Abstract: A driver can be configured to provide sensed phase currents as feedback to a controller to indicate the output currents from each phase of a switch mode power supply (SMPS). The driver can be configured to temperature compensate the sensed currents in one of two ways. If a temperature sensor is directly coupled to the driver, then the driver may be configured to temperature compensate the sensed currents from each phase based on a temperature measurement made by the temperature sensor. If a temperature sensor is not directly coupled to the driver, then the driver may be configured to temperature compensate the sensed current from each phase based on a temperature signal received from a bus coupled to the driver. The bus can communicate the temperature signal so that multiple drivers can utilize one temperature sensor.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting actions to be performed by an agent interacting with an environment. In one aspect, a system includes a language encoder model that is configured to receive a text string in a particular natural language, and process the text string to generate a text embedding of the text string. The system includes an observation encoder neural network that is configured to receive an observation characterizing a state of the environment, and process the observation to generate an observation embedding of the observation. The system includes a subsystem that is configured to obtain a current text embedding of a current text string and a current observation embedding of a current observation. The subsystem is configured to select an action to be performed by the agent in response to the current observation.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting actions to be performed by an agent interacting with an environment. In one aspect, a system includes a language encoder model that is configured to receive a text string in a particular natural language, and process the text string to generate a text embedding of the text string. The system includes an observation encoder neural network that is configured to receive an observation characterizing a state of the environment, and process the observation to generate an observation embedding of the observation. The system includes a subsystem that is configured to obtain a current text embedding of a current text string and a current observation embedding of a current observation. The subsystem is configured to select an action to be performed by the agent in response to the current observation.

Abstract: A diffractive waveguide stack includes first, second, and third diffractive waveguides for guiding light in first, second, and third visible wavelength ranges, respectively. The first diffractive waveguide includes a first material having first refractive index at a selected wavelength and a first target refractive index at a midpoint of the first visible wavelength range. The second diffractive waveguide includes a second material having a second refractive index at the selected wavelength and a second target refractive index at a midpoint of the second visible wavelength range. The third diffractive waveguide includes a third material having a third refractive index at the selected wavelength and a third target refractive index at a midpoint of the third visible wavelength range. A difference between any two of the first target refractive index, the second target refractive index, and the third target refractive index is less than 0.005 at the selected wavelength.

Abstract: A diffractive waveguide stack includes first, second, and third diffractive waveguides for guiding light in first, second, and third visible wavelength ranges, respectively. The first diffractive waveguide includes a first material having first refractive index at a selected wavelength and a first target refractive index at a midpoint of the first visible wavelength range. The second diffractive waveguide includes a second material having a second refractive index at the selected wavelength and a second target refractive index at a midpoint of the second visible wavelength range. The third diffractive waveguide includes a third material having a third refractive index at the selected wavelength and a third target refractive index at a midpoint of the third visible wavelength range. A difference between any two of the first target refractive index, the second target refractive index, and the third target refractive index is less than 0.005 at the selected wavelength.

Abstract: Gratings may be used in waveguides. Deep surface relief gratings (SRGs) may offer many advantages over conventional SRGs, an important one being a higher S-diffraction efficiency. Deep SRGs can be implemented as polymer surface relief gratings or evacuated periodic structures (EPSs). EPSs can be formed by first recording a holographic polymer dispersed liquid crystal (HPDLC) periodic structure. Removing the liquid crystal from the cured periodic structure provides a polymer surface relief grating. Polymer surface relief gratings have many applications including for use in waveguide-based displays.

Abstract: A system includes a first chuck operable to support a stencil including a plurality of apertures, a wafer chuck operable to support and move a wafer including a plurality of incoupling gratings, a first light source operable to direct light to impinge on a first surface of the stencil, and one or more second light sources operable to direct light to impinge on the wafer. The system also includes one or more lens and camera assemblies operable to receive light from the first light source passing through the plurality of apertures in the stencil and receive light from the one or more second light sources diffracted from the plurality of incoupling gratings in the wafer. The system also includes an alignment system operable to move the wafer with respect to the stencil to reduce an offset between aperture locations and incoupling grating locations.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting actions to be performed by an agent interacting with an environment. In one aspect, a system includes a language encoder model that is configured to receive a text string in a particular natural language, and process the text string to generate a text embedding of the text string. The system includes an observation encoder neural network that is configured to receive an observation characterizing a state of the environment, and process the observation to generate an observation embedding of the observation. The system includes a subsystem that is configured to obtain a current text embedding of a current text string and a current observation embedding of a current observation. The subsystem is configured to select an action to be performed by the agent in response to the current observation.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing multi-modal inputs using language models. In particular, the inputs include an image, and the image is encoded by an image encoder neural network to generate a sequence of image embeddings representing the image. The sequence of image embeddings is provided as at least part of an input sequence to that is processed by a language model neural network.

Abstract: A method of aligning a stencil to an eyepiece wafer includes providing the stencil, positioning the stencil with respect to a first light source, and determining locations of at least two stencil apertures. The method also includes providing the eyepiece wafer. The eyepiece wafer includes at least two eyepiece waveguides, each eyepiece waveguide including an incoupling grating and a corresponding diffraction pattern. The method further includes directing light from one or more second light sources to impinge on each of the corresponding diffraction patterns, imaging light diffracted from each incoupling grating, determining at least two incoupling grating locations, determining offsets between corresponding stencil aperture locations and incoupling grating locations, and aligning the stencil to the eyepiece wafer based on the determined offsets.

Abstract: A driver can be configured to provide sensed phase currents as feedback to a controller to indicate the output currents from each phase of a switch mode power supply (SMPS). The driver can be configured to temperature compensate the sensed currents in one of two ways. If a temperature sensor is directly coupled to the driver, then the driver may be configured to temperature compensate the sensed currents from each phase based on a temperature measurement made by the temperature sensor. If a temperature sensor is not directly coupled to the driver, then the driver may be configured to temperature compensate the sensed current from each phase based on a temperature signal received from a bus coupled to the driver. The bus can communicate the temperature signal so that multiple drivers can utilize one temperature sensor.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting actions to be performed by an agent interacting with an environment. In one aspect, a system includes a language encoder model that is configured to receive a text string in a particular natural language, and process the text string to generate a text embedding of the text string. The system includes an observation encoder neural network that is configured to receive an observation characterizing a state of the environment, and process the observation to generate an observation embedding of the observation. The system includes a subsystem that is configured to obtain a current text embedding of a current text string and a current observation embedding of a current observation. The subsystem is configured to select an action to be performed by the agent in response to the current observation.

Abstract: Methods are disclosed for fabricating molds for forming eyepieces having waveguides with integrated spacers. The molds are formed by etching deep holes (e.g., 5 ?m to 1000 ?m deep) into a substrate using a wet etch or dry etch. The etch masks for defining the holes may be formed with a thick metal layer and/or multiple layers of different metals. A resist layer may be disposed over the etch mask. The resist layer may be patterned to form a pattern of holes, the pattern may be transferred to the etch mask, and the etch mask may be used to transfer the pattern into the underlying substrate. The patterned substrate may be utilized as a mold onto which a flowable polymer may be introduced and allowed to harden. Hardened polymer in the holes may form integrated spacers. The hardened polymer may be removed from the mold to form a waveguide with integrated spacers.

Abstract: Improvements to gratings for use in waveguides and methods of producing them are described herein. Deep surface relief gratings (SRGs) may offer many advantages over conventional SRGs and Bragg gratings, an important one being a higher S-diffraction efficiency. In one embodiment, deep SRGs can be implemented as polymer surface relief gratings or evacuated Bragg gratings (EBGs). EBGs can be formed by first recording a holographic polymer dispersed liquid crystal (HPDLC) grating. Removing the liquid crystal from the cured grating provides a polymer surface relief grating. Polymer surface relief gratings have many applications including for use in waveguide-based displays.

Abstract: A driver can be configured to provide sensed phase currents as feedback to a controller to indicate the output currents from each phase of a switch mode power supply (SMPS). The driver can be configured to temperature compensate the sensed currents in one of two ways. If a temperature sensor is directly coupled to the driver, then the driver may be configured to temperature compensate the sensed currents from each phase based on a temperature measurement made by the temperature sensor. If a temperature sensor is not directly coupled to the driver, then the driver may be configured to temperature compensate the sensed current from each phase based on a temperature signal received from a bus coupled to the driver. The bus can communicate the temperature signal so that multiple drivers can utilize one temperature sensor.

Abstract: Improvements to gratings for use in waveguides and methods of producing them are described herein. Deep surface relief gratings (SRGs) may offer many advantages over conventional SRGs and Bragg gratings, an important one being a higher S-diffraction efficiency. In one embodiment, deep SRGs can be implemented as polymer surface relief gratings or evacuated Bragg gratings (EBGs). EBGs can be formed by first recording a holographic polymer dispersed liquid crystal (HPDLC) grating. Removing the liquid crystal from the cured grating provides a polymer surface relief grating. Polymer surface relief gratings have many applications including for use in waveguide-based displays.

Abstract: In some embodiments, a near-eye, near-eye display system comprises a stack of waveguides having pillars in a central, active portion of the waveguides. The active portion may include light outcoupling optical elements configured to outcouple image light from the waveguides towards the eye of a viewer. The pillars extend between and separate neighboring ones of the waveguides. The light outcoupling optical elements may include diffractive optical elements that are formed simultaneously with the pillars, for example, by imprinting or casting. The pillars are disposed on one or more major surfaces of each of the waveguides. The pillars may define a distance between two adjacent waveguides of the stack of waveguides. The pillars may be bonded to adjacent waveguides may be using one or more of the systems, methods, or devices herein. The bonding provides a high level of thermal stability to the waveguide stack, to resist deformation as temperatures change.

Abstract: A video processing system configured to analyze a sequence of video frames to detect objects in the video frames and provide information relating to the detected objects in response to a query. The query may comprise, for example, a request for a prediction of a future event, or of the location of an object, or a request for a prediction of what would happen if an object were modified. The system uses a transformer neural network subsystem to process representations of objects in the video.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a speaker neural network using one or more listener neural networks.

Abstract: Methods are disclosed for fabricating molds for forming eyepieces having waveguides with integrated spacers. The molds are formed by etching deep holes (e.g., 5 ?m to 1000 ?m deep) into a substrate using a wet etch or dry etch. The etch masks for defining the holes may be formed with a thick metal layer and/or multiple layers of different metals. A resist layer may be disposed over the etch mask. The resist layer may be patterned to form a pattern of holes, the pattern may be transferred to the etch mask, and the etch mask may be used to transfer the pattern into the underlying substrate. The patterned substrate may be utilized as a mold onto which a flowable polymer may be introduced and allowed to harden. Hardened polymer in the holes may form integrated spacers. The hardened polymer may be removed from the mold to form a waveguide with integrated spacers.