Photonic lantern spectrometer

Abstract

Photonic lantern spectrometers may comprise photonic lanterns comprising a plurality of tapered waveguides. The tapered waveguides may, for example, comprise a tapered optical fiber bundle or possibly tapered waveguides written in a bulk medium using 3D laser waveguide writing techniques. Spectroscopic information may be obtained by monitoring the light output from various output channels of the photonic lantern. Such output from the photonic lantern can be processed using a machine learning model to make inferences about the wavelength of light input into the photonic lantern spectrometer.

Description

UCRUZ.001WO / 2024-786-2 PATENT PHOTONIC LANTERN SPECTROMETER CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This non-provisional patent application claims priority under 35 U.S.C.119(e) to U.S. Provisional Patent Application No. 63 / 570,119, filed on March 26, 2024, and titled “PHOTONIC LANTERN SPECTROMETER,” which is hereby incorporated by reference herein in its entirety. BACKGROUND Field

[0002] This disclosure generally relates to tapered waveguides and / or tapered waveguide bundles, and tapered waveguide spectrometers and more particularly photonic lanterns and photonic lanterns employed as spectrometers. Description of Related Art

[0003] Photonic lanterns are fiber-based mode couplers. The photonic lantern comprises a fiber bundle including a plurality of optical fibers aggregated together and tapered. Consequently, the photonic lantern has a larger end and a smaller end. Light coupled into optical fibers at the smaller end propagates along the fibers in the bundle to the larger end. However, because of the taper or the increase in size from one end to the other, the fibers are increasingly separated with distance from the smaller end and location closer to the larger end. While there is initial cross talk between spatial modes at the smaller end, individual fibers at the larger end are isolated and carry only a single mode.

[0004] Photonic lanterns can be fabricated by smoothly tapering bundles (e.g., 7 or 19) of independent single-mode fibers down to the point where there is significant cross-talk between the single-mode channels. As the diameter decreases, the bundle transitions from operating as a plurality of single mode fibers at the first larger end into operating as if it were a multi-mode fiber at the second smaller end.

[0005] Conversely, when light with a complicated electric field distribution (e.g., multi-modal) is injected into the multi-mode end (smaller end), the lantern effectively extractsUCRUZ.001WO / 2024-786-2 PATENT the multi-modal input into a series of single-mode outputs that are relayed to the single-mode fibers as the light propagates to the larger end.

[0006] Notably the phase distribution of the input at the smaller end can be inferred from the intensity distribution of the output, single-mode channels at the larger end. Consequently, the photonic lanterns may be used as “wavefront sensors.”

[0007] Various implementations described here employ photonic lanterns for other applications as well. SUMMARY

[0008] Embodiments described herein have several features, no single one of which is solely responsible for their desirable attributes. Without limiting the scope of the inventions as expressed by the claims, some of the advantageous features will now be discussed briefly.

[0009] As discussed above, light may be injected into the smaller end of the tapered fiber bundle or photonic lantern and output from the plurality of fibers at the larger end. The mapping of input wavefront shapes (at the smaller end) to output channel intensities (at the larger end) is wavelength-dependent. As described herein, this wavelength dependent property may be employed to produce a reconstructive spectrometer.

[0010] Various implementations describe herein, for example, comprise a system for determining a spectral distribution of light. The system comprises a photonic lantern and a detector array. The photonic lantern comprises a tapered fiber bundle comprising a plurality of cores. The tapered fiber bundle has a smaller input end having an input face for injection of the light and a larger output end having an output face through which the light is emitted from at least one of the cores depending on the spectral distribution of the light. The cores are closer together at the input end than the output end such that the cores undergo cross-talk that, depending on the spectral distribution, will determine which of the at least one of the cores will emit light. The detector array is configured to capture an image using the light emitted from the output face. The captured image is dependent on the light emitted from the at least one of the cores that outputs light.

[0011] Some implementations describe herein comprise a system for determining a spectral distribution of light, where the system comprising a first stage comprising a photonic lantern and a second stage comprising a plurality of photonic lanterns. The first stageUCRUZ.001WO / 2024-786-2 PATENT comprising a photonic lantern comprising a tapered fiber bundle comprising a plurality of cores. The tapered fiber bundle has a smaller input end having an input face for injection of the light and a larger output end having an output face through which the light is emitted. The cores are closer together at the input end than the output end. The plurality of photonics lanterns in the second stage comprises tapered fiber bundles, each comprising a plurality of cores. The tapered fiber bundles have smaller input ends having input faces for injection of the light and larger output ends having output faces through which the light may be emitted from one or more of the cores depending on the spectral distribution of the light. The cores are closer together at the input ends than at the output ends such that the cores undergo cross-talk that, depending on the spectral distribution, will determine whether said cores will emit light from the output face. Different cores of the photonic lantern in the first stage at said output face are optically coupled to different of the plurality of photonic lanterns of the second stage.

[0012] Some implementations comprise a method implemented by a system of one or more processors. The method comprising obtaining information reflecting output from one or more photonic lanterns, computing a forward pass of the obtained information through a machine learning model; and obtaining the spectral distribution based on the machine learning model. A The photonic lantern in the one or more photonic lanterns comprise a tapered fiber bundle comprising a plurality of cores. The tapered fiber bundle has a smaller input end having an input face for injection of light and a larger output end having an output face through which the light is emitted from at least one of the cores depending on the spectral distribution of the light. The cores are closer together at the input end than at the output end such that the cores undergo cross- talk that, depending on the spectral distribution, will determine which of the at least one of the cores will emit light. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The following drawings and the associated descriptions are provided to illustrate embodiments of the present disclosure and do not limit the scope of the claims.

[0014] Figure 1 is a schematic side view of a photonic lantern spectrometer.

[0015] Figure 2 schematically illustrates an input face (left) and an output face (right) of the photonic lantern spectrometer.UCRUZ.001WO / 2024-786-2 PATENT

[0016] Figures 3A and 3B show example outputs of the output face of the photonic lantern spectrometer for input light having different spectral distributions.

[0017] Figure 4 shows spectral distributions, obtained from an off-the-shelf grating spectrometer, of two inputs beams prior to being coupled into the photonic lantern at different times to observe the difference in output of the photonic lantern for the two different input beams having different spectral distributions.

[0018] Figure 5 illustrates the result of subtracting a first image of the eighteen output single-mode channels when the first input wavelength (shown in Figure 4) is coupled into the input face of the photonic lantern and a second image of the eighteen output single-mode channels when the second input wavelength (shown in Figure 4) is coupled into the input face of the photonic lantern.

[0019] Figure 6 schematically illustrates a plurality of tapered fiber produced via mass production where repeating tapers are drawn in an automated procedure and later cleaved to make individual photonic lanterns (e.g., four lanterns in the example shown).

[0020] Figure 7 is a schematic representation of a two-layer (or two-stage) seven- channel lantern cascade.

[0021] Figure 8 is a schematic representation of an example end-to-end system.

[0022] Figure 9 is a plot on axes of flux (arbitrary units) and wavelength (in nanometers) showing the wavelength spectrum of light from an LED light source, light from said LED light source transmitted through an empty cuvette, and light from said LED light source transmitted through said cuvette filled with water.

[0023] Figure 10 is a plot on axes of normalized intensity (unitless) and lantern port number showing the ratio of the output photonic lantern intensities obtained by dividing the signal output by the photonic lantern from the different channels when light passed through the cuvette filled with water is injected into the photonic lantern as compared to the signal output by the photonic lantern when light passed through the empty cuvette is injected into the photonic lantern.

[0024] Figure 11 is a plot on axes of predicted wavelength (in nanometers), predicted by the photonic lantern spectrometer, and true wavelength (in nanometers), measured by a reference grating spectrometer showing successful recovery of input wavelengths of narrow lines.UCRUZ.001WO / 2024-786-2 PATENT

[0025] Figure 12A-12C schematically illustrate various configurations sensing the optical output from the channels of the photonic lantern.

[0026] Figure 13A schematically illustrates a method of using laser waveguide writing techniques for form a photonic lantern spectrometer.

[0027] Figure 13B schematically illustrates a photonic lantern spectrometer comprising waveguides formed using laser waveguide writing techniques. [0027A] Figure 14 schematically illustrates a photonic lantern spectrometer that includes an injection lantern through which light from a light source is propagated along a first direction to a sample and light from the sample is propagated along a second direction from the sample through the injection lantern. Figure 14 also shows optical fibers coupling the injection lantern to a plurality of photonic lantern. Figure 14 further shows a plurality of Bragg gratings included in optical fibers to reject light from the light source. DESCRIPTION OF CERTAIN EMBODIMENTS

[0028] Although certain preferred embodiments and examples may be disclosed herein, inventive subject matter extends beyond the specifically disclosed embodiments to other alternative embodiments and / or uses of the inventions, and to modifications and equivalents thereof. Thus, the scope of the inventions herein disclosed is not limited by any of the particular embodiments described below. For example, in any method or process disclosed herein, the acts or operations of the method or process may be performed in any suitable sequence and are not necessarily limited to any particular disclosed sequence.

[0029] For purposes of contrasting various embodiments with the prior art, certain aspects and advantages of these embodiments are described. Not necessarily all such aspects or advantages are achieved by any particular embodiment. Thus, for example, various embodiments may be carried out in a manner that achieves or optimizes one advantage or group of advantages as taught herein without necessarily achieving other aspects or advantages as may also be taught or suggested herein. Lantern Spectrometer

[0030] Figure 1 is a schematic side view of a photonic lantern spectrometer 10 comprising a photonic lantern 12. Figure 1 shows a side view of the photonic lantern 12UCRUZ.001WO / 2024-786-2 PATENT comprising a tapered fiber bundle. As illustrated, a plurality of fiber cores 14 surrounded by cladding 16 are reduced in size from the larger output end 18 to the smaller input end 20. The smaller input end 20 comprises an effective core 22 wherein light is guided within the cladding 16 which operate as a single effective core having sufficient effective core diameter to be support multiple modes. The individual cores 14 spread out toward the larger output end 18 such that the cores are sufficiently spaced apart to reduce cross-talk therebetween and guide light within the individual separate cores. Figure 1 also shows an input face or input 24 of the tapered optical fiber bundle 12 at the input end 20 and an output face or output 26 at an output end 18.

[0031] The source signal is delivered to the smaller input end 20 by a single-mode fiber 28, which can be coupled with relaxed tolerances to the input face or facet 24 of the photonic lantern 12. At the input 24 is an interior “guiding core” or effective core 22 formed by the cladding 16 which has a capillary tube 30 surrounding the bundle cores 14. After tapering, the single-mode cores 14 effectively vanish. The light is “pushed” into the single-mode cladding 16, which then acts like an effective “core” 22 for the multi-mode end. The guiding is provided by the capillary 30 which has now become the cladding. Accordingly, the lantern input 24 is effectively multi-modal, triggering interference patterns in a straight propagation region. This region then expands in two dimensions (e.g., X and Y) and the guiding multi-mode core 22 splits into multiple diverging, single-mode channels 14 (in this example, 7 core / channels) with distance away from the input end 20 and to the output end 18. Initially these channels 14 experience significant cross-talk, but as the bundle expands, the cores / channels separate further until they are independent or substantially without cross-talk therebetween, the separate cores individually guiding a single-mode with an intensity representative of the sampled and wavelength-dependent interference pattern. This basic behavior can be applied to a variety of problems and application some of which are discussed below.

[0032] In various implementations described herein, however, instead of feeding multi-modal light into the multi-modal end (smaller end) 20 of the photonic lantern 12, single- mode light is injected into the smaller end. As the single-mode light propagates down the multi- modal portion of the lantern 12, this light interferes with itself in a wavelength-dependent fashion. Notably, the intensity distribution of the output (of the larger end) 18, single-mode channels 14 contain information about the spectral content of the input signal introduced into the smaller end 20. The system 10, so configured, can operate as a spectrometer.UCRUZ.001WO / 2024-786-2 PATENT

[0033] More specifically, this system 10 is a reconstructive spectrometer. A computer algorithm can be used to learn the mapping between spectral content and output mode intensities. As will be described, the computer algorithm may be a machine learning model. For example, a neural network may be trained to output spectral content based on input of the output mode intensities as depicted in image(s). In this example, the neural network may output a vector of values, with the values indicating measures associated with individual frequencies within a frequency range. As an example, the measures may reflect intensity values and the individual frequencies within the frequency range may be separated via a threshold frequency step. Thus, the frequency range may be separated into frequency steps and the vector may describe the spectral content within the frequency range.

[0034] As discussed above, the taper 12 comprises a plurality of optical fibers 32 having individual cores 14 and cladding 20 surrounding the core. A glass capillary 30 surrounds the plurality of individual fibers 32, and this capillary may be held in a ferrule or coating or other substrate. The taper 12 is such that the plurality of fibers 32 becomes smaller and smaller with distance from the larger end 18 to the smaller end 20. At the smaller end 20, the cores 14 become close to each other and so small that the light is effectively guided within the cladding 16, which operates as a single effective core 22 having a core diameter that although small is sufficiently large to support multiple modes.

[0035] In various implementations of the systems 10 described herein, the small end 20 is the input end. Light from a light source is coupled into the smaller input end 20 of the tapered fiber bundle 12. The system 10 may further comprise coupling optics such as one or more optical fibers (e.g., single mode fiber) 28, lenses or mirrors to couple light from a light source into the smaller input end of the tapered fiber bundle or photonic lantern. In various implementations described herein, single mode light is coupled into the effective core 22 of the smaller input end 20 of the tapered fiber bundle 12. In this example, the effective core 22 is formed by the cladding 16 as the hexagonal array of tapered, single-mode cores 14 effectively vanish as their size is reduced with the taper of the fiber bundle 12. This light coupled into the fiber bundle 12 at the smaller end 20 propagates along the tapered region proximal the smaller end largely confined to the effective core 22. As the fiber bundle 12 progressively enlarges with distance away from the smaller end 20 and toward the large output end 18, the individual cores 14 begin to separate and the distance between them increases and the light is distributed into theUCRUZ.001WO / 2024-786-2 PATENT different separate cores depending, in part, on the wavelength spectrum of the light. At the larger output end 18 of the tapered fiber bundle 12, the individual cores surrounded by cladding 16 are sufficiently separated from each other such that cross-talk between them is reduced and is negligible. The separate cores 14 may be referred to herein as channels.

[0036] Figure 2 (left) schematically illustrates the input face 24 of the photonic lantern spectrometer 10. A single-mode beam is directed into the effective core 22 having an effective core diameter. In this example, the effective core 22 is formed by the cladding 16 as the hexagonal array of tapered, single-mode cores 14 effectively vanish as separate guiding structures as their size is reduced with the taper of the fiber bundle 12. This effective core 22 comprising the cladding 14 at the small end 20 acts as a multi-mode fiber waveguide. The effective core 22 is positioned inside the capillary 30 or other substrate, which simply holds the inner portion in place.

[0037] Figure 2 (right) schematically illustrated the output face or facet 26 of the photonic lantern spectrometer 10. The small circles in the hexagonal array represent the guiding, single-mode cores or channels 14. Single mode cladding 16 surrounds each core. An outer substrate, e.g., a capillary 30 is shown in which the single-mode bundle is inserted. As described above, measurements of the relative intensity of the different output cores 14 can be used to determine the wavelength spectrum of the single mode input light.

[0038] Figure 2 (left) shows the smaller input end 20 having the cores 14 sufficiently close together to produce a multi-mode effective core 22 as well as the larger output end 18 having the individual cores separated from each other and surrounded by respective claddings 16 so as to support separate respective individual single modes. As is discussed herein, introducing single mode light having a particular spectral distribution into the smaller input end 20 of the tapered fiber bundle 12 will cause different ones of the cores or channels 14 at the larger output end 18 to output light with different intensities. Figures 3A and 3B show examples of light output from the different cores or channels 14 at the output end 18 (e.g., output face 26) of the photonic lantern 12. Spots of light 31 are visible from various of the cores 14 of the photonic lantern 12. Such images, e.g., captured by a detector array imaging the output face 26 of the photonic lantern 12, may be referred to as lantern images, target images, or science images in some cases. Light having different spectral distributions or wavelengths was input into the input end 20 of the photonic lantern 12 to produce the different intensity patterns comprising theUCRUZ.001WO / 2024-786-2 PATENT plurality of different spots 31 that are shown in Figures 3A and 3B, respectively. As shown, different intensity levels are produced at different channels 14. Which core or channel 14 will output light and the intensity of such light output will depend on the spectral distribution of light input into the smaller input end 20 (e.g., into the input face 24 of the tapered fiber bundle 12). Accordingly, in various designs described herein, the system 10 further comprises a camera (e.g., detector array and imaging lens) at the larger output array to image the larger output end 18 (e.g., the larger output face 26 of the tapered fiber bundle 12) and monitor which cores or channels 14 are outputting light and the relative intensity of light emitted from the different cores / channels. By imaging this pattern of spots at the larger output end 18 of the tapered fiber bundle 12, the spectral distribution of the single mode light input into the smaller input end 20 of the tapered fiber bundle can be ascertained by an electronic processor programed to determine from the pattern of light the spectral distribution. Demonstration

[0039] The wavelength sensitivity of a photonic lantern was demonstrated by feeding two different narrow-wavelength beams of single mode light into the smaller input end 20 of the photonic lantern 12. To provide a single mode input, the two narrow wavelengths were coupled into a single-mode fiber 28 (in order to select only single-mode light) that was optically coupled to the input end 20 of the tapered fiber bundle 12. Before injecting these single-mode signals into the lantern 12, this light was first injected into a grating spectrometer from Ocean Optic. Figure 4 shows the two spectra 33a, 33b obtained from this off-the-shelf grating spectrometer. This spectral distribution shows a difference of 1 nm between the two beams.

[0040] The same two beams were injected into the photonic lantern 12 and images of the output face 26 were obtained using a camera. As discussed, single-mode intensities corresponding to the output of the respective single mode fibers / channels 14 are imaged with the camera. Two images (one from each wavelength input into the smaller input end 20 of the tapered fiber bundle 12) are subtracted to show that there are obvious differences in the intensity distributions of the 18 output single-mode channels 14, as seen in Figure 5.

[0041] Accordingly, this experiment demonstrates the utility of photonic lanterns 12 as reconstructive spectrometers.UCRUZ.001WO / 2024-786-2 PATENT Spectral Extraction

[0042] As discussed above, the system 10 includes a camera and / or the detector array (e.g., 2D detector array) and imaging optics (e.g., one or more lenses) configured to image the output of the photonic lantern. The detector array (e.g., 2D detector array) may comprise, for example, a CCD or CMOS detector array. In some implementations, the lens has a focal length and is disposed at a distance (e.g., a focal length) away from the detector array to image objects a distance away therefrom. Other configurations are possible. Zoom lenses and / or other lens configurations may be employed.

[0043] The system 10 may further include electronics or circuits or circuitry configured to receive signals from the camera and / or the detector array. The electronics may include one or more computer processors to extract the spectral information from a photonic lantern 12, for example, using a computer algorithm. In various implementations, for example, the electronics determines spectral information from the image of the output of the tapered fiber bundle 12. For example, the electronics can evaluate the image of the output face 26 of the taper fiber bundle 12, which varies based on which cores 14 output light and how bright such light is.

[0044] Additionally, in certain configurations of the system a wavelength dispersive prism such as a low-dispersion prism is positioned in the optical path between the output or output end 18 of the lantern (e.g., output lantern face 26) and the camera. The wavelength dispersive prism may comprise an optically transmissive or transparent material that has an index of refraction that varies with wavelength. Accordingly, the wavelength dispersive prism will refract or redirect different wavelengths of light different amounts. As a result, the prism receives light from the cores 14 of the larger output end 18 of the tapered fiber bundle 12 and refracts or redirects different wavelengths in the output of the different cores different amounts creating an elongated area or spot of light with different wavelengths at different locations along the length of that area. The wavelength dispersive prism at the output of the tapered fiber bundle 12 thus elongates the “spots” of the different respective channel’s output into short segments, with a wavelength dependence along the segment. The camera may be configured to image these elongated output spots or areas. Consequently, the prism can provide further wavelength discrimination. Different patterns of light in the image captured by the camera which include different cores or channels 12 being bright and others not, as well as the distribution of light in the elongated spots for the different cores or channels can be employed by the electronics toUCRUZ.001WO / 2024-786-2 PATENT ascertain the spectral distribution of the light input into the smaller input end of the photonic lantern. Although a prism is discussed above, other wavelength dispersive optical elements such as, for example, diffractive optical elements like gratings may be employed.

[0045] In certain implementations, the images can be evaluated by the electronics, e.g., processor(s), using one or more machine learning algorithms. In some embodiments, the detectors may include individual photodetectors, e.g., which are separated from each other. For example, the individual photodetectors may measure strength (e.g., intensity or other measure) associated with light output by individual cores, core regions, channels, etc. These measurements may be used as input to the one or more machine learning algorithms, for example via forming the measurements into an input vector, via providing the individual measurements to individual input nodes, and so on. The analysis may be divided into a training step and then spectral extraction (e.g., inference step).

[0046] Training may be accomplished by using one or more stable and narrow wavelength band light sources such as a laser source (e.g., a monochromator) having known wavelength distributions, e.g., light having a spectral band with a central wavelength and a bandwidth, possibly a narrow bandwidth. A set of different wavelength beams, e.g., having different central wavelengths, within the chosen wavelength range of the device can be separately input into the smaller input end 20 of the photonic lantern 12 while images of the output end 18 are captured by the camera thereby producing a set of images that are paired with known wavelengths. Thus, the images may reflect input and the associated central wavelengths may be used to form ground truth information. As described above, in some embodiments images may not be used and measurements associated with a plurality of individual photodetectors (e.g., photodetectors not included in an imaging sensor such as in a 2D detector array in an imaging sensor) may be used. The spacing between different wavelengths (e.g., central wavelengths) and wavelength bandwidth of the source (at any given wavelength or central wavelength) may be selected based on the desired spectral resolution of the final, extracted spectra (e.g., as limited by the device design and sensitivity). The ground truth may include a multitude of zero, or substantially zero, values for individual wavelengths within the wavelength range along with a measure associated with a central wavelength (e.g., an intensity value). In some embodiments, the ground truth may represent a normalized intensity distribution associated with an input laser source. In some embodiments, the ground truth may representUCRUZ.001WO / 2024-786-2 PATENT actual intensity values. The extracted spectrum is the measurement from the lantern 12 after processing, for example, by the algorithm(s) used to correlate the pattern of light output from the output face 26 of the photonic lantern 12 with a wavelength spectral distribution or information. For the same photonic lantern spectrometer 10, it may be possible to extract spectra with different spectral resolution and wavelength ranges, as these depend on the nature of algorithmic training (as well as the limitations of the lantern 12 itself).

[0047] This plurality of images and associated input wavelengths can provide an initial training data set. In various implementations, this set of images may be expanded upon by numerically summing various combinations of individual scans, effectively creating, for example, a mock broad-band spectrum. Training could alternatively or in addition include providing light to the input 20 of the photonic lantern 12 that is output from a variable broad- band light source. In various such implementations, the photonic lantern 12 could be configured to receive the light from the broad-band light source while the spectrum of the broad-band light source (and hence the wavelength spectrum input into the photonic lantern) can be monitored simultaneously (e.g., using a beam-splitter and an optical spectral analyzer). Similar to the above, the ground truth associated with the broad-band light source may include a vector of values reflecting measured, or normalized, intensities for individual frequencies within the wavelength range.

[0048] The extraction process may involve an initial image normalization step followed by a machine learning (ML) algorithm, which is trained using the training set described above.

[0049] An example of such an ML algorithm is a vision neural network (e.g., 2D- CNN (convolutional neural network) followed by one or more fully-connected network dense layers, or an attention-based network, such as a vision transformer that leverages, as an example, an embedding or linear projection followed by attention-layer(s) and output layer(s)). In some embodiments, the output layer may not include a non-linear activation function such that numeral values may be directly predicted without transformation. Additionally, a convolutional or attention-based network may be used for input that includes individual measurements associated with photodetectors (e.g., not included in an imaging sensor). In some embodiments, a non-linear activation function may be used. For example, a sigmoid function may be used toUCRUZ.001WO / 2024-786-2 PATENT compress the values between 0 and 1. As another example, a softmax function may be used to normalize the output such that relative intensity may be determined.

[0050] Another example of such an ML algorithm is a dense, or fully-connected, network. For example, images associated with input wavelengths may be input into a fully- connected network. In this example, a portion of the image may be understood to include the above-described spots which represent the imaged output from the photonic lantern 12. Thus, pixel values associated with this portion may be provided as input. Optionally, the spots may be positioned e.g., consistently positioned within sub-areas of the image. These sub-areas may be used as the input. In some embodiments, the pixels forming the sub-areas may be concatenated. In some embodiments, information characterizing the spots may be used as input. For example, measures of intensity or brightness of respective spots may be used. As another example, size measures associated with each spot may be used. Additionally, a dense or fully-connected network may be used for input that includes individual measurements associated with photodetectors (e.g., not included in an imaging sensor).

[0051] Once trained, a previously unseen image captured by the camera, possibly referred to herein as a lantern image, target image, or science image (see. e.g., Figures 3A and 3B), that is the output intensity pattern produced by the light exiting the various cores 14 at the larger end 18 of the photonic lantern 12, possibly being transmitted through a wavelength dispersive prism or diffractive optical element (e.g. grating), can be fed to the ML algorithm. Similarly, in implementations in which individual photodetectors are used, the input may include individual measurements associated with the photodetectors (e.g., not included in an imaging sensor). For example, a forward pass through the above-described ML algorithm may be computed. The output of the dense layer, or final layer of the ML algorithm, is then a regression vector with predicted intensity values as a function of the (trained) wavelength steps. This output corresponds to the extracted wavelength spectrum of the previously unseen, input light source that was used to couple into the input end 20 of the photonic lantern 12. Advantages

[0052] Miniaturized spectrometers have numerous applications. Advantageously, photonic lantern spectrometers 10 such as described herein may be lightweight, compact, have a small footprint or any combination of these. The photonic lantern spectrometers 10 are alsoUCRUZ.001WO / 2024-786-2 PATENT advantageously “built-in” to an optical fiber, a format that is commonly used to collect light. Use of a fiber-based device such as the tapered fiber bundle spectrometer 10 can avoid a lossy and difficult to manufacture interface between a fiber collector and a 2D planar device, like a chip.

[0053] Additionally, the lantern 12 advantageously has a larger effective input diameter and can accept light entering within a larger cone of angles as compared to single-mode interfaces to single-mode fibers or 2D planar devices, like a chip. As a result, the manufacturing tolerances for alignment of light sources to the input of a photonic lantern spectrometer 10 can be significantly relaxed while maintaining high sensitivity as well as potentially stability over changes in stress state and temperature. Design and Fabrication Basic Design

[0054] Variation in the design of the photonic lantern spectrometer system 10 can be employed for different applications. For example, the desired operating wavelength range may inform the design and / or selection of the single-mode cores 14 which are ultimately bundled and tapered under heat and pressure to form the photonic lantern 12. In some cases, single-mode cores 14 with diameters or mode field diameters larger, significantly larger, or even exactly matched to the mode field diameter of the longest operating wavelength of the spectrometer 10 may be beneficial. Such matching can improve single-mode behavior at this wavelength (and potentially or even likely across the full wavelength range, assuming the wavelength range is not too broad). The spectral information from internal optical interference in the taper region of the photonic lantern 12 may potentially be encoded into exactly the number of single-mode cores 14 as a result.

[0055] In other cases, mode field diameters may be used for individual lantern cores 14 when fabricating it that are larger than the longest operating wavelength so as to provide a plurality of spatial modes. Several spatial modes will then propagate through the output channels 14 as the fiber cores exhibit few-mode behavior. The spectral interference pattern can then be sampled with a greater number of spatial modes, and while the few modes traveling through an individual output core 14 cannot be distinguished from measurements of the overall core’sUCRUZ.001WO / 2024-786-2 PATENT output intensity alone, these modes may still be separated by other approaches, such as, for example, by measuring and / or learning the shape of the near-field light distributions of the different output channels or by modestly dispersing the output with a prism or grating as discussed elsewhere herein.

[0056] Other design considerations for the fabrication of lantern spectrometers 10 include the taper ratio and the material used for the outer glass tube or capillary. While a small degree of tapering can allow the lantern to function, the final tapering parameters may be influenced by the input effective core diameter and associated numerical aperture. For a lantern spectrometer 10, these may be designed to be “large” enough to enable efficient and / or easy (e.g., low tolerance) coupling with single-mode input, but specifics may vary depending on the particular application. The effective core size, width and / or diameter, may be, for example, 20 micrometers or microns (µm), 25 µm, 30 µm, 35 µm, 40 µm, 45 µm, 50 µm, 55 µm, 60 µm, or any range formed by any of these values such as 35 µm to 45 µm or 25 µm to 55 µm or 20 µm to 60 µm or possibly larger or smaller. The input numerical aperture (NA) is determined by the materials used and could be 0.22 or as large as 0.35 or as small as 0.05, or possibly larger or smaller.

[0057] The “performance” of the spectrometer may also be improved or modified by providing some variation in the sizes and properties of the single-mode cores. Such variation may introduce distortions in the spectral patterns arising within the taper region and therefore yield different and potentially improved sampling of the spectral information. A specified spatial pattern of varying core sizes or other differences in the core types (e.g., indices of refraction) could illicit specific, desired behavior in the output intensities that would improve the ability of downstream algorithms to recover spectral input. Such designs may be influenced by the specific application. For example, a lantern spectrometer 10 sensitive to a narrow emission line may have a make-up of core properties and sizes that differs from a lantern spectrometer designed for more broadband or continuous spectral analysis. Another application may be a lantern spectrometer 10 whose spectral resolution varies with wavelength in a specified and desired way (e.g., combining higher resolution for certain wavelengths with the ability to measure a larger wavelength range at lower resolution for different wavelengths). In some implementations, for example, the variation in the core size may be 5% or more, for example 2%, 3%, 4%, 5%, 6%, 7%, 8%, 8%, 9%, 10%,UCRUZ.001WO / 2024-786-2 PATENT 11% 12%, or any range formed by any of these values (e.g., 5%-8% or 5%-10% or 5%-11%) or possibly larger or smaller.

[0058] Finally, different types of fibers 32 employing different kinds of glass may be used to construct lantern spectrometers that work at wavelengths where silica does not transmit (e.g., fluoride-based fibers for infrared wavelengths). Fabrication

[0059] Figure 6 shows a schematic representation of mass production where repeating tapers 34a, 34b, 34c, 34d are drawn in an automated procedure and later cleaved to make individual lanterns 12. In this example, four such tapers 34a-34d are shown.

[0060] Accordingly, for the mass production of lantern spectrometers 10, a single undrawn fiber bundle of cores 14 may be tapered on a glass-working station or fiber-draw tower in repeating segments 34a, 34b, 34c, 34d as shown in Fig. 3, which could then be cleaved to obtain many lanterns 12 at once.

[0061] Individual lanterns (e.g., tapered fiber bundles) 12 can be coupled to an input fiber 28 such as a single mode fiber through standard techniques such as, for example, barrel alignment or UV-cured epoxy. The alignment tolerances are low given the large effective core size at the smaller end 20 of the tapered fiber bundle 12, which is a major benefit of lantern spectrometers 10. In various implementations, the input fiber 28 may be misaligned with respect to the effective core 22 formed by the single mode core cladding to excite a greater range of interference patterns. For example, the center of the input fiber 28 can be offset with respect to the center of the plurality of cores 14 and / or the center of the cladding 16 or effective core 22 by multiples (e.g., 1, 2, 3, 4, or more) of one-half of the wavelength of light at the median of the desired operating wavelength range, up to a maximum offset equal to the radius of the lantern’s effective core. Additionally or in the alternative, the angle of the input beam may be misaligned to the propagation axis of the lantern 12. The angle of the input beam and / or of the input optical fiber 28 may be misaligned with respect to the axis of the longitudinal axis of the lantern by at least or at most 5%, 6%, 8%, 10%, 12%, 15%, 18%, 20%, 24%, 25%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100% of the acceptance angle of the photonic lantern at the smaller input end 20 or any range formed by any of these values (e.g., 5%-10% or 5%-20% or 5%-30% or 5%- 40% or 10%-20% or 10%-30% or 10%-40% or 10%-50% or 10%-80% or 10%-90% of theUCRUZ.001WO / 2024-786-2 PATENT acceptance angle of the photonic lantern) of possibly larger or smaller angle. For example, the input beam may be injected into the input end 20 of the photonic lantern 12 at the steepest accepted angle possible. Such misalignment may increase the likelihood of the light to self- interfere or interfere with itself.

[0062] Additionally, in various implementations, a specific spatial arrangement of single-mode cores 14 in the manufacturing “preform” would allow for this input-to-lantern coupling to become part of the overall fabrication process. This could be accomplished by inserting a transition in the preform between the bundle of cores 14 and a single core of larger diameter. For example, at the transition point a single cane with a desired, slightly higher index of refraction could be positioned within a bundle of canes with slightly lower index of refraction. The ratio of the indices of refraction would be chosen based on the desired numerical aperture of the final input single-mode fiber, while the cane diameter of the cane with the slightly higher index of refraction would be chosen based on the desired core size of the final input single-mode fiber. The bundle of canes opposite the transition point from the lantern would span the same diameter as the bundle of canes making up the preform of the lantern. After tapering is applied to both the lantern preform and the input cane bundle preform, the higher-index cane will become the core of the single-mode fiber input while the surrounding canes in the input cane bundle become the cladding of that single-mode fiber. The tapering process results in fusing the single-mode input to the lantern making this input injection interface very robust. Alternatively, the lantern taper could be drawn down to an extreme degree at which point the tapered end 20 then acts like a single-mode fiber. This may be desirable for applications where plenty of source light is available and can be easily injected into the (single-mode) lantern spectrometer input 20 (see below discussion of Diffuse Spectroscopy application).

[0063] A cascade of lanterns 40 such as shown below in Figure 7 may be desirable for multiple reasons. The cascade 40 may comprise a first stage (or layer) 42 comprising a single photonic lantern (e.g., tapered fiber bundle) 12 to which a second stage (or layer) 44 comprising a plurality of photonic lanterns (e.g. tapered fiber bundles) 12a-12g are concatenated or optically coupled. The input 20 of the photonic lanterns 12a-12g in the second stage 42 can be optically coupled to respective channels 14 at the output 18 of the photonic lantern 12 in the first stage 42. In the example shown, for example, the photonic lantern 12 in the first stage can have seven (7) cores or channels 14. The second stage 44 therefore includes seven (7) photonic lanterns 12a-UCRUZ.001WO / 2024-786-2 PATENT 12g, the inputs 20 of which are coupled to the output channels of the photonic lantern in the first stage. The result is that the 7 channels 14 in the first stage 42 are distributed into a total of 49 channels at the output 18 of the second stage 44 as one channel at the output of the first stage 12 is coupled into seven channels in the second stage. Although two stages are shown, more than two stages, for example, three, four, five, or possibly more may be used. The final number of outputs 18 will exponentially increase with the number of stages.

[0064] Total number of output channels = (number of channels in one of the tapered bundles)number of stages

[0065] (In some implementations, however, a photonic lantern 12a-12g need not be optically coupled to each channel 14 though. In such designs, the total number of output channels 14 may thus be reduced. Additionally, the number of channels in the photonic lantern may be different.) Coupling the output 18 of one lantern 12 to the input of another 12a-12g can again be accomplished easily (with low tolerance) using standard alignment and bonding techniques after or following a fanout of the single-mode output. However, for mass production, it may be worthwhile to develop a preform that contains “bundles of bundles” positioned in such a way that upon tapering, the desired lantern cascade 40 is produced.

[0066] As discussed above, Figure 7 is a schematic representation of a 2-layer (or 2- stage) 7-channel lantern cascade 40 with N2= 72= 49 final outputs 18. As discussed herein, the photonic lanterns 12a-12g in the second stage 44 may be physically as well as optically coupled to the photonic lantern 12 in the first stage 42 although a gap is shown in the schematic representation depicted in Figure 7.

[0067] The ML algorithms described herein may optionally be trained based on the above-described cascade of lanterns 40. For example, input light of known spectrum may be provided to the input lantern 12. Images may then be obtained from the output 18 of the 7 lanterns 12a-12b and used to train the ML algorithms as described herein.

[0068] An alternative approach for producing complex lantern cascades could be 3D laser writing of waveguides within an appropriate medium such as, for example, discussed in connection with Figures 13A and 13B.UCRUZ.001WO / 2024-786-2 PATENT Spectral Resolution and Wavelength Range

[0069] Specific spectroscopic applications may call for a desired spectral resolution and wavelength range. For a given spectrometer, there may be a design or performance trade-off between these two properties because spectral resolution represents precision while wavelength range represents dynamic range. In a lantern spectrometer 10, these two properties can be tied to the “spectral sampling density” (e.g., the number of lantern cores or channels 14) and the nature of the spectroscopic training and extraction.

[0070] Considering wavelength range first, the resultant wavelength range may be determined by the choice of materials (see above), but will be further limited by the ability of the lantern spectrometer 10 to sample spectral information. With a limited number of cores or channels 14, a limited sampling of the spectral pattern can be achieved. If the wavelength range is too large, the spectral response (in terms of the relative output intensity distribution of all the cores 14) as a function of wavelength may begin to repeat, confusing the ability to extract the unique, input spectrum. Part of the solution to this problem is to insert a bandpass filter in the spectrometer’s optical system with a specified, limited wavelength range such that the spectral response over that wavelength range is unique.

[0071] One way to improve the wavelength range is to increase the spectral sampling density, which can depend on the number of cores or channels 14 in the photonic lantern 12 (see below). With increased sampling, wavelengths that previously had duplicate output signals become unique. An increased spectral sampling density is also useful for increasing the spectral resolution because increased sampling of a complex pattern improves the precision at which that pattern can be measured.

[0072] In practice, the wavelength range and resolution may also depend on the approach used to train the software which extracts the spectrum. This too can be adjusted or optimized to achieve the desired performance. For example, for a given photonic lantern spectrometer 10, using lower-resolution training data (e.g., widening the width of a single- wavelength step for the training monochromator or widening the bandwidth of the light of the laser source used for training) may allow for unique spectral extraction over a wider wavelength range, but at lower spectral resolution. Higher resolution training for that same lantern 12 may increase the extracted spectral resolution, but entail a restricted wavelength range to ensure uniqueness.UCRUZ.001WO / 2024-786-2 PATENT Spectral Sampling Density

[0073] As referred to above, the spectral sampling density corresponds to the number of independent measurements (samples) of the wavelength-dependent interference pattern arising within the photonic lantern 12. The spectral sampling density, which can depend on the number of cores or channels 14 in the photonic lantern 12, limits or affects both wavelength range and spectral resolution. As discussed above, spectral sampling density can be increased by increasing the number of lantern cores 14. Lanterns 12 with several hundred cores 14 have been made successfully. However, for mass production or because of cost or hardware constraints on fabrication, such a large number of cores 14 in a single lantern 12 may not always be possible or practical to achieve.

[0074] An alternative is to “cascade” a hierarchy of lanterns 12 with a more modest core numbers such as, for example, described above with respect to Figure 7. For example, the (single-mode) output channels 14 of one nineteen core (19-core) lantern 12 could each be directed into nineteen (19) additional lanterns, each with nineteen (19) cores. The effective number of final output cores 14 sampling the spectral information arising in the cascade 40 is then 192= 361. If these outputs were directed to another layer of nineteen core (19-core) lanterns downstream, the total number of cores 14 would grow to 193= 6859. A lantern cascade 40 could be produced as part of the drawing / tapering process or 3D laser-writing process as referenced above and such as, for example, discussed in connection with Figures 13A and 13B can be used to make the lanterns 12. Manual alignment and bonding of the interfaces at each layer may not be needed in such cases.

[0075] Another way to increase the spectral sampling density is to modestly disperse the lantern outputs such as discussed above with a wavelength dispersive element such as a prism or diffractive optical element (e.g., a grating). A converging lens may be used to optically couple to the dispersive element (e.g., possibly collimating the output beams before being incident on the dispersive element, e.g., a grating or prism). Downstream optics (e.g., camera optics) could focus and image a dispersed image of the lantern output face 26 on a detector or plurality of detectors (e.g., detector array). Instead of obtaining spots at each core location, line segments may be recorded whose intensity profile (along the line segment) would provide additional information about the spectral content of the individual outputs. The use of one or more wavelength dispersive elements allows a more unique set of output patterns to be analyzed,UCRUZ.001WO / 2024-786-2 PATENT equivalent to increasing the sampling density. The line segment profiles could be incorporated into the spectral extraction which learns the mapping between output light and input spectrum.

[0076] Increasing spectral sampling density may also motivate a lantern design with cores 14 that support multiple modes at the design wavelength. Here, the additional information in the few-mode output can be extracted, for example, using a wavelength dispersing element such as a prism or grating as described above. Different modes would contribute somewhat different line segment profiles. In addition, the spatial profile of the output spots (or line segments) may contain information about which of the few-modes is excited in each output channel. The extraction software can be written to include these spatial profiles in the way it learns the spectral mapping, thereby increasing the number of spectral samples.

[0077] The spectral sampling density also may potentially be affected by the launch conditions of single-mode light into the lantern 12 (e.g., incidence angle, spatial profile, focus, etc.). These may be adjusted or optimized to achieve a desired sampling. For example, the single-mode source may comprise a narrow gaussian beam compared to the large effective diameter of the lantern input. This gaussian beam may be intentionally offset with respect to the center of the photonic lantern. Additionally, or in the alternative, the output gaussian beam may be at an angle with respect to the photonic lantern’s propagation axis. The light may be collected, but the nature of the interference pattern and the sampling thereof may change, possibly in a beneficial way.

[0078] A final consideration of spectral sampling involves the flux used to achieve some specified signal-to-noise. When the source surface brightness is limited, it may be difficult to obtain a sufficient flux level with a single-mode fiber input to the lantern spectrometer. Multiple lantern spectrometers 10 could be used in parallel to increase the total flux observed (see below, e.g., Figure 8 as well as discussion on Spectrum of a Focused Spot), but because each lantern 12 is imaged by a separate set of detector pixels, there will be a read-noise penalty if the surface brightness is too low. In this case, incoherent multi-mode input (e.g., from an incoherent light source) to the lantern spectrometer may be used. The assumption is that the spectral content of each mode in the multi-modal signal is virtually identical at the precision of interest. In this case, each additional input mode will excite its own series of interference patterns, and as the input mode number grows, these patterns will increasingly overlap (but not interfere because they are incoherent). This will degrade the effective spectral sampling densityUCRUZ.001WO / 2024-786-2 PATENT of the lantern spectrometer by some factor, but depending on the nature of the initial sampling, this factor may be less than the number of injected input modes and therefore provide an overall signal-to-noise gain. Without subscribing to any particular scientific theory, in some cases, the different lantern spectrometers 12a-12g would see different respective superpositions of input modes. Furthermore, sampling enhancement schemes that are more compact, such as mapping shape information of the output spatial profiles, could be beneficial because they involve distributing the output light over fewer pixels. In particular, the shape of the optical outputs of the individual cores / channels 14, e.g., the shape of the individual spots of light 31, such as shown for example Figures 3A and 3B, can be used to provide more information regarding the spectral distribution, when the lantern spectrometer design is such that the wavelengths of interest result in few-mode or multi-mode behavior in the output channels. In this case, the machine learning can incorporate the shape of the spots of light, whose wavelength-dependent deformation encodes the few-mode or multi-mode behavior 31, that are output by the different cores / channels 14 at the output end 18 of the photonic lantern and that are imaged by the detector array in the training and in making inferences as to the wavelength spectrum of the incoherent light input into the photonic lantern spectrometer 10. As the output light from the cores / channels 14 at the output end 18 is not single mode, but are multimode, the spots of light 31 and the images thereof captured by the detector array will not be perfectly gaussian but will be distorted and have a distorted shape. This multimode behavior is wavelength dependent. The distortions in shape of the spots 31 contain wavelength information. The spatial shape of the individual spots of light 31 output by the cores 14 of the photonic lantern 12 at the output 18 thereof that are captured by the detector array can be mapped to the spectral inference. Use of the shape of the spots 31 can therefore provide assistance in ascertaining the wavelength spectrum of the light input into the photonic lantern spectrometer 10. Applications Spectrum of a Focused Spot

[0079] Many types of spectroscopy involve analysis of a focused beam or “spot.” This might be because of a desire for a high surface-brightness illumination source, which benefits from concentrating illumination light over a small area. Examples include reflectance,UCRUZ.001WO / 2024-786-2 PATENT transmission, fluorescence, and Raman spectroscopy of small-area and small-volume samples. When microscopy is additionally included, the illumination zones become smaller still. Another example would be the sampling over some distance of a concentrated source such as a laser probe through a liquid or gaseous medium, the beam from a source-light beacon, or indeed the observation of a star through a telescope. Finally, there are examples where the signal is inherently confined to a fiber or waveguide, as expected in telecommunications applications, fiber probe technologies, and nanoscale “laboratories.”

[0080] In these cases, it may be possible to achieve a diffraction-limited focus and inject the source signal directly into the single-mode input of a lantern spectrometer.

[0081] In practice, however, the diffraction limit is often difficult to achieve and regardless, an optimal injection is challenging because of tight tolerances for alignment and mismatches between the delivered beam shape and the shape for optimal coupling. When it is important to gather the increased or maximum flux possible in such cases, a conventional photonic lantern may be used to capture the focused input. This photonic lantern has the job of accepting somewhat multi-modal light and extracting from it single-mode light which it then delivers via the lantern’s output channels. Each of these (single-mode) outputs would then be directed to its own lantern spectrometer. The lantern spectrometer outputs would be bundled or close-packed so that a single camera could image the signals from all lantern spectrometers simultaneously. Alternatively multiple cameras could be used. Also, as discussed herein, the outputs single-mode fiber intensities could be measured by separate individual photodetectors such as individual photodiodes (or photoconductors) for respective individual channels or cores 14.

[0082] In this case, the designs of the injection lantern and the lantern spectrometers will be different, since they are solving different problems. However, because the injection lantern introduces a wavelength dependence to its output, obtaining the final spectrum would involve accounting for this, e.g., through training and analysis of different separate components or the whole end-to-end system. Raman Filtering

[0083] Raman spectroscopy is a particular application of this technology. Raman requires intense illumination from a narrow-band source (typically a laser). One or a fewUCRUZ.001WO / 2024-786-2 PATENT channels of a photonic lantern may be used in reverse to inject illumination light onto a sample, leaving the remaining channels to collect the shifted, Raman signal. See, for example, Figure 14 discussed below. As described above, these collection channels would then feed a set of lantern spectrometers.

[0084] An additional, independent advance can be made with this configuration, regardless of how the Raman stimulation beam is conveyed to the sample. Because the collection channels of the “injection lantern” described above are single-mode, they can be coupled to fiber Bragg filters designed to reject the specific frequency of the illumination laser. Alternatively, if the injection lantern output includes a “fan-out” of single-mode fibers (or waveguides), these may be laser-written to obtain the Bragg filter effect. A rejection filter is common to Raman systems, but using Bragg filters would enable an extremely compact designed that is well suited to mass production. It would replace the need to insert and epoxy a separate filter optic within the optical system, easing manufacture and reducing the risk of scratching and breaking.

[0085] If an injection lantern is not required (see below), single-mode fibers may be used to inject light into lantern spectrometers. Fiber Bragg filters may be incorporated into these single-mode inputs to reject Raman stimulation light. Diffuse Spectroscopy

[0086] In some cases, the source light is diffuse with sufficient surface brightness to make it worthwhile to simply sample the source light with a single-mode fiber which is then directed to a lantern spectrometer. Reflected sunlight would be one example. To increase the signal-to-noise on the final spectrum, a bundle of single-fibers (or a multi-core fiber) could be used to gather the signal and direct it to a series of lantern spectrometers arranged in parallel. Given the small sizes involved (e.g., 50-100 um), many spectrometer outputs could be imaged simultaneously with a single camera / detector package. Alternatively, separate cameras may be employed. It may be desirable to vary the designs of the lantern spectrometers within this packaged array, for example by adjusting the single-mode core sizes. This would allow different lanterns to sample different wavelength ranges and spectral resolutions. Under the assumption that the source spectrum varies negligibly across the inputs, the result would be a final “average” spectrum of the source sampled over a wider wavelength range and at greater spectral resolution than could be achieved by a single lantern spectrometer.UCRUZ.001WO / 2024-786-2 PATENT End-to-End System Considerations

[0087] Regardless of the design and fabrication approach of the photonic lanterns and fiber-routing elements, a complete device may include additional components at the input and output ends. On the input, a lens, lenslet array, or other optic or optics can shape and focus source light into the input channel(s) of the spectrometer. There may be a spacing optic or optics that allows for illumination light to strike a sample at a specified angle and reflect back before it is collected by a focusing fore-optic or optics, perhaps additionally shaping the beam to improve the coupling to single-mode inputs. A filter may be inserted here, e.g., as discussed above, to reject unwanted light (such as the Raman stimulation beam).

[0088] The light output from a lantern spectrometer or lantern spectrometer array is ultimately recorded by a detector, optical detector, photodetector, etc. (e.g., photodiode or photoconductor) or sensor array (e.g., CMOS or CCD) or a plurality of such detectors or detector arrays or a plurality of individual photodetectors such as a plurality of individual photodiodes not included in a photodetector array or photodiode array. (The terms detector, optical detector, photodetector may be used interchangeably herein.) If any dispersion is desired, an optical system may be included before the detector, possibly a collimating lens, a dispersing element, and a focusing optic for focusing onto the detector array (or sensor). If a dispersing element is not needed, it may be possible to abut the sensor array or camera directly to the output face of the lantern spectrometer or spectrometer array. Typical output core diameters are 5-8 um, well suited to 3X sampling by modern CMOS devices with 2-3 um pixels. This would result in an extremely compact design, potentially fitting a powerful spectrometer in a size smaller than a pen cap. Through fiber manufacture and waveguide writing techniques, it may be possible to reduce the size further still to, for example, 1-2 mm, enabling devices suitable for medical wearables, although larger or smaller sized are possible.

[0089] Finally, in addition to read-out electronics and data transfer from the sensor, it may be desirable to include a thermoelectric cooler on the backside of the detector or detectors to reduce dark current and improve performance at low light levels.

[0090] Figure 8 shows a schematic representation of an example end-to-end system 50 for a focused-spot application. Starting from left, an incoming illumination beam represented by the arrow 52 coming from an illumination source such as a laser (not shown) illuminates a sample 54 over a small area. Reflected light 56 passes rightward through a spacer 57 beforeUCRUZ.001WO / 2024-786-2 PATENT encountering converging fore-optics 58 which may focus the beam into the multi-mode input of an “injection lantern” 60. The injection lantern 60 extracts single-mode output from the source light via a plurality of single mode output channels 14 and directs these single mode channels to a series of lantern spectrometers or spectrometer lanterns 12a-12g. The plurality of output channels 14 of the injection lantern 60 may be optical coupled to different ones of the spectrometer lanterns 12a-12g, via an input 20 thereof. The spectrometer lanterns 12a-12g can also have a plurality of output cores or channels 14. The injection and spectrometer lanterns 60, 12a-12g can have different designs. The output of one lantern spectrometer 12c is pictured. Output light is collimated with collimating optics 61, passed through a wavelength dispersing or dispersive optic or element or optics 62 such as a prism or diffractive optical element such as a grating and re-imaged using focusing or imaging optics 63 possibly with demagnification onto a detecting sensor (e.g., a CMOS sensor) 64 as represented by a black bar. Although only one of the photonic lanterns 12c is depicted as having a 2D optical detector array 64, imaging optics 61, 63, and wavelength dispersion optics 62, other ones, most, or potentially all of the photonics lanterns may have respective 2D optical detector arrays 64, imaging optics 61, 63 and wavelength dispersion optics 62. Other configurations are also possible. Additional Demonstrations Water absorption detection

[0091] Figures 9 and 10 demonstrate the ability of the photonic lantern spectrometer 10 to detect molecular absorption signals, specifically the presence of H2O via absorption of a light beam passing through a cuvette filled with water. Figure 9 depicts ground-truth spectra obtained with an off-the-shelf grating spectrometer (Ocean Optic NIRQuest). In particular, Figure 9 shows the emission spectrum 72 for an LED source (blue curve). Figure 9 also shows the emission spectra 74 (orange curve) of light from the LED after passing through an empty cuvette. Figure 9 additionally shows a spectra 76 (green curve) of the light from the LED after passing through the cuvette is filled with water. As illustrated, the water causes significant absorption of the light from the LED.

[0092] Light from the LED passed through the cuvette with and without water in the cuvette (with spectra 76, 74 shown in Figure 9) was injected into the photonic lanternUCRUZ.001WO / 2024-786-2 PATENT spectrometer 10. Figure 10 shows the normalized difference in output photonic lantern intensities obtained by dividing the signal output by the photonic lantern when light passed through the cuvette filled with water (spectra 76 in Figure 9) is injected into the photonic lantern by the signal output by the photonic lantern when light passed through the empty cuvette (spectra 74 shown in Figure 9) is injected into the photonic lantern. The plot in Figure 10 represents different experimental runs 78a, 78b, 78c, 78d and demonstrates that a consistent pattern is observed at the output of the photon lantern for the cuvette and water filled cuvette. This pattern encodes the water absorption detection and can be learned by an ML algorithm. The other points represent small changes to the experimental setup. Spectral recovery of a single label

[0093] The accuracy of the photonic lantern spectrometer was tested using a monochromator having a known wavelength output to control the input spectrum to the photonic lantern spectrometer. Figure 11, in particular, shows the results of neural-network based recovery of peak-emission wavelength measured with the photonic lantern spectrometer. These experimental results illustrate learning of a single label.

[0094] Figure 11 is a plot on axes of predicted wavelength (in nanometers), predicted by the photonic lantern spectrometer, and true wavelength (in nanometers), measured by a reference grating spectrometer that shows successful recovery of input wavelengths of narrow lines. Both a curve 82 plotting the predicted wavelength (vertical or Y-axis) versus true wavelength (horizontal or X-axis) is shown together with a line 84 corresponding to plot of the line Y = X. A monochromator was used to generate a set of narrow “emission” lines (~0.5 nm full width half maximum, FWHM) from 850-900 nm. The monochromator output was fed into a single-mode fiber 22 in order to spatially filter the light before injection into the photonic lantern spectrometer. Some of the output was also directed into an Ocean Optic NIRQuest grating spectrometer. Analysis of the spectra from this grating spectrometer provided the “truth” data set.

[0095] The neural network was trained on every other data point, labeled simply by the wavelength of the emission peak. Figure 11 illustrates how all wavelengths are recovered with reasonable accuracy. Accordingly, Figure 11 shows how a neural network architecture canUCRUZ.001WO / 2024-786-2 PATENT be trained and recover spectral information, demonstrating how the photonic lantern spectrometer acts as a “reconstructive spectrometer.” New Applications Cost benefit of limited-parameter recovery

[0096] In many sensing applications, the spectroscopic signal of interest can be modeled with just a few parameters. For example, the water absorption signal shown in Figure 9 depends primarily on the path-integrated concentration of H2O in the beam, as well as the ambient pressure and temperature. Additional parameters could account for systematics in the instrumentation (e.g., a varying sensitivity curve) and the presence of other absorbing species.

[0097] In these cases, an ML algorithm coupled with our photonic lantern spectrometer can be trained directly on the data in order to learn the parameters of interest. There is no need to first reconstruct the spectrum and then fit that spectrum with a model. With sufficient training data, this approach will yield a more accurate and precise result.

[0098] However, a primary benefit may be cost. By eliminating the need to record a complete spectrum, uses of (often expensive) detector arrays need not be required. Instead, less expensive photodiodes or photodetectors can be coupled directly to the photonic lantern output channels 14. A relatively small number of such photodiodes or photodetectors or optical detectors as compared to the number of photodiodes in a CCD or CMOS 2D detector array. For example, as few as 3 (or as many as several hundred) photodiodes or photoconductors or photodetectors or optical detectors may be employed. However, in many instances, this reduced number of photodetectors / optical detectors is less than that in a 2D detector array used in a camara.

[0099] Cost savings could be significant in regimes where detector arrays are expensive (e.g., infrared wavelengths). The use of photodiode-coupled lantern spectrometers for gas sensing is also an alternative to expensive tunable diode lasers (TDLs) that must maintain extremely fine wavelength precision in order to sample particular spectral features.

[0100] Accordingly, any of the systems, devices, methods, designs, and implementations, described herein can employ individual photodiodes or photodetectors or optical detectors or detectors as opposed to a 2D detector array. In some cases, a linear, 1D arrayUCRUZ.001WO / 2024-786-2 PATENT may be employed. In some cases, a smaller 2D array, such as a 2x2, 3x3, 4x4, 2x3, 2x4, 3x4, etc. array may be employed that is smaller than used in imaging cameras. Accordingly the number of photodetectors or photodiodes or photoconductors (or optical detector or detectors) may be less than 4000, 3000, 2000, 1500, 1000, 800, 600, 500, 400, 300, 200, 100, 75, 50, 25, 20, 15, 12, 10, 8, 6, 5, 4, 3, or 2, or any range formed by any of these values, may be used, in some implementations. Gas sensor

[0101] Many gaseous molecular species of interest have prominent absorption features between 1-15 micrometers or microns. For the reasons described above, our photonic lantern spectrometer 10 can be used to enable sensing of these gases at reduced cost. These gases include greenhouse gases such as H2O, CH4, CO2, and N2O, and many other gases that are valuable to industry and sometimes hazardous (e.g., NH3, HCN, HF, HCl). Thus, a gas sensor utilizing the photonic lantern spectrometer 10 could be used to measure gas products, improve or optimize processing, and detect leaks.

[0102] In an outdoor setting, the gas sensor combined with micrometeorological data and modeling could be used to measure diffuse gas fluxes (e.g., flows of CO2or CH4). Wavefront and Turbulence sensing

[0103] As discussed above, the spectral signal from the photonic lantern 12 can be obtained by spatially filtering the injected light (e.g., by passing it through a single-mode fiber). However, it is possible to model both the spectral signal and variations in the input wavefronts at the same time if the spatial mode filter is removed, if, for example, the model can be sufficiently constrained by the number of output channels 14 in the lantern 12. As referred to above, wavefront sensing with photonic lanterns 12 has been described in the literature.

[0104] For example, the photonic lantern spectrometer 10 can be used to measure gas concentrations as well as the level of turbulence along the line of sight (because turbulence distorts wavefronts). This could be valuable for leak detection. In the case of hazardous gas leaks, a turbulence sensor can be validated by test emission of non-hazardous gases.

[0105] Another example is the ability to simultaneously determine surface texture (via wavefront distortion) and composition through reflection spectroscopy. A sensor under aUCRUZ.001WO / 2024-786-2 PATENT car may, for example, be able to detect both road surface texture and distinguish the presence of water, oil, and ice. Example Detection Configurations

[0106] A variety of different system designs are possible for obtaining output from the photonic lantern 12. As discussed above, for example, a detector array such as a two- dimensional (2D) detector array such as a CCD or CMOS detector or sensor array may be employed. Figure 12A, for example, shows a detector array such as a two-dimensional (2D) detector array configured to capture an image of the output 18 of the photonic lantern 12. A lens 86, for example, is shown forming an image of the output 18 of the photonic lantern 12 such as an image of the output face or facet 26 of the photonic lantern onto an 2D detector array 88. In particular, light output from the cores or channels 14 of the photonic lantern 12 is imaged onto the 2D detector array 88 by the lens 86. Images such as shown in Figures 3A or 3B comprising a plurality of spots of light 31 may, for example, be formed on the detector array 88 by the lens 88. Although a single positive lens 88 is shown, other designs are possible. For example, a plurality of lenses may be employed to image the output of the photonic lantern 12 onto a detector array 88. Accordingly, the lens or lenses or other optics 86 employed to form an image onto the detector array 88 may be referred to as imaging or focusing optics. The combination of the imaging or focusing optics 86 and the 2D detector array 88 may be referred to as a camera.

[0107] Other configurations, however, are possible. Figure 12B, for example, shows the 2D detector array 88 butt-coupled to the output 18, e.g., output face or facet 26, of the photonic lantern. The detector array 88 may for example be positioned close to, adjacent to and / or in contact with the output face or facet 26 of the photonic lantern 12 to collect light from the various cores or channels 14 in the photonic lantern.

[0108] In various implementations, the 2D detector array 88, such as shown in Figures 12A and 12B may comprises more than 100 pixels. The 2D detector array 88, may comprise, for example, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1200, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 6000, 8000 or more pixels, or any range formed by any of these values or possible more or less pixels.

[0109] Figure 12C shows a configuration with separate photodetectors or photosensors, such as photodiodes, 90a, 90b, 90c not included in an image sensor such as a two-UCRUZ.001WO / 2024-786-2 PATENT dimensional detector array configured for imaging (e.g., with more than 100 pixels), configured to receive light output by the photonic lantern 12. In particular, photodetectors (e.g., photodiodes) 90a, 90b, 90c are configured to receive light from the individual the cores or channels 14 of the photonic lantern 12. In the example shown, separate optical fibers 32a, 32b, 32c extend to the photodetectors 90a, 90b, 90c, respectively. In some implementations, this optical fiber 32a, 32b, 32c may be the optical fiber of the photonic lantern 12. For example, a portion of the optical fiber used to form the tapered fiber bundle 12 may not be tapered. These fibers 32a, 32b, 32c that extend from the tapered portion may be optically coupled to the individual respective photodetectors, 90a, 90b, 90c.

[0110] Although the photodetectors 90a, 90b, 90c are optically coupled to the different channels 14 using optical fiber 32a, 32b, 32c, other designs are possible. For example, the photodetectors 90a, 90b, 90c may be butt-coupled to the optical channels 14 of the fiber bundle 12. The photodetectors 90a, 90b, 90c may, of example, close to, adjacent to and / or in contact with the output face or facet 26 and / or the cores / channels 14 of the photonic lantern 12.

[0111] Similarly, instead of the optical fiber 32a, 32b, 32c comprising optical fiber of the tapered fiber bundle 12, separate optical fibers could be optically coupled to the channels 14 of the photonic lantern. Individual detectors may be butt-coupled to the separate optical fibers.

[0112] As discussed above, employing individual photodetectors 90a, 90b, 90c may be less expensive than an imaging sensor comprising a 2D detector array 88 (e.g., with more than 10, 20, 30, 40, 50, 60, 80, 90, 100, 150, 200, 400, 800, 1000, 1200, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 6000, 8000, or more pixels or any range formed by any of these values or possibly larger). Although single photodetectors 90a, 90b, 90c are optically coupled to individual channels 14 of the photonic lantern 12 in Figure 12B, in other designs more than a single photodetector may be optically coupled to an individual channel. Nevertheless, having photodetectors 90a, 90b, 90c not integrated in an imaging sensor such as a larger detector array 88 such as for an imaging sensor (e.g., imaging sensor array) such as separate and / or individual photodetectors may be less expensive than a larger 2D sensor array such as for an imaging sensor (e.g., imaging sensor array). The number of photodetectors therefore may be less than used in such 2D sensor arrays such as CCD or CMOS imaging sensors such as used in a CCD or CMOS camera, and may, for example, be less than 100 pixels (e.g., less than 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1200, 1500, 2000, 2500, 3000, 3500, 4000, 4500 pixels). Accordingly, theUCRUZ.001WO / 2024-786-2 PATENT number of photodetectors configured to receive output from a channel (e.g., single output channel) may be in some examples less than 100, 90, 80, 70, 60, 50, 40, 30, 25, 20, 15, 12, 10, 9, 8, 6, 5, 4, 3, 2, or 1 or any range form by any of these values. For example, the number of photodetectors configured to receive output from a channel (e.g., single output channel) may be in some examples no more than 20, 18, 15, 14, 12, 10, 9, 8, 6, 5, 4, 3, 2, or 1 or any range form by any of these values. Waveguide Writing

[0113] Although fiber bundles photonic lantern spectrometers 10 comprising tapered fiber bundles 12 are described above, the photonic lantern is not limited to waveguides comprising optical fiber. The waveguide may comprise a waveguide core 14 within bulk cladding material 92 such within a substrate, layer, or medium of another shape (see, e.g., Figures 13A and 13B). Such a waveguide 32 may be produced by three-dimensional (3D) laser writing such as, for example, using ultrafast laser inscription (ULI). See, for example, “Ultrafast laser inscription of an integrated photonic lantern”, Thomson et al, Optics Express, Vol.19, Issue 6, pp.5698-5705 (2011), https: / / doi.org / 10.1364 / OE.19.005698. As illustrated in Figure 13A, in some implementations a laser such as a pulsed laser, e.g., a femtosecond pulsed laser, may output a laser beam 96 that is directed to the substrate or layer of material 92. In some implementations, the laser beam 96 may be focused, for example, using focusing optics such as a microscope objective 98. The laser beam 96 may be focused down to a location within the substrate or layer of material 92.

[0114] In one example of ultrafast laser inscription, extremely short (e.g., 10−13s) pulses of light are focused within the volume of a dielectric material that is transparent at the wavelength of the incident light pulses. Without subscribing to any particular scientific theory, in some cases a nonlinear light matter interaction occurs at the laser focus. As a result, the refractive index of the material 92 may be altered enabling the formation of optical waveguide structures within the material. Thus, if the material (e.g., substrate or medium) 92, laser 94, laser beam 96 or any combination thereof is translated in three dimensions such that the laser focus moved relative to the material, arbitrary shaped optical waveguides 32 and microstructure may be fabricated, etched or inscribed within the volume of the material. In this manner, higher indexUCRUZ.001WO / 2024-786-2 PATENT waveguide core regions 14 may be formed in the lower index “cladding” 16, e.g., bulk cladding material 92.

[0115] Other types of laser inscription and laser writing techniques are possible and can be used. The method of forming the photonic lantern 12 in a substrate or layer of material 92 is thus not limited to those described herein such as ultrafast laser inscription. In various implementations, nevertheless, the index of refraction of a portion of the material 92 can be altered, e.g., increased, with respect to other portions of the material. Other methods, however, are possible. For example, potentially the waveguide could be written by lowering the refractive index in the cladding region as compared to the waveguide core region. Additionally, although the substrate or layer of material 92 is shown as a rectangular slab the host material or medium may have other shapes (e.g., three-dimensional shapes).

[0116] Accordingly, in various implementations, photonic lanterns 12 such as described herein may be formed in a substrate or layer of material 92 such a transparent dielectric using laser writing (e.g., ultrafast laser inscription). Likewise, any feature or combination of features described above with respect to photonic lantern spectrometers 10 and photonic lanterns 12 comprising optical fiber such as tapered fiber bundles may be implemented in waveguides 32 included with cladding material such as formed by laser writing.

[0117] Figure 13B, for example, illustrates a photonic lantern 12 and / or a photonic lantern spectrometer 10 within a substrate 92 that may be formed in three dimensions using laser waveguide writing. The photonic lantern spectrometer 10 comprises a plurality of waveguides 32 that are tapered such that the photonic lantern 12 has a smaller input end 20 and a larger output end 18. The waveguides 32 comprise core regions 14 that may, for example, have a higher index than the adjacent cladding 16. The photonic lantern 12 shown in Figure 13B, comprising a plurality of tapered waveguides (e.g., a tapered waveguide bundle) 32, may operate in a manner such as described above with respect to the photonic lantern comprising a tapered fiber bundle. Similarly, a single mode waveguide 28 is included at the input 20 of the photonic lantern 12.

[0118] As illustrated in Figure 13B, a lens or imaging optics 86 and a 2D optical detector array 88 such as a CCD or CMOS detector array are positioned with respect to the output face 26 of the photonic lantern 12 to obtain images of the output 18 of the photonic lantern 12 such as described above with respect to the fiber based photonic lantern discussedUCRUZ.001WO / 2024-786-2 PATENT above and shown, for example, in Figure 12A. Other configurations such as any of the other configurations discussed herein with respect to fiber based photonic lanterns 12 and photonic lantern spectrometers 10 may be employed such as the configurations shown in Figures 12B and 12C. Similarly, cascaded systems 40 such as shown in Figure 7 as well as plurality of photonic lanterns 12a-12g in combination with a plurality of tapered waveguides or a tapered waveguide bundle (e.g., injection lantern 60) such as shown in Figure 8 may employ waveguides 32 formed in a substrate or layer of cladding material 92 formed by laser waveguide writing techniques. And although the waveguide 92 written in the cladding substrate or layer 92 do not comprise optical fiber, the photonic lanterns 12, photonic lantern spectrometers 10, and other components therein may operate and / or be configured similarly to the fiber based photonic lanterns, photonic lantern spectrometer and fiber components such as described herein. Additional Example Configurations

[0119] Figure 14 demonstrates that a wide range modifications and variations may be included systems that employ photonic lanterns and photonic lantern spectrometers. For example, in the system shown in Figure 14 includes an injection lantern 64 optically coupled to a plurality of photonic lantern 12a, 12b, 12c, 12d, 12e, by a plurality of optical fiber 32a, 32b, 32c, 32d, 32e. In some implementations, these optical fibers 32a, 32b, 32c, 32d, 32e can be part of the optical fibers 32 used to form the tapered optical fiber bundle forming the photonic lantern 12 that is the injection lantern 64. As shown, the system further comprises a light source (e.g., laser) that is optically coupled to one or more channels 14 of the injection lantern 12, in this example, via one of the optical fiber 32a. In particular, the light source (e.g., laser) 94 is coupled to the injection lantern 64 at the second larger end 18 thereof such that light from the light source propagates within the lantern to the first smaller end 20. This light exits the smaller end 20 and illuminates the sample 50. Light from the sample 54 is then collected by a plurality of channels 14 at the smaller (input) end of the injection lantern 64 and propagates to the larger (output) end 18 thereof. Light from respective ones of the channels or cores 14 are coupled into respective photonic lanterns 12a, 12b, 12c, 12d, 12e via the optical fibers 32a, 32b, 32c, 32e. Accordingly, light from the light source 94 travels in the reverse direction through the injection lantern 64 to smaller end 20 and to the sample 54, while light from the sample travels in the forward directionUCRUZ.001WO / 2024-786-2 PATENT through the injection lantern to the larger end and the plurality of lanterns 12a, 12b, 12c, 12d, 12e.

[0120] In the example shown, the plurality of optical fibers 32a, 32b, 32d, 32c, 32e include one or more optical filters 96 such as one or more bandpass filters, blocking filters, notch filters, long pass filters, short pass filters, etc. These filters 96 may comprise a Bragg grating. The Bragg filters 96 may in some cases be formed (e.g., inscribed) in the optical fibers 32a, 32b, 32d, 32c, 32e. Alternatively, the optical filters, e.g., Bragg filters, 96 may be separate from the optical fibers 32a, 32b, 32d, 32c, 32e.

[0121] In applications such as Raman spectroscopy, the optical filter (e.g., Bragg filters) may attenuate one or more wavelengths of light from the light source. The optical filter may, for example, comprise a notch filter having a notch at the output wavelength of the light source, e.g., of the laser.

[0122] Figure 14 also shows collection optics 58 collecting light from the sample 54 and coupling the light into the input 20 of the injection lantern 60 as well as an optical sensor array 64 receiving light output by one of the photonics lanterns 12c. Imaging optics 61, 63 are shown forming an image of the output face 26 of the photonic lantern 12c onto the 2D optical detector array 64 or otherwise collecting light from the output 18 of the photonic lantern 12c. Wavelength dispersion optics (e.g., a prism) 62 are shown in the optical path between the output 18 of the photonic lantern 12c and the optical detector array 64. Although only one of the photonic lanterns 12c is depicted as having a 2D optical detector array 64, imaging optics 61, 63, and wavelength dispersion optics 62, other ones, most, or potentially all of the photonics lanterns may have respective 2D optical detector arrays 64, imaging optics 61, 63 and wavelength dispersion optical 62. Other configurations are also possible. Examples

[0123] The following is a numbered list of example implementation that are within the scope of this disclosure. The examples that are listed should in no way be interpreted as limiting the scope of the potential implementations. Various features of the examples that are listed can be removed, added, or combined to form additional implementations, which are part of this disclosure.UCRUZ.001WO / 2024-786-2 PATENT Part 1A 1. A system for determining a spectral distribution of light, said system comprising: a photonic lantern comprising a tapered waveguide bundle comprising a plurality of core regions comprising a plurality of respective output channels of said photonic lantern, said tapered waveguide bundle having a smaller input end having an input for injection of said light and a larger output end having an output through which said light is emitted from at least one of said output channels depending on the spectral distribution of said light, said core regions being closer together at said input end than said output end such that said core regions undergo cross-talk that, depending on said spectral distribution, will determine which of said at least one of said output channel will emit light; and a plurality of optical detectors configured to sense light emitted from said output. 2. The system of Example 1, further comprising electronics configured to determine said spectral distribution from said light emitted from said output and sensed by said plurality of optical detectors. 3. The system of Example 1 or 2, wherein said optical detectors comprise photodetectors not included in an imaging sensor. 4. The system of Example 3, wherein said photodetectors comprise photodiodes. 5. The system of Example 3 or 4, wherein said photodetectors are optically coupled to respective channels of said photonic lantern. 6. The system of any of Examples 3-5, wherein said photodetectors are butt coupled to respective core regions. 7. The system of any of Examples 3-5, further comprising a dispersive optical element between said core region and said photodetector. 8. The illumination system of Example 1, wherein said plurality of optical detectors comprise at least one detector array configured to capture an image using said light emitted from said output, said captured image dependent on the light emitted from said at least one of said core regions that outputs light. 9. The system of Example 8, further comprising electronics configured to determine said spectral distribution from one or more images captured by said detector array using said light emitted from said output.UCRUZ.001WO / 2024-786-2 PATENT 10. The system of Example 8 or 9, further comprising a lens configured to image said output onto said detector array. 11. The system of any of Examples 8-10, further comprising a wavelength dispersive optical element positioned in an optical path between said output and said detector array. 12. The system of Example 11, further comprising a lens configured to form an image onto said detector array using said light from said wavelength dispersive element. 13. The system of Example 11 or 12, further comprising a collimating lens configured to increase collimation of light from said output directed to said wavelength dispersive optical. 14. The system of any of Examples 11-13, wherein said wavelength dispersive element comprises a prism or grating. 15. The system of any of Examples 8-14, further comprising at least one additional photonic lantern in parallel with said photonic lantern, said at least one additional photonic lantern comprising: a tapered waveguide bundle comprising a plurality of core regions comprising a plurality of respective output channels of said photonic lantern, said tapered waveguide bundle having a smaller input end having an input for injection of said light and a larger output end having an output through which said light is emitted from at least one of said output channels depending on the spectral distribution of said light, said core regions being closer together at said input end than said output end such that said core regions undergo cross-talk that, depending on said spectral distribution, will determine which of said at least of said output channels will emit light. 16. The system of Example 15, wherein said detector array is configured to capture an image using said light emitted from said output of said at least one additional photonic lantern, said captured image dependent on the light emitted from said at least one of said output channels that output light from said at least one additional photonic lantern. 17. The system of Example 15, further comprising an additional detector array configured to capture an image using said light emitted from said output of said at least one additional photonic lantern, said captured image dependent on the light emitted from said at least one of said output channels that output light from said at least one additional photonic lantern. 18. The system of any of the Examples 8-17, wherein said detector array abuts the output of the tapered waveguide bundle.UCRUZ.001WO / 2024-786-2 PATENT 19. The system of any of the Examples 8-18, wherein said detector array comprises a CMOS or CCD detector array. 20. The system of any of Examples 1-7, further comprising at least one additional photonic lantern in parallel with said photonic lantern, said at least one additional photonic lantern comprising: a tapered waveguide bundle comprising a plurality of core regions comprising a plurality of respective output channels of said photonic lantern, said tapered waveguide bundle having a smaller input end having an input for injection of said light and a larger output end having an output through which said light is emitted from at least one of said output channels depending on the spectral distribution of said light, said core regions being closer together at said input end than said output end such that said core regions undergo cross-talk that, depending on said spectral distribution, will determine which of said at least of said output channels will emit light. 21. The system of Example 20, wherein said plurality of optical detectors is configured sense light emitted from said output of tapered waveguide bundle of said at least one additional photonic lantern. 22. The system of Example 20, further comprising additional optical detectors configured to sense light emitted from said output of tapered waveguide bundle of said at least one additional photonic lantern. 23. The system of any of the examples above, further comprising an optical filter configured to filter light coupled into said input. 24. The system of any of the examples above, wherein said optical filter comprises a bandpass filter. 25. The system of any of the examples above, wherein said optical filter comprises a notch filter. 26. The system of any of the examples above, wherein said optical filter comprises a high pass filter or a low pass filter. 27. The system of any of the examples above, wherein said optical filter comprises a Bragg grating. 28. The system of any of the examples above, further comprising an optical fiber configured to couple light into said input.UCRUZ.001WO / 2024-786-2 PATENT 29. The system of Example 28, wherein said optical fiber comprises a single mode fiber. 30. The system of Example 28 or 29, wherein said optical fiber comprises a fiber Bragg grating. 31. The system of any of Examples 28-30, wherein said an optical fiber configured to couple light into said input is misaligned with respect to an effective core formed by said closely spaced plurality of core regions at said input. 32. The system of any of Examples 28-31, wherein said an optical fiber configured to couple light into said input is at an average non-zero angle with respect to said input end. 33. The system of any of the examples above, further comprising collecting optic configured to couple light into said input. 34. The system of Example 33, wherein said collecting optic comprises a lens. 35. The system of any of the examples above, wherein said at least two of said core regions have different sizes. 36. The system of any of the examples above, wherein said at least two of said core regions have different indices of refraction. 37. The system of any of the examples above, wherein said at least two of said core regions comprise different glass. 38. The system of any of the examples above, further comprising a cooler to cool said detector array. 39. The system of any of the examples above, wherein said plurality of core regions at said input operate as a single mode waveguide. 40. The system of any of the examples above, wherein said tapered waveguide bundle comprises a tapered fiber bundle. 41. The system of any of the examples above, wherein said core regions comprise cores. 42. The system of any of the examples above, wherein said input comprises an input face of said tapered waveguide bundle. 43. The system of any of the examples above, wherein said output comprises an output face of said tapered waveguide bundle.UCRUZ.001WO / 2024-786-2 PATENT 44. The system of any of the Examples 1-39, wherein said core regions are in a substrate or layer of cladding material. 45. The system of any of the Examples 1-39, wherein said tapered waveguide bundle is formed by 3D laser printing, laser inscribing or laser waveguide writing. Part 1B 1. A system for determining a spectral distribution of light, said system comprising: a first stage comprising a photonic lantern comprising a tapered waveguide bundle comprising a plurality of core regions comprising a plurality of respective output channels of said photonic lantern, said tapered waveguide bundle having a smaller input end having an input for injection of said light and a larger output end having an output through which said light is emitted, said core regions being closer together at said input end than said output end; and a second stage comprising a plurality of photonic lanterns comprising tapered waveguide bundles, each comprising a plurality of core regions comprising a plurality of respective output channels of said photonic lantern, said tapered waveguide bundles having smaller input ends having inputs for injection of said light and larger output ends having outputs through which said light may be emitted from one or more of said channels depending on the spectral distribution of said light, said core regions being closer together at said input ends than said output ends such that said core regions undergo cross-talk that, depending on said spectral distribution, will determine whether said output channel will emit light from said output, wherein different channels of said photonic lantern in said first stage at said output are optically coupled to different of said plurality of photonic lanterns of said second stage. 2. The system of Example 1, further comprising a plurality of optical detectors configured to sense light emitted from said outputs of said plurality of photonic lanterns in said second stage. 3. The system of Example 2, further comprising electronics configured to determine said spectral distribution from said light emitted from said outputs and sensed by said plurality of optical detectors.UCRUZ.001WO / 2024-786-2 PATENT 4. The system of Example 2 or 3, wherein said optical detectors comprise photodetectors not included in an imaging sensor. 5. The system of Example 4, wherein said photodetectors comprise photodiodes. 6. The system of Example 4 or 5, wherein said photodetectors are optically coupled to respective channels of said photonic lanterns in said second stage. 7. The system of any of Examples 4-6, wherein said photodetectors are butt coupled to respective core regions of said photonic lanterns in said second stage. 8. The system of any of Example 4-6, further comprising a dispersive optical element between said core region and said photodetector. 9. The illumination system of Example 2, wherein said plurality of optical detectors comprises at least one detector array configured to capture one or more images using said light emitted from one or more of said outputs of said photonic lanterns in said second stage, said captured image dependent on the light emitted from said at least one of said optical channels that outputs light. 10. The system of Example 9, further comprising electronics configured to determine said spectral distribution from one or more images captured by said one or more detector arrays using said light emitted from said output. 11. The system of Example 9 or 10, further comprising one or more lenses configured to image said output onto said one or more detector arrays. 12. The system of any of Examples 9-11, further comprising a wavelength dispersive optical element positioned in an optical path between said output and said detector array. 13. The system of Example 12, further comprising a lens configured to form an image onto said detector array using said light from said wavelength dispersive element. 14. The system of Example 12 or 13, further comprising a collimating lens configured to increase collimation of light from said output directed to said wavelength dispersive optical. 15. The system of any of Examples 12-14, wherein said wavelength dispersive element comprises a prism or grating. 16. The system of any of Examples 9-15, wherein said at least one detector array comprises a detector array configured to capture an image using said light emitted from said outputs of a plurality of said photonic lanterns in said second stage, said captured imageUCRUZ.001WO / 2024-786-2 PATENT dependent on the light emitted from said at least one of said output channels that output light from said plurality of photonic lantern. 17. The system of any of Examples 9-15, said at least one detector array comprises a plurality of detector arrays configured to capture images using said light emitted from said output of a plurality of photonic lantern in said second stage, said captured images dependent on the light emitted from said at least one of said output channels that output light from said plurality of photonic lanterns in said second stage. 18. The system of any of the Examples 9-17, wherein said detector array abuts the output of the tapered waveguide bundle. 19. The system of any of the Examples 9-18, wherein said detector array comprises a CMOS or CCD detector array. 20. The system of any of the examples above, further comprising an optical filter configured to filter light coupled into the input of said tapered waveguide bundle of a photonic lantern in said second stage. 21. The system of any of the examples above, wherein said optical filter comprises a bandpass filter. 22. The system of any of the examples above, wherein said optical filter comprises a notch filter. 23. The system of any of the examples above, wherein said optical filter comprises a high pass filter or a low pass filter. 24. The system of any of the examples above, wherein said optical filter comprises a Bragg grating. 25. The system of any of the examples above, further comprising an optical fiber configured to couple light into the input of said tapered waveguide bundle of a photonic lanterning said second stage. 26. The system of Example 25, wherein said optical fiber comprises a single mode fiber. 27. The system of Example 25 or 26, wherein said optical fiber comprises a fiber Bragg grating.UCRUZ.001WO / 2024-786-2 PATENT 28. The system of any of Examples 25-27, wherein said an optical fiber configured to couple light into said input is misaligned with respect to an effective core formed by said closely spaced plurality of core regions at said input. 29. The system of any of Examples 25-28, wherein said an optical fiber configured to couple light into said input is at an average non-zero angle with respect to said input end. 30. The system of any of the examples above, further comprising collecting optic configured to couple light into the input of said tapered waveguide bundle of said photonic lantern in said first stage. 31. The system of Example 30, wherein said collecting optic comprises a lens. 32. The system of any of the examples above, wherein said at least two of said core regions in a tapered waveguide bundle in a photonic lantern in said second stage have different sizes. 33. The system of any of the examples above, wherein said at least two of said core regions in a tapered waveguide bundle in a photonic lantern in said second stage have different indices of refraction. 34. The system of any of the examples above, wherein said at least two of said core regions in a tapered waveguide bundle in a photonic lantern in said second stage comprise different glass. 35. The system of any of the examples above, further comprising a cooler to cool said detector array. 36. The system of any of the examples above, wherein said plurality of core regions at said input of a tapered waveguide bundle of a photonic lantern in said second stage operates as a single mode waveguide. 37. The system of any of the examples above, further comprising a light source coupled to at least one channel at the output of the photonic lantern in the first stage such that light propagates in the reverse direction from said output to said input of said tapered waveguide bundle of siad photonic lantern in said first stage. 38. The system of any of the examples above, wherein said light source comprises a laser. 39. The system of any of the examples above, wherein said tapered waveguide bundles comprise tapered fiber bundles.UCRUZ.001WO / 2024-786-2 PATENT 40. The system of any of the examples above, wherein said core regions comprise cores. 41. The system of any of the examples above, wherein said inputs comprises input faces of said tapered waveguide bundles. 42. The system of any of the examples above, wherein said outputs comprises output faces of said tapered waveguide bundles. 43. The system of any of the examples above, wherein said core regions are in a substrate or layer of cladding material. 44. The system of any of the examples above, wherein said tapered waveguide bundles are formed by 3D laser printing, laser inscribing or laser waveguide writing. Part 1C 1. A method implemented by a system of one or more processors, the method comprising: obtaining information reflecting output from one or more photonic lanterns, wherein a photonic lantern comprises a tapered waveguide bundle comprising a plurality of core regions comprising a plurality of respective output channels of said photonic lantern, said tapered waveguide bundle having a smaller input end having an input for injection of light and a larger output end having an output through which said light is emitted from at least one of said optical channels depending on the spectral distribution of said light, said cores being closer together at said input end than said output end such that said core regions undergo cross-talk that, depending on said spectral distribution, will determine which of said at least one of said optical channels will emit light; computing a forward pass of the obtained information through a machine learning model; and obtaining the spectral distribution based on the machine learning model. 2. The method of Example 1, wherein the obtained information the output from the at least one said optical channels. 3. The method of Example 1, wherein the obtained information is one or more images depicting spots associated with output from the at least one said optical channels.UCRUZ.001WO / 2024-786-2 PATENT 4. The method of Example 1, wherein the obtained information includes the shape of spots associated with output from the at least one said optical channels. 5. The method of Example 1, wherein the machine learning model is a neural network. 6. The method of Example 5, wherein the neural network is a convolutional neural network. 7. The method of Example 5, wherein the neural network is a fully-connected network. 8. The method of Example 1, wherein the machine learning model is trained based on a training data set comprises individual images and associated ground truth, wherein the ground truth reflects a vector of values, and wherein each value is associated with an individual frequency within a frequency range. 9. The method of Example 8, wherein individual values that form a vector of values are individual intensity values associated with an individual frequency within the frequency range. 10. The method of Example 1, wherein the spectral distribution is a vector of values, wherein each value is associated with an individual frequency within a frequency range. 11. The method of any of the examples above, wherein obtaining information reflecting output from one or more photonic lanterns comprises sensing light from different of the optical channels using optical detectors comprising photodetectors not included in an imaging sensor. 12. The method of Example 11, wherein said photodetectors comprise photodiodes. 13. The method of any of the examples above, wherein obtaining information reflecting output from one or more photonic lanterns comprising sensing light from different of the optical channels using at least one detector array configured to capture an image using said light emitted from said output, said captured image dependent on the light emitted from said at least one of said optical channels that outputs light. 14. The method of any of the examples above, wherein said tapered waveguide bundles comprise tapered fiber bundles. 15. The method of any of the examples above, wherein said core regions comprise cores.UCRUZ.001WO / 2024-786-2 PATENT 16. The method of any of the examples above, wherein said inputs comprises input faces of said tapered waveguide bundles. 17. The method of any of the examples above, wherein said outputs comprises output faces of said tapered waveguide bundles. 18. The method of any of the examples above, wherein said core regions are in a substrate or layer of cladding material. 19. The method of any of the examples above, wherein said tapered waveguide bundles are formed by 3D laser printing, laser inscribing or laser waveguide writing. Part 2 1. A system for determining a spectral distribution of light, said system comprising: a photonic lantern comprising a tapered fiber bundle comprising a plurality of cores, said tapered fiber bundle having a smaller input end having an input face for injection of said light and a larger output end having an output face through which said light is emitted from at least one of said cores depending on the spectral distribution of said light, said cores being closer together at said input end than said output end such that said cores undergo cross-talk that, depending on said spectral distribution, will determine which of said at least one of said cores will emit light; and a detector array configured to capture an image using said light emitted from said output face, said captured image dependent on the light emitted from said at least one of said cores that outputs light. 2. The system of Example 1, further comprising electronics configured to determine said spectral distribution from an image captured by said detector array using said light emitted from said output face. 3. The system of Example 1 or 2, further comprising a lens configured to image said output face onto said detector array. 4. The system of Example 1 or 2, wherein said detector array abuts said output face. 5. The system of Example 1 or 2, further comprising a wavelength dispersive optical element positioned in an optical path between said output face and said detector array. 6. The system of Example 5, further comprising a lens configured to form an image onto said detector array using said light from said wavelength dispersive element.UCRUZ.001WO / 2024-786-2 PATENT 7. The system of Example 5 or 6, further comprising a collimating lens configured to increase collimation of light from said output face directed to said wavelength dispersive optical. 8. The system of any of Examples 5-7, wherein said wavelength dispersive element comprises a prism or grating. 9. The system of any of Examples 1-8, further comprising an optical fiber configured to couple light into said input face. 10. The system of Example 9, wherein said optical fiber comprises a single mode fiber. 11. The system of Example 9 or 10, wherein said optical fiber comprises a fiber Bragg grating. 12. The system of any Examples 1-8, further comprising collecting optic configured to couple light into said input face. 13. The system of Example 12, wherein said collecting optic comprises a lens. 14. The system of any of the examples above, wherein said detector array comprises a CMOS or CCD detector array. 15. The system of any of the examples above, further comprising at least one additional photonic lantern in parallel with said photonic lantern, said at least one additional photonic lantern comprising: a tapered fiber bundle comprising a plurality of cores, said tapered fiber bundle having a smaller input end having an input face for injection of said light and a larger output end having an output face through which said light is emitted from at least one of said cores depending on the spectral distribution of said light, said cores being closer together at said input end than said output end such that said cores undergo cross-talk that, depending on said spectral distribution, will determine which of said at least of said cores will emit light. 16. The system of Example 15, wherein said detector array is configured to capture an image using said light emitted from said output face of said at least one additional photonic lantern, said captured image dependent on the light emitted from said at least one of said cores in said plurality of cores that output light from said at least one additional photonic lantern. 17. A system for determining a spectral distribution of light, said system comprising:UCRUZ.001WO / 2024-786-2 PATENT a first stage comprising a photonic lantern comprising a tapered fiber bundle comprising a plurality of cores, said tapered fiber bundle having a smaller input end having an input face for injection of said light and a larger output end having an output face through which said light is emitted, said cores being closer together at said input end than said output end; and a second stage comprising a plurality of photonic lanterns comprising tapered fiber bundles, each comprising a plurality of cores, said tapered fiber bundles having smaller input ends having input faces for injection of said light and larger output ends having output faces through which said light may be emitted from one or more of said cores depending on the spectral distribution of said light, said cores being closer together at said input ends than said output ends such that said cores undergo cross-talk that, depending on said spectral distribution, will determine whether said cores will emit light from said output face, wherein different cores of said photonic lantern in said first stage at said output face are optically coupled to different of said plurality of photonic lanterns of said second stage. 18. The system of Example 17, further comprising at least one detector array configured to capture an image using said light emitted from said output face of at least one of said plurality of photonic lanterns, said captured image dependent whether said cores outputs light. 19. A method implemented by a system of one or more processors, the method comprising: obtaining information reflecting output from one or more photonic lanterns, wherein a photonic lantern comprises a tapered fiber bundle comprising a plurality of cores, said tapered fiber bundle having a smaller input end having an input face for injection of light and a larger output end having an output face through which said light is emitted from at least one of said cores depending on the spectral distribution of said light, said cores being closer together at said input end than said output end such that said cores undergo cross-talk that, depending on said spectral distribution, will determine which of said at least one of said cores will emit light;UCRUZ.001WO / 2024-786-2 PATENT computing a forward pass of the obtained information through a machine learning model; and obtaining the spectral distribution based on the machine learning model. 20. The method of Example 19, wherein the obtained information is one or more images depicting spots associated with output from the at least one said cores. 21. The method of Example 19, wherein the machine learning model is a neural network. 22. The method of Example 21, wherein the neural network is a convolutional neural network. 23. The method of Example 21, wherein the neural network is a fully-connected network. 24. The method of Example 19, wherein the machine learning model is trained based on a training data set comprises individual images and associated ground truth, wherein the ground truth reflects a vector of values, and wherein each value is associated with an individual frequency within a frequency range. 25. The method of Example 24, wherein individual values that form a vector of values are individual intensity values associated with an individual frequency within the frequency range. 26. The method of Example 19, wherein the spectral distribution is a vector of values, wherein each value is associated with an individual frequency within a frequency range. 27. The system of any of Examples 1-18, wherein said at least two of said cores have different sizes. 28. The system of any of Examples 1-18, wherein said at least two of said cores have different indices of refraction. 29. The system of any of Examples 1-18, wherein said at least two of said cores comprise different glass. 30. The system of any of Examples 9-11, wherein said an optical fiber configured to couple light into said input face is misaligned with respect to an effective core formed by said closely spaced plurality of cores at said input face. 31. The system of any of Examples 9-11, wherein said an optical fiber configured to couple light into said input face at an average non-zero angle with respect to said input end.UCRUZ.001WO / 2024-786-2 PATENT 32. The system of any of Examples 1-18, further comprising a cooler to cool said detector array. 33. The system of any of Examples 1-18, wherein said plurality of cores at said input face operate as a single mode waveguide. Additional Information

[0124] A wide variety of other variations, for example, to any of the embodiments or implementation disclosed herein, are possible. Components can be added, removed, and / or rearranged. Similarly, in any method or process disclosed herein, steps or operations can be added, removed, and / or rearranged. Other configurations are possible.

[0125] Configurations other than those described herein are possible. The structures, devices, systems, and methods may include additional components, features, and steps and any of these components, features, and steps may be excluded and may or may not be replaced with others. The arrangements may be different. Reference throughout this specification to “some embodiments,” “certain embodiments,” or “an embodiment” means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least some embodiments. Thus, appearances of the phrases “in some embodiments” or “in an embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment and may refer to one or more of the same or different embodiments. Furthermore, the particular features, structures or characteristics may be combined in any suitable manner, as would be apparent to one of ordinary skill in the art from this disclosure, in one or more embodiments.

[0126] As used in this application, the terms “comprising,” “including,” “having,” and the like are synonymous and are used inclusively, in an open-ended fashion, and do not exclude additional elements, features, acts, operations, and so forth. Also, the term “or” is used in its inclusive sense (and not in its exclusive sense) so that when used, for example, to connect a list of elements, the term “or” means one, some, or all of the elements in the list.

[0127] Similarly, it should be appreciated that in the above description of embodiments, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. This method of disclosure,UCRUZ.001WO / 2024-786-2 PATENT however, is not to be interpreted as reflecting an intention that any claim require more features than are expressly recited in that claim. Rather, inventive aspects lie in a combination of fewer than all features of any single foregoing disclosed embodiment.

[0128] Some examples of some of certain embodiments disclosed and / or described herein are listed below. These examples of various embodiments are not meant to be limiting, but rather illustrate some of the embodiments of systems and methods that that are disclosed by this description and accompanying figures. Such embodiments can include:

[0129] Although the inventions presented herein have been disclosed in the context of certain preferred embodiments and examples, it will be understood by those skilled in the art that the inventions extend beyond the specifically disclosed embodiments to other alternative embodiments and / or uses of the inventions and obvious modifications and equivalents thereof. Thus, it is intended that the scope of the inventions herein disclosed should not be limited by the particular embodiments described above.

Claims

UCRUZ.001WO / 2024-786-2 PATENT WHAT IS CLAIMED IS:

1. A system for determining a spectral distribution of light, said system comprising: a photonic lantern comprising a tapered fiber bundle comprising a plurality of cores, said tapered fiber bundle having a smaller input end having an input face for injection of said light and a larger output end having an output face through which said light is emitted from at least one of said cores depending on the spectral distribution of said light, said cores being closer together at said input end than said output end such that said cores undergo cross-talk that, depending on said spectral distribution, will determine which of said at least one of said cores will emit light; and a detector array configured to capture an image using said light emitted from said output face, said captured image dependent on the light emitted from said at least one of said cores that outputs light.

2. The system of Claim 1, further comprising electronics configured to determine said spectral distribution from an image captured by said detector array using said light emitted from said output face.

3. The system of Claim 1, further comprising a lens configured to image said output face onto said detector array.

4. The system of Claim 1, further comprising a wavelength dispersive optical element positioned in an optical path between said output face and said detector array.

5. The system of Claim 4, further comprising a lens configured to form an image onto said detector array using said light from said wavelength dispersive element.

6. The system of Claim 4, further comprising a collimating lens configured to increase collimation of light from said output face directed to said wavelength dispersive optical.

7. The system of Claim 4, wherein said wavelength dispersive element comprises a prism or grating.

8. The system of Claim 1, further comprising an optical fiber configured to couple light into said input face.

9. The system of Claim 8, wherein said optical fiber comprises a single mode fiber.

10. The system of Claim 8, wherein said optical fiber comprises a fiber Bragg grating.UCRUZ.001WO / 2024-786-2 PATENT depending on said spectral distribution, will determine whether said cores will emit light from said output face wherein different cores of said photonic lantern in said first stage at said output face are optically coupled to different of said plurality of photonic lanterns of said second stage.

17. The system of Claim 16, further comprising at least one detector array configured to capture an image using said light emitted from said output face of at least one of said plurality of photonic lanterns, said captured image dependent whether said cores outputs light.

18. A method implemented by a system of one or more processors, the method comprising: obtaining information reflecting output from one or more photonic lanterns, wherein a photonic lantern comprises a tapered fiber bundle comprising a plurality of cores, said tapered fiber bundle having a smaller input end having an input face for injection of light and a larger output end having an output face through which said light is emitted from at least one of said cores depending on the spectral distribution of said light, said cores being closer together at said input end than said output end such that said cores undergo cross-talk that, depending on said spectral distribution, will determine which of said at least one of said cores will emit light; computing a forward pass of the obtained information through a machine learning model; and obtaining the spectral distribution based on the machine learning model.

19. The method of Claim 18, wherein the obtained information is one or more images depicting spots associated with output from the at least one said cores. -54-depending on said spectral distribution, will determine whether said cores will emit light from said output face wherein different cores of said photonic lantern in said first stage at said output face are optically coupled to different of said plurality of photonic lanterns of said second stage.

17. The system of Claim 16, further comprising at least one detector array configured to capture an image using said light emitted from said output face of at least one of said plurality of photonic lanterns, said captured image dependent whether said cores outputs light.

18. A method implemented by a system of one or more processors, the method comprising: obtaining information reflecting output from one or more photonic lanterns, wherein a photonic lantern comprises a tapered fiber bundle comprising a plurality of cores, said tapered fiber bundle having a smaller input end having an input face for injection of light and a larger output end having an output face through which said light is emitted from at least one of said cores depending on the spectral distribution of said light, said cores being closer together at said input end than said output end such that said cores undergo cross-talk that, depending on said spectral distribution, will determine which of said at least one of said cores will emit light; computing a forward pass of the obtained information through a machine learning model; and obtaining the spectral distribution based on the machine learning model.

19. The method of Claim 18, wherein the obtained information is one or more images depicting spots associated with output from the at least one said cores.

Images