Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a machine learning model on training data. In one aspect, one of the methods include: obtaining a training data set comprising a plurality of training inputs; obtaining data defining an original search space of a plurality of candidate data augmentation policies; generating, from the original search space, a compact search space that has one or more global hyperparameters; and training the machine learning model on the training data using one or more final data augmentation policies generated from the compact search space.

Abstract: A backplane includes: a substrate including a circuit structure layer, a first reflective layer disposed on a bearing surface of the substrate, a plurality of light-emitting diode chips, and a plurality of optical structures. The first reflective layer includes a plurality of through holes spaced apart. A light-emitting diode chip in the plurality of light-emitting diode chips is located in one of the plurality of through holes. The plurality of light-emitting diode chips are electrically connected to the circuit structure layer. The circuit structure layer is configured to drive the plurality of light-emitting diode chips to emit light. An optical structure in the plurality of optical structures covers the light-emitting diode chip, a light incident surface of the optical structure is in contact with a light exit surface of the light-emitting diode chip, and a light exit surface of the optical structure is a curved surface.

Abstract: The embodiments of the disclosure provide a method for interacting with a virtual world, a host, and a computer readable storage medium. The method includes: determining a hand gesture and accordingly determining an indicator in the virtual world; in response to determining that the hand gesture is a pinch gesture, performing a specific function corresponding to a position indicated by the indicator in the virtual world or a specific object indicated by the indicator in the virtual world.

Abstract: The disclosure belongs to the technical field of liquid crystal materials, and specifically relates to a polymerizable compound, a preparation method therefor, and the use thereof. The polymerizable compound has a structure as shown in general formula I. Compared with the existing polymerizable RM, the polymerizable compound has the advantages of a good solubility, a faster polymerization rate, a more complete extent of polymerization, and less residues, thereby improving the problem of poor display to a great extent.

Abstract: Methods, computer systems, and apparatus, including computer programs encoded on computer storage media, for processing a perspective view range image generated from sensor measurements of an environment. The perspective view range image includes a plurality of pixels arranged in a two-dimensional grid and including, for each pixel, (i) features of one or more sensor measurements at a location in the environment corresponding to the pixel and (ii) geometry information comprising range features characterizing a range of the location in the environment corresponding to the pixel relative to the one or more sensors. The system processes the perspective view range image using a first neural network to generate an output feature representation. The first neural network comprises a first perspective point-set aggregation layer comprising a geometry-dependent kernel.

Abstract: The present disclosure provides a fixed-position defect doping method for a micro-nanostructure based on a self-alignment process, including: S1, sequentially forming a sacrificial layer and a photoresist layer on a surface of a crystal substrate; S2, performing a lithography on the photoresist layer to form a mask hole according to a micro-nano pattern; S3, performing an isotropic etching on the sacrificial layer through the mask hole, and amplifying the micro-nano pattern to the sacrificial layer; S4, performing an ion implantation doping on an exposed crystal surface below the mask hole; S5, removing the photoresist layer, and depositing a mask material; S6, removing the sacrificial layer, and transferring a micro-nano amplified pattern in the sacrificial layer to a mask material pattern; and S7, etching an exposed crystal surface, and removing the mask material on the surface and forming a specific defect by annealing.

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 embodiments of the disclosure provide a method for interacting with a virtual world, a host, and a computer readable storage medium. The method includes: determining a hand gesture and accordingly determining an indicator in the virtual world; in response to determining that the hand gesture is a pinch gesture, performing a specific function corresponding to a position indicated by the indicator in the virtual world or a specific object indicated by the indicator in the virtual world.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing point cloud data using dynamic voxelization. When deployed within an on-board system of a vehicle, processing the point cloud data using dynamic voxelization can be used to make autonomous driving decisions for the vehicle with enhanced accuracy, for example by combining representations of point cloud data characterizing a scene from multiple views of the scene.

Abstract: A method (100) for producing a particulate titanium alloy product can include preparing (110) a composite particulate oxide mixture with TiO2 powder and at least one alloying element powder. The composite particulate oxide mixture can be co-reduced (120) using a metallic reducing agent under a hydrogen atmosphere at a reduction temperature for a reduction time sufficient to produce a hydrogenated titanium alloy product. The hydrogenated titanium alloy product can then be heat treated (130) under a hydrogen atmosphere and a heat treating temperature to reduce pore size and specific surface area to form a heat treated hydrogenated titanium product. The heat treated hydrogenated titanium product can be deoxygenated (140) to reduce residual oxygen to less than 0.2 wt % to form a deoxygenated hydrogenated titanium product as a particulate. The deoxygenated hydrogenated titanium product can optionally be dehydrogenated (150) to form the titanium alloy product as a particulate.

Abstract: A flow channel assembly (200) of a fluid micro-injection device has a fluid seat (210), a nozzle mounting plate (240), a nozzle (220), a nozzle platen (250) and a fluid supply joint (230). The nozzle (220) is connected to the fluid seat (210) by the nozzle mounting plate (240). The nozzle platen (250) is connected with the nozzle mounting plate (240) to secure the nozzle (220). The fluid supply joint (230) and the fluid seat (210) are connected to control the fluid flowing to the nozzle (220).

Abstract: A fluid micro-injection device comprises an execution system (100) and a flow channel assembly (200) connected with the execution system (100). The execution system (100) includes a base body (110), a movable member (120), an executor, an adjusting member (130) and a plurality of clearance sheets. The flow channel assembly (200) have a fluid seat (210), a nozzle (220) and a fluid supply joint (230).

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing point cloud data representing a sensor measurement of a scene captured by one or more sensors to generate an object detection output that identifies locations of one or more objects in the scene. When deployed within an on-board system of a vehicle, the object detection output that is generated can be used to make autonomous driving decisions for the vehicle with enhanced accuracy.

Abstract: An execution system (100) for a fluid micro-injection device and a fluid micro-injection device are provided. The execution system (100) has a base body (110), a movable member (120), an executor and an adjusting member (130). The adjusting member (130) is disposed in the executor mounting cavity (110) and connected with the executor. The adjusting member (130) is adjustable for adjusting a pre-tightening force of the executor.

Abstract: An execution system (100) for a fluid micro-injection device have a base body (110), a movable member (120), an executor, an adjusting member (130) and a plurality of clearance sheets. The movable member (120) is movably disposed on the base body (110). The executor and the adjusting member (130) are disposed in the base body (110). The plurality of clearance sheets are disposed between the base body (110) and the adjusting member (130) to adjust a pre-tightening force of the executor.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for agent trajectory prediction using temporal-spatial interaction predictions.

Abstract: Methods, computer systems, and apparatus, including computer programs encoded on computer storage media, for performing object detection. The system obtains a respective range image corresponding to each point cloud in a set of point clouds captured by one or more sensors. The system processes each range image using a segmentation neural network to generate range image features and a segmentation output. The system generates a feature representation of the set of point clouds from only the feature representations of the foreground points. The system processes the feature representation of the set of point clouds using a prediction neural network to generate a prediction characterizing the set of point clouds.

Abstract: A fluid micro-injection device including an execution system, a fluid channel assembly and a movable unit. The execution system includes: a base body defining an executor mounting chamber and an adjustor mounting chamber; an executor; an adjusting seat, at least a portion of the adjusting seat extending through the positioning hole and into the adjustor mounting chamber, and the adjusting seat being provided with a channel; and an adjustor, at least a portion of the adjustor being connected with the adjusting seat to adjust the distance between the adjusting seat and the executor. The fluid channel assembly includes: a fluid seat detachably connected with the base body. At least a portion of the movable unit is provided on the adjusting seat.

Abstract: Methods, computer systems, and apparatus, including computer programs encoded on computer storage media, for performing three-dimensional auto-labeling on sensor data. The system obtains a sensor data segment that includes a temporal sequence of three-dimensional point clouds generated from sensor readings of an environment by one or more sensors. The system identifies, from the sensor data segment, (i) a plurality of object tracks that each corresponds to a different object in the environment and (ii) for each object track, respective initial three-dimensional regions in each of one or more of the point clouds in which the corresponding object appears. The system generates, for each object track, extracted object track data that includes at least the points in the respective initial three-dimensional regions for the object track.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing point cloud data using dynamic voxelization. When deployed within an on-board system of a vehicle, processing the point cloud data using dynamic voxelization can be used to make autonomous driving decisions for the vehicle with enhanced accuracy, for example by combining representations of point cloud data characterizing a scene from multiple views of the scene.