Abstract: A bending portion of an endoscope and the endoscope includes a hollow tube, wherein a plurality of groups of gap portions are disposed on the hollow tube; the plurality of groups of gap portions extend in sequence along an axial direction of the hollow tube; each group of the gap portions includes at least one gap; the gaps are distributed in a circumferential direction of the hollow tube; an included angle of a preset angle is present along the length direction of the gap between a length direction of one end or a length direction of two ends of the gap and a radial direction of the hollow tube; and the radial direction of the hollow tube is perpendicular to the axial direction of the hollow tube.

Abstract: An electronic device includes a package structure having a side in a plane of orthogonal first and second directions, a conductive lead, and a plating layer on a bottom side of the conductive lead and including a first portion exposed along the side of the package structure and a second portion extending inward along a third direction that is orthogonal to the first and second directions between a lateral side of the conductive lead and the package structure. A method of fabricating an electronic device includes performing an electric etch process to etch a surface of a conductive lead exposed along a side of a package structure, and performing a plating process that forms a plating layer on an etched surface of the conductive lead.

Abstract: Methods, systems, and apparatus for generating a panoptic segmentation label for a sensor data sample. In one aspect, a system comprises one or more computers configured to obtain a sensor data sample characterizing a scene in an environment. The one or more computers obtain a 3D bounding box annotation at each time point for a point cloud characterizing the scene at the time point. The one or more computers obtain, for each camera image and each time point, annotation data identifying object instances depicted in the camera image, and the one or more computers generate a panoptic segmentation label for the sensor data sample characterizing the scene in the environment.

Abstract: A method includes obtaining, by a processing device, input data derived from a set of sensors of an autonomous vehicle (AV), generating, by the processing device using a set of lane detection classifier heads, at least one heatmap based on a fused bird's eye view (BEV) feature generated from the input data, obtaining, by the processing device, a set of polylines using the at least one heatmap, wherein each polyline of the set of polylines corresponds to a respective track of a first set of tracks for a first frame, and generating, by the processing device, a second set of tracks for a second frame after the first frame by using a statistical filter based on a set of extrapolated tracks for the second frame and a set of track measurements for the second frame, wherein each track measurement of the set of track measurements corresponds to a respective updated polyline obtained for the second frame.

Abstract: A basket body and a stone retrieval basket includes a plurality of metal sleeves arranged in parallel, and an anti-slip pattern is cut in an outer surface of each metal sleeve. Each two adjacent metal sleeves are connected by a metal wire, the metal wire is slidably inserted into one metal sleeve, and a front end of the metal wire extends out of a front end opening of the metal sleeve, and extends into a front end opening of a adjacent metal sleeve for fixation. In an open state, the plurality of metal sleeves and the metal wires are combined to form a conical basket. In the present invention, the sleeves of the basket body are made of metal material, which enhances surface friction, and improves a grasping efficiency of the basket and the durability of repeated stone retrieval.

Abstract: A service board includes a flexible board, at least one chip, and at least one optical cage layer. Each optical cage layer includes a plurality of optical cages. Each optical cage has an opening that faces a panel side and that is configured to connect to an optical module. The optical cage includes a housing and a plurality of signal pins. One end of the flexible board is electrically connected to tail ends of the plurality of signal pins, and another end of the flexible board is electrically connected to one chip, thereby shortening a connection path between the optical module and the chip, and reducing a link loss. Therefore, integration of the service board is improved, assembly is convenient, and a high-density connection and a simplified design of the service board are implemented, so that the service board is applicable to a high-density and high-speed transmission scenario.

Abstract: A semiconductor device and a method for fabricating it are disclosed. In the method, second and first substrates are bonded to obtain increased operating efficiency. Moreover, a heat dissipation unit, which includes a heat-dissipating semiconductor layer and a first heat-dissipating metal channel extending through the heat-dissipating semiconductor layer and dielectric layers on a surface thereof, is bonded to a surface of the first substrate to accelerate dissipation of heat generated during operation of the second and first substrates, imparting improved heat dissipation capacities to the semiconductor device.

Abstract: A negative pressure suction sheath includes a sheath, a negative pressure joint, and a dilator; the negative pressure joint includes a main channel cavity and a negative pressure suction channel cavity located at one side of the main channel cavity, a proximal end of the sheath is connected to a distal end of the main channel cavity, the dilator is inserted from a proximal end of the main channel cavity and extends out of a distal end of the sheath, and is detachably connected to the main channel cavity; the main channel cavity is provided with a side channel port, the negative pressure suction channel cavity is communicated with the side channel port; a cavity wall of the main channel cavity is provided with a pressure regulating port. Therefore, an operator only needs to adjust the size of the pressure regulating port to adjust the negative pressure.

Abstract: A controllable bending mechanism of an endoscope includes a hollow tube, wherein a plurality of groups of gap sets are disposed on the hollow tube; the plurality of groups of gap sets extend along an axial direction of the hollow tube; each group of gap sets includes a first gap portion and a second gap portion, wherein the first gap portion is distributed in a first circumferential direction of the hollow tube; the second gap portion is distributed in a second circumferential direction of the hollow tube; the first gap portion includes two gaps; the length of each gap is equivalent to 1/2 of that of the circumference; the second gap portion includes two gaps; the length of each gap is equivalent to 1/2 of that of the circumference.

Abstract: A mesh basket device capable of measuring the size of stones includes: a mesh basket, a push rod, a mode switching device, an opening size adjustment device, and an adjusting amount measuring device. The push rod is configured to drive the mesh basket to move. The mode switching device is used to switch between a measurement mode and a stone extraction mode. In the stone extraction mode, the opening size of the mesh basket remains unchanged during the movement of the mesh basket. The opening size adjustment device is connected to the mesh basket. In the measurement mode, the opening size adjustment device is configured to adjust the opening size of the mesh basket during the movement. The adjusting amount measuring device is configured to measure the adjusting amount of the opening size adjustment device, which is in a corresponding relationship with the opening size of the mesh basket.

Abstract: The described aspects and implementations enable efficient detection and classification of objects with machine learning models that deploy a bird's-eye view representation and are trained using depth ground truth data. In one implementation, disclosed are system and techniques that include obtaining images, generating, using a first neural network (NN), feature vectors (FVs) and depth distributions pixels of images, wherein the first NN is trained using training images and a depth ground truth data for the training images. The techniques further include obtaining a feature tensor (FT) in view of the FVs and the depth distributions, and processing the obtained FTs, using a second NN, to identify one or more objects depicted in the images.

Abstract: A method includes obtaining, by a processing device, input data derived from a set of sensors associated with an autonomous vehicle (AV). The input data includes camera data and radar data. The method further includes extracting, by the processing device from the input data, a plurality of sets of bird's-eye view (BEV) features. Each set of BEV features corresponds to a respective timestep. The method further includes generating, by the processing device from the plurality of sets of BEV features, an object flow for at least one object. Generating the object flow includes performing at least one of: multi-frame temporal aggregation or multi-frame dense motion estimation. The method further includes causing, by the processing device, a driving path of the AV to be modified in view of the object flow.

Abstract: A flexible endoscope includes a disposable endoscope body structure and a reusable camera assembly. The endoscope body structure includes a handle portion, a flexible inserting portion, and a line connecting portion. The flexible inserting portion is directly or indirectly connected to the handle portion. The line connecting portion includes a receptacle located in the handle portion and an external connecting portion located outside the handle portion. The receptacle is directly or indirectly connected with the external connecting portion. The handle portion is provided with a camera assembly access port for the camera assembly to access. After accessing and being assembled to the camera assembly access port, the camera assembly access port is able to insert into the receptacle to transmit electrical energy and/or a signal with the external connecting portion with the receptacle.

Abstract: The present invention provides a rigid endoscope including a disposable endoscope body structure and a reusable camera assembly. The endoscope body structure includes a handle portion, an endoscope body inserting portion and a line connecting portion, wherein the endoscope body inserting portion is directly or indirectly connected to the handle portion, the line connecting portion includes a receptacle located in the handle portion and an external connecting portion located outside the handle portion, and the receptacle is directly or indirectly connected with the external connecting portion; the handle portion is provided with a camera assembly access port for the camera assembly to access; after accessing and being assembled to the camera assembly access port, the camera assembly access port is able to be inserted into the receptacle to transmit electrical energy and/or a signal with the external connecting portion by using the receptacle.

Abstract: In examples, a method for manufacturing a package comprises coupling first and second semiconductor dies to a first surface of a conductive terminal; applying a dry film to a second surface of the conductive terminal opposite the first surface; removing a portion of the dry film contacting the second surface to form a dry film opening, the dry film opening having a linear, non-curved edge extending along a width of the conductive terminal; etching the second surface through the dry film opening; removing the dry film; plating the second surface; and sawing through the conductive terminal to form the package.

Abstract: The described aspects and implementations enable efficient detection and classification of objects with machine learning models that deploy a bird's-eye view representation and are trained using depth ground truth data. In one implementation, disclosed are system and techniques that include obtaining images, generating, using a first neural network (NN), feature vectors (FVs) and depth distributions pixels of images, wherein the first NN is trained using training images and a depth ground truth data for the training images. The techniques further include obtaining a feature tensor (FT) in view of the FVs and the depth distributions, and processing the obtained FTs, using a second NN, to identify one or more objects depicted in the images.

Abstract: A method includes obtaining, by a processing device, input data derived from a set of sensors of an autonomous vehicle (AV), generating, by the processing device using a set of lane detection classifier heads, at least one heatmap based on a fused bird's eye view (BEV) feature generated from the input data, obtaining, by the processing device, a set of polylines using the at least one heatmap, wherein each polyline of the set of polylines corresponds to a respective track of a first set of tracks for a first frame, and generating, by the processing device, a second set of tracks for a second frame after the first frame by using a statistical filter based on a set of extrapolated tracks for the second frame and a set of track measurements for the second frame, wherein each track measurement of the set of track measurements corresponds to a respective updated polyline obtained for the second frame.

Abstract: A train control method includes: acquiring a current control level of the train and outputting the current control level to a train traction control system; acquiring current train operation data and calculating an evaluation score according to the current train operation data by the train traction control system; and inputting the current train operation data and the evaluation score into a neural network learning system to adjust the current control level of the train to obtain a final outputted control level.

Abstract: A semiconductor package includes a semiconductor component and a plurality of leads electrically connected to the semiconductor component. Each of the leads has a first surface, and has a second surface opposite from the first surface, with a solderable metal on the first surface and the second surface. The solderable metal has a first average thickness on the first surfaces, and has a second average thickness on the second surfaces. The second average thickness is 10 percent to 80 percent of the first average thickness. The semiconductor package is formed by concurrently electroplating the solderable metal on the first surfaces and on the second surfaces. The solderable metal is electroplated on the first surfaces with a first average current, and is electroplated on the second surfaces with a second average current. The second average current is 10 percent to 80 percent of the first average current.

Abstract: A speech control method includes: receiving a target voice instruction; performing semantic parsing on the target voice instruction, and acquiring a semantic parsing result; and in response to that the semantic parsing result matches a keyword, responding to the target voice instruction by using an application corresponding to the keyword; or in response to that the semantic parsing result does not match a keyword, determining a target service type based on the semantic parsing result, and acquiring an application list corresponding to the target service type; determining a recommended application from the application list by using a first recommendation policy; and responding to the target voice instruction by using the recommended application.