Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating trajectory predictions for one or more target agents, e.g., a vehicle, a cyclist, or a pedestrian, in an environment. In one aspect, one of the methods include: obtaining scene context data characterizing a scene at a current time point in an environment that includes multiple target agents; generating, from the scene context data, an encoded representation of the scene in the environment; and generating, by a diffusion model based on the encoded representation, a respective trajectory prediction output that predicts a respective future trajectory for each of the multiple target agents after the current time point.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for estimating a 3-D pose of an object of interest from image and point cloud data. In one aspect, a method includes obtaining an image of an environment; obtaining a point cloud of a three-dimensional region of the environment; generating a fused representation of the image and the point cloud; and processing the fused representation using a pose estimation neural network and in accordance with current values of a plurality of pose estimation network parameters to generate a pose estimation network output that specifies, for each of multiple keypoints, a respective estimated position in the three-dimensional region of the environment.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training a semantic segmentation neural network for point clouds. One of the methods includes: obtaining a plurality of training points divided into a respective plurality of components; obtaining, for each of the respective plurality of components, data identifying a ground truth category for one or more labeled point; processing each training points using a semantic segmentation neural network to generate a semantic segmentation that includes a respective score for each of the plurality of categories; determining a gradient of a loss function that penalizes the semantic segmentation neural network for generating, for points in the component, non-zero scores for categories that are not the ground truth category for any labeled point in the component; and updating, using the gradient, the parameters of the semantic segmentation neural network.

Abstract: The present disclosure relates to an object recognition system and method. The system comprises: an illumination apparatus configured to generate an illumination light for illuminating an object; a lighting sensor configured to acquire a first lighting parameter of an ambient light in a recognition environment in which the object is located; an imaging apparatus configured to acquire imaging data of the object; a first controller communicatively connected with the illumination apparatus and the lighting sensor, and configured to control a second lighting parameter of the illumination light generated by the illumination apparatus according to the first lighting parameter, until the first lighting parameter of the ambient light is within a preset parameter range; and a processor communicatively connected with the imaging apparatus and the first controller, and configured to recognize the object according to imaging data of the object under the ambient light within the preset parameter range.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating realistic full-scene point clouds. One of the methods includes obtaining an initial scene point cloud characterizing an initial scene in an environment; obtaining, for each of one or more objects, an object point cloud that characterizes the object; and processing a first input comprising the initial scene point cloud and the one or more object point clouds using a first neural network that is configured to process the first input to generate a final scene point cloud that characterizes a transformed scene that has the one or more objects added to the initial scene.

Abstract: Embodiments of the present disclosure relate to automated validation of medical data. Some embodiments of the present disclosure provide a method for medical data validation. The method comprises obtaining target medical data generated in a medical test and obtaining a machine learning model for validating medical data. The machine learning model represents an association between the medical data and validation results, the validation results indicating information about predetermined actions to be performed on the medical data. The method further comprises determining a target validation result for the target medical data by applying the target medical data to the machine learning model, the target validation result indicating information about a target action selected from the predetermined actions to be performed on the target medical data. Through the solution, it is possible to achieve automated medical data validation with high accuracy and efficiency as well as reduced manual efforts.

Abstract: An aluminum borate whisker reinforced and toughened non-metallic matrix composite is provided, which specifically includes a non-metallic material reinforced and toughened with aluminum borate whiskers. The composite exhibits a higher bending strength and fracture toughness and a higher wear resistance. A method for preparing the composite is also provided. The method includes mixing the aluminum borate whiskers and the non-metallic material to form a mixture; and sintering the mixture by a vacuum hot press method, or molding the mixture.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing sensor data, e.g., laser sensor data, using neural networks. One of the methods includes obtaining a temporal sequence of multiple three-dimensional point clouds generated from sensor readings of an environment collected by one or more sensors within a given time period, each three-dimensional point cloud comprising a respective plurality of points in a first coordinate system; processing, using a feature extraction neural network, an input that comprises data derived from the temporal sequence of multiple three-dimensional point clouds to generate a feature embedding; receiving a query that specifies one time point within the given time period; and generating, from the feature embedding and conditioned on the query, one or more outputs that characterize one or more objects in the environment at the time point specified in the received query.

Abstract: The present invention discloses a flame simulating device, comprising a mist generating chamber, an atomizing head, an air orifice and a nozzle. The inside of the mist generating chamber is provided with a liquid and the atomizing head, the atomizing head being capable of atomizing the liquid inside the mist generating chamber, the two sides of the nozzle being set as Coanda curved surfaces, the cross section of the air orifice being in a constricted shape and providing an air flow blown upward such that under the Venturi effect, the air flow blown upward will guide and attract the mist from inside the mist generating chamber to vent out and flow into a nozzle inlet; the upper surface of the mist generating chamber is provided with a breathing port, and the breathing port directly faces the atomizing head. Due to the Coanda curved surface on the side of the nozzle, the mist flows along both sides of the nozzle under the Coanda effect and then vents out of the nozzle.

Abstract: The present disclosure provides a dropout voltage generation circuit, a switching power supply and a dropout voltage generation method. The circuit includes a reference current generation unit, a first regulation unit and a second regulation unit. The reference current generation unit is configured to generate a plurality of reference currents according to the output current of the switching power supply. The first regulation unit is configured to generate a regulation current according to a first regulation code and the plurality of reference currents. The second regulation unit is configured to generate the dropout voltage according to a second regulation code and the regulation current, wherein the first regulation code and the second regulation code can be configured at least according to a target value of load line resistance. The present disclosure achieves higher accuracy and wider range of regulation of load line resistance with adjustable gain.

Abstract: Provided is a group of peripheral blood gene markers for screening benign and malignant pulmonary micro-nodules, comprising: gene sequences characterized by micronodular lung carcinoma as shown in SEQ ID NOs. 1-6, wherein the gene sequences exhibit differential expression in the peripheral blood of micronodular lung carcinoma patients and non-micronodular lung carcinoma patients. In addition, also provided is the use of the above-mentioned gene markers in preparing a product for early screening micronodular lung carcinoma. The gene markers of the invention used for early screening micronodular lung carcinoma, have high sensitivity and strong specificity; besides, they take peripheral blood, which is the easiest to collect in clinic, as the test sample.

Abstract: A control method of a switching circuit, a control circuit of the switching circuit and the switching circuit are provided. The control circuit includes a slope buffer and a first operational amplifier. The slope buffer receives a first voltage reference, and controls slopes of a rising edge and a falling edge to generate a second voltage reference. The first operational amplifier receives an output feedback voltage and a reference voltage, and performs an operational amplification to obtain a compensation voltage. When the first voltage reference has the falling edge, the reference voltage is coupled to the first voltage reference through a first switch, and the second voltage reference is coupled to an output voltage through a second switch. When the first voltage reference has the rising edge, the reference voltage is coupled to the second voltage reference through a third switch.

Abstract: A streamer type flame simulation device includes a casing, some cross-flow fans disposed inside the casing, a first streamer, a second streamer, a supplementary air-supplying device and a light source. The supplementary air-supplying device is arranged in the split joint area between the two cross-flow fans. Since a supplementary air-supplying device is provided between two cross-flow fans for supplying air, the deficiency where air was unable to deliver at the space occupied by the drive motor of the cross-flow fans can be avoided. Various parts of the first streamer and the second streamer can be blown, thus preventing the flame from being cut off and enhancing the overall coordination, which greatly improves the viewing effect of the flame area.

Abstract: The present invention discloses a regulating mechanism regulating the flows from a plurality of gas outlets in a fuel gas valve, comprising an actuator, a push rod, a presser plate assembly, regulating rods and a valve body. Wherein the valve body has a plurality of gas outlets, a regulating rod is provided in each regulating port which exists between the gas inlet and each gas outlet, a resetting device is provided between each regulating rod and the valve body, and the actuator drives the push rod which is around by the regulating ports to move along the axis. The push rod causes the presser plate assembly to move up and down to press or relax the regulating rods and simultaneously change the degree of opening between the head portion of each regulating rod and the corresponding regulating port so that the flow from each gas outlet changes synchronously.

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: An energy-saving paddlewheel aerator is provided. A power supply system of the paddlewheel aerator includes a switching power supply converting main power to direct current, the switching power supply being connected with two terminals of a power mechanism of the paddlewheel aerator and supplying power to the paddlewheel aerator; and a solar power supply module, two ends of the solar power supply module being connected in parallel with a large-capacity capacitor, and two ends of the large-capacity capacitor being connected with the two terminals of the power mechanism of the paddlewheel aerator respectively to supply power to the paddlewheel aerator. A rated output voltage of a solar panel with sufficient power is higher than an output voltage of the switching power supply, and the rated output voltage of the solar panel with insufficient power is lower than the output voltage of the switching power supply.

Abstract: Single cell ribonucleic acid sequencing data has provided numerous avenues to monitor and study organism more thoroughly at a cellular level, including spatial arrangement of cells. An approach to predicting cellular immune response based on cellular spatial features may be presented herein. The approach may include utilizing ribonucleic acid sequence data for a single cell (“scRNA-seq”) or cell from a tissue. The approach may also include extracting spatial features of the single cell using the scRNA-seq data including cell-to-cell interactions and relative distance between cells. The approach may include predicting an immune response of a cell or cells based on the extracted spatial features.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing data compression and data decompression on lidar range images.

Abstract: A method, computer system, and a computer program product may perform document reference and reference update. One or more processors may assign marker information for a reference of a reference source. The reference may reference a target portion of a target document. The one or more processors may determine identification information for the target portion. The determined identification information may be based on content in the target portion and context information for the target portion in the target document. The one or more processors may generate a mapping of at least the marker information, the identification information, and a relative location of the target portion within the target document for use in the referencing of the target portion by the reference source.