some alternative embodiments, the luminaires 102-108 may be suspended luminaires, scones, etc. In some example embodiments, the luminaire 110 may be a desk lamp or another type of luminaire without departing from the scope of this disclosure. In some alternative embodiments, the area 114 may have a different shape than shown without departing from the scope of this disclosure. In some alternative embodiments, multiple persons may be in the area 114 without departing from the scope of this disclosure.

[0071] In some alternative embodiments, the method 800 may include more or fewer steps than shown without departing from the scope of this disclosure. In some alternative embodiments, the steps of the method 800 may be performed in a different order than shown without departing from the scope of this disclosure.

[0072] Although particular embodiments have been described herein in detail, the descriptions are by way of example. The features of the example embodiments described herein are representative and, in alternative embodiments, certain features, elements, and / or steps may be added or omitted. Additionally, modifications to aspects of the example embodiments described herein may be made by those skilled in the art without departing from the scope of the following claims, the scope of which are to be accorded the broadest interpretation so as to encompass modifications and equivalent structures.

Claims

CLAIMS:

1. A method (800) for detection of facing direction of a person using a lighting system (100), the method comprising: receiving (806), by a first luminaire (102) and a second luminaire (104), a sound (130) produced by a person (120) in an area (114); determining (810) spatial directivity signatures of the sound based on sound levels of the sound as received by at least the first luminaire and the second luminaire, wherein the sound levels of the sound are at least two frequencies of the sound; and estimating (812) a facing direction of the person based on the spatial directivity signatures of the sound by executing an artificial intelligence (Al) model (714) trained using spatial directivity signature data labelled with facing directions.

2. The method of Claim 1, further comprising performing (814) a fall detection of the person (120) based on a sequence of facing directions of the person including the facing direction of the person, wherein the sequence of facing directions is determined based on a time series of spatial directivity signatures of one or more sounds (130, 602-610) made by the person (120).

3. The method of Claim 2, further comprising detecting (802) the person in the area and estimating a location of the person.

4. The method of Claim 3, further comprising tracking (804) the person in the area.

5. The method of Claim 2, further comprising confirming a result of the fall detection of the person based on a type of the one or more sounds.

6. The method of Claim 2, further comprising disaffirming a result of the fall detection of the person based on a speed of the person in the area determined using radarbased and / or radiofrequency(RF)-sensing-based detection.

7. The method of Claim 1, further comprising filtering (808) the sound to remove noise from the sound.

8. The method of Claim 1, wherein the facing direction of the person is estimated further based on an additional spatial directivity signature determined based on additional sound levels of the sound (130) as received by a third luminaire (106, 108, 110) and wherein the additional sound levels of the sound are at the at least two frequencies of the sound.

9. The method of Claim 1, wherein the spatial directivity signatures of the sound comprise first spatial directivity signatures of the sound and second spatial directivity signatures of the sound, wherein the first spatial directivity signatures are determined by the first luminaire (102) or the server (112) based on first sound levels of the sound at the at least two frequencies of the sound as received by the first luminaire (102), and wherein the second spatial directivity signatures of the sound are determined by the second luminaire (104) or the server (112) based on second sound levels of the sound at the at least two frequencies of the sound as received by the second luminaire (104).

10. The method of Claim 1, wherein estimating the facing direction of the person is performed by the first luminaire (102), the second luminaire (104), or a server (112) configured to communicate with the first luminaire and the second luminaire.

11. A lighting system (100) for detection of a facing direction of a person, the system comprising: a first luminaire (102, 700) comprising a first processor (702) and a first microphone (706), wherein the first luminaire is configured to receive a sound (130) produced by a person (120) in an area (114) via the first microphone and wherein the lighting system is configured to determine a first spatial directivity signature of the sound based on first sound levels of the sound at at least two frequencies of the sound as received by the first luminaire; and a second luminaire (104, 700) comprising a second processor (702) and a second microphone (706), wherein the second luminaire is configured to receive the sound produced by the person via the second microphone, wherein the lighting system is configured to determine a second spatial directivity signature of the sound based on second sound levels of the sound at the at least two frequencies of the sound as received by the second luminaire,and wherein the lighting system is configured to execute an artificial intelligence (Al) model (714) to estimate a facing direction of the person based on at least the first spatial directivity signature and the second spatial directivity signature.

12. The lighting system (100) of Claim 11, wherein the lighting system is further configured to perform a fall detection of the person based on a sequence of facing directions of the person (120) and wherein the sequence of facing directions is determined based on a time series of spatial directivity signatures of one or more sounds made by the person.

13. The lighting system (100) of Claim 11, further comprising a third luminaire(106, 108, 110), wherein the system is configured to estimate the facing direction of the person further based on a third spatial directivity signature of the sound determined by the third luminaire based on third sound levels of the sound at the at least two frequencies of the sound as received by the third luminaire.

14. The lighting system (100) of Claim 11, wherein the first processor (702) is configured to execute the Al model to estimate the facing direction of the person (120) based on at least the first spatial directivity signature of the sound and the second spatial directivity signature of the sound.

15. The lighting system (100) of Claim 11, wherein the server (112) is configured to execute the Al model (714) to estimate the facing direction of the person based on at least the first spatial directivity signature of the sound and the second spatial directivity signature of the sound.

Images