Abstract: A light beam processor includes a first light beam amplifying apparatus, a second light beam amplifying apparatus, and a light source generating apparatus. The light source generating apparatus is configured to emit a light source beam to the first light beam amplifying apparatus. The first light beam amplifying apparatus is configured to perform an initial amplification on the light source beam to obtain an amplified light beam. The second light beam amplifying apparatus is configured to perform a secondary amplification on the amplified light beam to obtain a target light beam. The target light beam is used as an output light beam of the light beam processor.

Abstract: Techniques for sensing faces based on capturing low-resolution images and machine learning are disclosed. One or more images are captured by a face identification sensor. The face identification sensor includes a low-resolution camera, a microcontroller, and an external memory. At least two machine learning models operate simultaneously on the microcontroller. A first neural network operating on the microcontroller detects one or more faces in the image that was captured. The detecting includes a first confidence score associated with each face in the one or more faces. A second neural network operating on the microcontroller recognizes the one or more faces, wherein the recognizing is based on a face ID. A second confidence score is assigned to each of the one or more faces that were recognized. The face ID is saved to the external memory, where the second confidence score that was assigned is below a threshold.

Abstract: Techniques for sensing surfaces based on reflected light and machine learning are disclosed. A housing is used, which includes a first light source and a first photosensor. The first light source is mounted to project light downward and the first photosensor is mounted to capture light reflected upward. Data from the first photosensor is used with a machine learning model. The housing is moved a minimum distance along the surface. The distance allows for detection, by the first photosensor, of reflected light from the first light source off the surface. Light is sent from the first light source. Reflected light from the first light source is captured by the first photosensor. An output of the first photosensor is interpreted by the machine learning model. The interpreting recognizes a surface texture. A composition of the surface is identified based on the texture.

Abstract: Disclosed is a dual-link wireless ad hoc network and a security defense method in an emergency scene, aiming at comprehensively improving its security defense capability. The method comprises: sending, by a source node, the secret key and other messages which are not security defense messages through the second link; detecting, by a destination node, abnormal messages from the acquired valid messages after matching with abnormal message feature library, filtering the abnormal messages out, and quickly broadcasting the features of new abnormal messages through the first link; checking, by a new node to be added to the network, the identity and hardware state, authorizing the new node without abnormality, and broadcasting the authorization result information through the first link; adding, by other nodes receiving the security defense messages, the features of the new abnormal messages to their own abnormal message feature library, and allowing the entry of the new node.

Abstract: Disclosed is a dual-link wireless ad hoc network and a security defense method in an emergency scene, aiming at comprehensively improving its security defense capability. The method comprises: sending, by a source node, the secret key and other messages which are not security defense messages through the second link; detecting, by a destination node, abnormal messages from the acquired valid messages after matching with abnormal message feature library, filtering the abnormal messages out, and quickly broadcasting the features of new abnormal messages through the first link; checking, by a new node to be added to the network, the identity and hardware state, authorizing the new node without abnormality, and broadcasting the authorization result information through the first link; adding, by other nodes receiving the security defense messages, the features of the new abnormal messages to their own abnormal message feature library, and allowing the entry of the new node.

Abstract: A cyber-physical energy optimization control system for a hybrid electric vehicle includes an information layer which is configured to realize vehicle and road condition information collection, hybrid control unit (HCU) threshold optimization and threshold wireless update loading, and an optimized object plug-in hybrid electric bus (PHEB) as a physical layer.