Abstract: This disclosure is directed to a liquid cooling device having at least two collecting tanks and at least one pair of heat-exchange plates. The tanks are separated from each other. Each of the collecting tanks has a joint tube. Each of the heat-exchange plates is in elongated shape and the collecting tanks are connected serially by the heat-exchange plates. The two collecting tanks are connected by the pair of heat-exchange plates. Each of the heat-exchange plates has a channel extended along the longitudinal direction thereof. The channels in the heat-exchange plates are connected to the collecting tanks at two ends of the heat-exchange plates, respectively. The longitudinal directions of the channels of the heat-exchange plates between the collecting tanks are parallel to each other.

Abstract: A display device having a substrate; a first thin film transistor (TFT) on the substrate, the first TFT having a first active layer, a first gate insulator, and a first gate electrode; a second TFT on the substrate, the second TFT having a second active layer, a second gate insulator and a second gate electrode. The first gate insulator is disposed between the first gate electrode and the first active layer, and the first gate insulator is in contact with the first active layer. The second gate insulator is disposed between the second gate electrode and the second active layer, and the second gate insulator is in contact with the second active layer. The first active layer is a different material than the second active layer, and a hydrogen concentration of the second gate insulator is less than a hydrogen concentration of the first gate insulator.

Abstract: An optical fingerprint recognition device includes a light-emitting diode (LED) array and a fingerprint sensing device. The LED array includes a central LED area and an edge LED area, and configured to display a light source pattern in response to a fingerprint sensing request. The light source pattern includes a central portion and a surrounding portion. During displaying the light source pattern, a plurality of red display subpixels of the central LED area are not illuminating and a plurality of red display subpixels of the edge LED area are illuminating. The fingerprint sensing device generates a first fingerprint image according to a plurality of first sensing signals obtained from a plurality of first sensing pixel area, and the first fingerprint image is adapted to be used for examining whether a finger which triggers the fingerprint sensing request is real or fake.

Abstract: The present invention provides a flexible glass and manufacturing method thereof. The flexible glass includes a first straight part and a second straight part on two opposite ends thereof, a recess formed between the first straight part and the second straight part, and a pre-bent curve connection part disposed corresponding to the recess. The first straight part and the second straight part are not arranged on the same plane. The flexible glass has a first lateral side and a second lateral side, and the recess sinks from the first lateral side toward the second lateral side. Therefore, the flexible glass is provided with a greater bendability.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for controlling an agent. One of the methods includes receiving a current observation characterizing a current state of the environment as of the time step; generating an embedding of the current observation; processing scene memory data comprising embeddings of prior observations received at prior time steps using an encoder neural network, wherein the encoder neural network is configured to apply an encoder self-attention mechanism to the scene memory data to generate an encoded representation of the scene memory data; processing the encoded representation of the scene memory data and the embedding of the current observation using a decoder neural network to generate an action selection output; and causing the agent to perform the selected action.

Abstract: The present disclosure describes a storage device including a top panel, a bottom panel, a back panel, a front panel, and two side panels configured to form an enclosed volume. The storage device further includes multiple slots disposed at inner surfaces of the two side panels and configured to hold a substrate, a gas diffuser disposed at an inner surface of the back panel and configured to provide a purge gas to the enclosed volume, an isolation gas device disposed on an inner surface of the top panel and adjacent to a top portion of the front panel, and an isolation gas line configured to connect the isolation gas device to the gas diffuser. The isolation gas device is configured to inject the purge gas into a front portion of the storage device and in a direction from the top panel toward the bottom panel.

Abstract: The present disclosure describes a storage device including a top panel, a bottom panel, a back panel, a front panel, and two side panels configured to form an enclosed volume. The storage device further includes multiple slots disposed at inner surfaces of the two side panels and configured to hold a substrate, a gas diffuser disposed at an inner surface of the back panel and configured to provide a purge gas to the enclosed volume, an isolation gas device disposed on an inner surface of the top panel and adjacent to a top portion of the front panel, and an isolation gas line configured to connect the isolation gas device to the gas diffuser. The isolation gas device is configured to inject the purge gas into a front portion of the storage device and in a direction from the top panel toward the bottom panel.

Abstract: A time-resolved fluorescence immunochromatography test paper card for detecting butralin, which comprises a cover body and a housing body, wherein the cover body is provided with a test hole, a loading hole and a through-hole, an isolating mechanism is arranged in the test hole and the loading hole, the isolating mechanism comprises a first isolating ring and a second isolating ring, the top lateral walls of which are respectively provided with a first lug boss and a second lug boss, one end of the upper surface of the working board is concave towards the inner of the working board to provide a groove, there is a nitrocellulose membrane, a binding pad, a sample pad and a mark zone successively provided between the water absorbing block and the other end of the working board, and the lateral wall at one end of the working board is provided with a bump.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for controlling an agent. One of the methods includes receiving a current observation characterizing a current state of the environment as of the time step; generating an embedding of the current observation; processing scene memory data comprising embeddings of prior observations received at prior time steps using an encoder neural network, wherein the encoder neural network is configured to apply an encoder self-attention mechanism to the scene memory data to generate an encoded representation of the scene memory data; processing the encoded representation of the scene memory data and the embedding of the current observation using a decoder neural network to generate an action selection output; and causing the agent to perform the selected action.

Abstract: The present disclosure describes a method for controlling gas supplies and an example system for performing the method. The method includes providing a first setting to configure a gas supply device to supply a first gas mixture to a substrate carrier holding a first substrate. The method further includes receiving critical dimension (CD) data measured on the first substrate after the first substrate completes a process operation. The method further includes, in response to the CD data being outside a predetermined range, providing a second setting to configure the gas supply device to supply a second gas mixture to the substrate carrier holding a second substrate that has yet to undergo the process operation.

Abstract: A portable pesticide application equipment with accurate application, which includes a pesticide tank. Rollers are arranged at the bottom of the pesticide tank. An inclined pesticide guide plate is arranged below the filter. A side connector is arranged at one side of the pesticide tank. The side connector is connected with a long hose. The long hose is connected with a pesticide discharging device. The pesticide discharging device includes a pesticide control box, two sides of which are hinged with a first iron handle and a second iron handle. A first nozzle is connected with the bottom end of the first iron handle while a second nozzle is connected with the bottom end of the second iron handle. The upper end of the first iron handle is inserted into a first support rod while the upper end of the second iron handle is inserted into a second support rod.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for controlling an agent. One of the methods includes receiving a current observation characterizing a current state of the environment as of the time step; generating an embedding of the current observation; processing scene memory data comprising embeddings of prior observations received at prior time steps using an encoder neural network, wherein the encoder neural network is configured to apply an encoder self-attention mechanism to the scene memory data to generate an encoded representation of the scene memory data; processing the encoded representation of the scene memory data and the embedding of the current observation using a decoder neural network to generate an action selection output; and causing the agent to perform the selected action.

Abstract: The present disclosure describes a storage device including a top panel, a bottom panel, a back panel, a front panel, and two side panels configured to form an enclosed volume. The storage device further includes multiple slots disposed at inner surfaces of the two side panels and configured to hold a substrate, a gas diffuser disposed at an inner surface of the back panel and configured to provide a purge gas to the enclosed volume, an isolation gas device disposed on an inner surface of the top panel and adjacent to a top portion of the front panel, and an isolation gas line configured to connect the isolation gas device to the gas diffuser. The isolation gas device is configured to inject the purge gas into a front portion of the storage device and in a direction from the top panel toward the bottom panel.

Abstract: A time-resolved fluorescence immunochromatography test paper card for detecting butralin, which comprises a cover body and a housing body, wherein the cover body is provided with a test hole, a loading hole and a through-hole, an isolating mechanism is arranged in the test hole and the loading hole, the isolating mechanism comprises a first isolating ring and a second isolating ring, the top lateral walls of which are respectively provided with a first lug boss and a second lug boss, one end of the upper surface of the working board is concave towards the inner of the working board to provide a groove, there is a nitrocellulose membrane, a binding pad, a sample pad and a mark zone successively provided between the water absorbing block and the other end of the working board, and the lateral wall at one end of the working board is provided with a bump.

Abstract: A portable pesticide application equipment with accurate application, which includes a pesticide tank. Rollers are arranged at the bottom of the pesticide tank. An inclined pesticide guide plate is arranged below the filter. A side connector is arranged at one side of the pesticide tank. The side connector is connected with a long hose. The long hose is connected with a pesticide discharging device. The pesticide discharging device includes a pesticide control box, two sides of which are hinged with a first iron handle and a second iron handle. A first nozzle is connected with the bottom end of the first iron handle while a second nozzle is connected with the bottom end of the second iron handle. The upper end of the first iron handle is inserted into a first support rod while the upper end of the second iron handle is inserted into a second support rod.

Abstract: Implementations are directed to training a machine learning model that, once trained, is used in performance of robotic grasping and/or other manipulation task(s) by a robot. The model can be trained using simulated training examples that are based on simulated data that is based on simulated robot(s) attempting simulated manipulations of various simulated objects. At least portions of the model can also be trained based on real training examples that are based on data from real-world physical robots attempting manipulations of various objects. The simulated training examples can be utilized to train the model to predict an output that can be utilized in a particular task—and the real training examples used to adapt at least a portion of the model to the real-world domain can be tailored to a distinct task. In some implementations, domain-adversarial similarity losses are determined during training, and utilized to regularize at least portion(s) of the model.

Abstract: Implementations are directed to training a machine learning model that, once trained, is used in performance of robotic grasping and/or other manipulation task(s) by a robot. The model can be trained using simulated training examples that are based on simulated data that is based on simulated robot(s) attempting simulated manipulations of various simulated objects. At least portions of the model can also be trained based on real training examples that are based on data from real-world physical robots attempting manipulations of various objects. The simulated training examples can be utilized to train the model to predict an output that can be utilized in a particular task—and the real training examples used to adapt at least a portion of the model to the real-world domain can be tailored to a distinct task. In some implementations, domain-adversarial similarity losses are determined during training, and utilized to regularize at least portion(s) of the model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for controlling an agent. One of the methods includes receiving a current observation characterizing a current state of the environment as of the time step; generating an embedding of the current observation; processing scene memory data comprising embeddings of prior observations received at prior time steps using an encoder neural network, wherein the encoder neural network is configured to apply an encoder self-attention mechanism to the scene memory data to generate an encoded representation of the scene memory data; processing the encoded representation of the scene memory data and the embedding of the current observation using a decoder neural network to generate an action selection output; and causing the agent to perform the selected action.

Abstract: Implementations are directed to training a machine learning model that, once trained, is used in performance of robotic grasping and/or other manipulation task(s) by a robot. The model can be trained using simulated training examples that are based on simulated data that is based on simulated robot(s) attempting simulated manipulations of various simulated objects. At least portions of the model can also be trained based on real training examples that are based on data from real-world physical robots attempting manipulations of various objects. The simulated training examples can be utilized to train the model to predict an output that can be utilized in a particular task—and the real training examples used to adapt at least a portion of the model to the real-world domain can be tailored to a distinct task. In some implementations, domain-adversarial similarity losses are determined during training, and utilized to regularize at least portion(s) of the model.