Abstract: Some implementations related to using a large language model (LLM) in generating (and potentially refining) a plan for the execution of a long-horizon robotic task. Various implementations include processing, using the LLM, a free-form natural language instruction and textual feedback to generate LLM output. In many implementations, the free-form natural language instruction describes the robotic task. In additional or alternative implementations, the textual feedback can include task-specific feedback, passive scene description feedback, active scene description feedback, one or more additional or alternative types of environmental feedback, and/or combinations thereof. In some implementations, the system can select one or more robotic skills to perform based on the LLM output.

Abstract: A method for applying a surface treatment to a plugged honeycomb body comprising porous wall includes: mixing particles of an inorganic material with a liquid vehicle and a binder material to form a liquid-particulate-binder stream; mixing the liquid-particulate-binder stream with an atomizing gas, directing the liquid-particulate-binder stream into an atomizing nozzle thereby atomizing the particles into liquid-particulate-binder droplets comprised of the liquid vehicle, he binder material, and the particles; conveying the droplets toward the plugged honeycomb body by a gaseous carrier stream, wherein the gaseous carrier stream comprises a carrier gas and the atomizing gas; evaporating substantially all of the liquid vehicle from the droplets to form agglomerates comprised of the particles and the binder material; depositing the agglomerates onto the porous walls of the plugged honeycomb body; wherein the deposited agglomerates are disposed on, or in, or both on and in, the porous walls.

Abstract: A method for applying a surface treatment to a plugged honeycomb body comprising porous wall includes: mixing particles of an inorganic material with a liquid vehicle and a binder material to form a liquid-particulate-binder stream; mixing the liquid-particulate-binder stream with an atomizing gas, directing the liquid-particulate-binder stream into an atomizing nozzle thereby atomizing the particles into liquid-particulate-binder droplets comprised of the liquid vehicle, the binder material, and the particles; conveying the droplets toward the plugged honeycomb body by a gaseous carrier stream, wherein the gaseous carrier stream comprises a carrier gas and the atomizing gas; evaporating substantially all of the liquid vehicle from the droplets to form agglomerates comprised of the particles and the binder material; depositing the agglomerates onto the porous walls of the plugged honeycomb body; wherein the deposited agglomerates are disposed on, or in, or both on and in, the porous walls.

Abstract: A device for advanced nitrogen and phosphorus removal in sewage treatment includes a pre-denitrification zone, an anaerobic zone, an anoxic zone, an aerobic zone, a sedimentation zone, a biological filtration zone, and a clear water zone, where a sludge return system is provided between the pre-denitrification zone and the sedimentation zone; a nitrification liquid return system is provided between the anoxic zone and the aerobic zone; a filler layer is provided in the biological filtration zone, and the filler layer divides a cavity in the biological filtration zone to form an upper water inlet cavity and a lower water outlet cavity; a backwash aeration pipe is provided in the water outlet cavity, and a backwash water outlet is formed in the water inlet cavity; and the backwash water outlet is connected to a sludge concentration and storage tank or the pre-denitrification zone.

Abstract: A method for upgrading client software includes receiving an upgrade instruction for client software deployed on a computing node, where the client software is used to access one or more storage nodes in the storage system based on a received input/output I/O instruction; performing I/O suspension on the received I/O instruction in response to the upgrade instruction; and upgrading the client software based on the upgrade instruction.

Abstract: A method of producing soot, including: combusting a first fuel stream and a first oxidizer at a burner face; combusting a second fuel stream and a second oxidizer at the burner face, wherein the second fuel stream and the second oxidizer are premixed in advance of the burner face and a second equivalence ratio of the second fuel stream and the second oxidizer is less than about 1; and combusting a silicon-containing fuel into a plurality of soot particles, wherein the second fuel stream and the second oxidizer are combusted between the first fuel stream and the silicon-containing fuel. Applying this method of producing soot to deposit a preform suitable for the manufacture of optical fibers.

Abstract: This application provides an image rendering method and apparatus, an electronic device, a computer-readable storage medium, and a computer program product. The method comprises the following steps: acquiring texture data of a virtual object; performing differential fitting processing on the texture data to obtain fitted texture data; baking the fitted texture data into a texture image to obtain a rendered texture image; and performing, based on the rendered texture image, image rendering processing corresponding to the virtual.

Abstract: A bone model combination method includes obtaining a plurality of bone models, each of the bone models comprising a pixel set, and each bone model corresponding to a common skeleton. The method further includes determining, for two of the bone models, a set of common bones that is an intersection of bones included in both of the two bone models. The method further includes determining, for common bones in the intersection, common bone chains, generating common bone chain duplicates on the basis of the common bone chains, and binding the common bone chain duplicates at upper common bones of the common bone chains. The method further includes building a mapping between the common bone chains and a pixel set included in a corresponding bone model, and building a mapping between the common bone chain duplicates and a pixel set included in a corresponding bone model.

Abstract: This application discloses a soft rasterizing method and apparatus, a device, a medium, and a program product, belonging to the field of computer technologies.

Abstract: This application provides an image rendering method performed by an electronic device. The method includes: acquiring first texture data of a virtual object and conversion parameters corresponding to second texture data of the virtual object; performing, based on the conversion parameters and the first texture data, fitting rendering to obtain a fitted rendering image comprising the virtual object; updating the conversion parameters and the first texture data based on a rendering loss between the fitted rendering image and a reference rendering image comprising the virtual object , wherein the reference rendering image is a rendering image that is obtained by performing rendering based on the second texture data; and performing, based on the updated conversion parameters and the updated first texture data, real-time rendering to obtain a target rendering image comprising the virtual object.

Abstract: A three-dimensional model rendering method performed by a computer device is disclosed. The method includes: acquiring material information of a plurality of triangle primitives in a three-dimensional model, at least two triangle primitives in the plurality of triangle primitives having different material information; generating model parameters of the three-dimensional model based on the material information of the triangle primitives; and rendering a two-dimensional image of the three-dimensional model based on the rendering parameters of the rasters.

Abstract: The described aspects and implementations enable efficient calibration of a sensing system of a vehicle. In one implementation, disclosed is a method and a system to perform the method of obtaining a plurality of target outputs generated by processing a training input using a respective teacher machine learning model (MLM) of a plurality of teacher MLMs. The training input includes a representation of one or more objects, and each of the plurality of target outputs includes a classification of the objects among a respective set of classes of a plurality of sets of classes. The method further includes using the training input and the plurality of target outputs to train a student MLM to classify the one or more objects among each of the plurality of sets of classes.

Abstract: Methods, systems, and apparatus for processing point clouds using neural networks to perform a machine learning task. In one aspect, a system comprises one or more computers configured to obtain a set of point clouds captured by one or more sensors. Each point cloud includes a respective plurality of three-dimensional points. The one or more computers assign the three-dimensional points to respective voxels in a voxel grid, where the grid of voxels includes non-empty voxels to which one or more points are assigned and empty voxels to which no points are assigned. For each non-empty voxel, the one or more computers generate initial features based on the points that are assigned to the non-empty voxel. The one or more computers generate multi-scale features of the voxel grid, and the one or more computers generate an output for a point cloud processing task using the multi-scale features of the voxel grid.

Abstract: The present disclosure provides a method for processing visibility data of a three-dimensional model performed by a computer device. The method includes: obtaining original values of visibility data of a plurality of sampling points of the three-dimensional model; determining visibility data of a plurality of vertexes of the three-dimensional model based on a first error function measuring differences between recovery values of the visibility data of the sampling points from interpolation of the visibility data of the vertexes of the three-dimensional model and the original values of the visibility data of the sampling points and change rates of the recovery values of the visibility data of the sampling points; and storing the visibility data of each vertex of the three-dimensional model. This method reduces the space for storing the visibility data, relieves the pressure of storing the visibility data, and improves the efficiency of rendering the three-dimensional model.

Abstract: Implementations process, using a large language model, a free-form natural language (NL) instruction to generate to generate LLM output. Those implementations generate, based on the LLM output and a NL skill description of a robotic skill, a task-grounding measure that reflects a probability of the skill description in the probability distribution of the LLM output. Those implementations further generate, based on the robotic skill and current environmental state data, a world-grounding measure that reflects a probability of the robotic skill being successful based on the current environmental state data. Those implementations further determine, based on both the task-grounding measure and the world-grounding measure, whether to implement the robotic skill.

Abstract: Disclosed are a power grid-friendly control method and system for a data center cooling system. The method includes: dividing an equipment room into a plurality of pieces of subspace in advance, with a plurality of servers and a cooling device comprised in each piece of subspace; obtaining total power consumption of the servers in each piece of subspace; performing priority sorting on the total power consumption of the servers in each piece of subspace; obtaining a real-time power supply energy consumption value of a power grid; determining, whether an energy supply is sufficient; and if the energy supply is sufficient, skipping turning off the cooling device in the subspace; or if the energy supply is not sufficient, turning off a cooling device, till it is determined that the energy supply is sufficient, thus minimizing impact of turning off the cooling system on stability.

Abstract: A method of producing bi-modal particles includes the steps of igniting a first precursor gas using a primary burner thereby producing a first plurality of particles of a first size, fluidly transporting the first plurality of particles down a particle tube, igniting a second precursor gas using a secondary burner thereby producing a second plurality of particles of a second size, flowing the second plurality of particles into the first plurality of particles, and capturing the first and second plurality of particles.

Abstract: A method of forming an optical fiber preform includes the steps: igniting a burner having a fume tube assembly to produce a first spray size of silicon dioxide particles; depositing the silicon dioxide particles on a core cane to produce a soot blank; and adjusting an effective diameter of an aperture of the fume tube assembly to produce a second spray size of the silicon dioxide particles. The second spray size is larger than the first spray size.

Abstract: Disclosed are a power grid-friendly control method and system for a data center cooling system. The method includes: dividing an equipment room into a plurality of pieces of subspace in advance, with a plurality of servers and a cooling device comprised in each piece of subspace; obtaining total power consumption of the servers in each piece of subspace; performing priority sorting on the total power consumption of the servers in each piece of subspace; obtaining a real-time power supply energy consumption value of a power grid; determining, whether an energy supply is sufficient; and if the energy supply is sufficient, skipping turning off the cooling device in the subspace; or if the energy supply is not sufficient, turning off a cooling device, till it is determined that the energy supply is sufficient, thus minimizing impact of turning off the cooling system on stability.

Abstract: A method of producing soot, including: combusting a first fuel stream and a first oxidizer at a burner face; combusting a second fuel stream and a second oxidizer at the burner face, wherein the second fuel stream and the second oxidizer are premixed in advance of the burner face and a second equivalence ratio of the second fuel stream and the second oxidizer is less than about 1; and combusting a silicon-containing fuel into a plurality of soot particles, wherein the second fuel stream and the second oxidizer are combusted between the first fuel stream and the silicon-containing fuel. Applying this method of producing soot to deposit a preform suitable for the manufacture of optical fibers.