Abstract: Various embodiments include methods for processing an image from an apparatus camera to recognize potentially malicious attacks on the camera. Various embodiments may include processing an image received from a camera of the apparatus using a plurality of different trained image processing models or vision pipelines to obtain a plurality of different image processing outputs, and performing a plurality of consistency checks on the plurality of different image processing outputs. Such consistency checks compare two or more selected outputs of the plurality of different outputs to detect inconsistencies that may be associated with or due to an attack on the camera. Indications of an attack on a camera may be reported to and considered by an autonomous driving system of the apparatus or otherwise addressed in one or more mitigation actions.

Abstract: Systems and techniques are described for securely managing maneuver identifications (IDs) for vehicle-based communications. For example, a method may include transmitting a message associated with a potential maneuver by the vehicle, determining a maneuver identification (ID) for the potential maneuver based on the message, and transmitting a first maneuver message (including the maneuver ID) for the potential maneuver. In another example, a method may include receiving a first maneuver message including a maneuver ID, determining that the first maneuver ID is a new maneuver ID, and determining, based on the first maneuver ID being a new maneuver ID, whether the first maneuver ID matches a second maneuver ID determined based on a previously-received message. In another example, a method may include receiving maneuver messages that include a same maneuver identification (ID) and one or more source IDs and determining whether the source ID(s) of the maneuver messages match.

Abstract: Various embodiments include methods for identifying inconsistencies in images that could be due to malicious attacks. Various embodiments may include receiving a plurality of camera images from one or more cameras of an apparatus (e.g., a vehicle), performing a plurality of different processes on the plurality of images to detect different types of image inconsistencies, using results of the plurality of different processes on the plurality of images to recognize a vision attack and performing one or more mitigation actions in response to recognizing a vision attack. The plurality of different processes may include temporal consistency checks on the plurality of images spanning a period of time, inconsistency counter checks on the plurality of images that determine whether a number of inconsistencies in camera images satisfies a threshold, past history checks on the plurality of images comparing objects previously recognized to objects recognized in currently obtained images.

Abstract: A system is provided for generating an agentic system. The system receives a request including natural language text input from an interaction interface, generates an agent design prompt including an agent task description, agent output instructions, and scored agents based on the request, and inputs the agent design prompt into an agent design generator to generate one or more agents. A tool design prompt is generated to include the one or more generated agents, a tool task description, tool output instructions, and scored tools. The tool design prompt is then inputted into a tool generator to generate one or more tools, and the system generates and outputs the agentic system including the one or more generated agents and the one or more generated tools.

Abstract: This document relates to facilitating interactions between a user and multiple agents, such as agents that provide different generative machine learning capabilities. In certain disclosed implementations, user input is received and two or more agents are selected to perform a particular task for the user. Then, interactions with the agents are coordinated on behalf of the user based on agent definitions of the selected agents. A result of the particular task is received and provided to the user. For instance, the user input can be received via a chat interface that allows the user to conduct a single chat session that involves the particular task being performed by multiple agents, thus giving the user the appearance of interacting with a single agent.

Abstract: Embodiments include methods performed by a vehicle-to-everything (V2X) system for protecting proprietary data within misbehavior reports. The method includes receiving a misbehavior report from a V2X system participant, obtaining sensor data from the received misbehavior report indicating that a misbehavior condition has occurred, determining whether the sensor data indicating that a misbehavior condition has occurred includes encrypted proprietary information. In response to determining that the sensor data includes encrypted proprietary information, the V2X system identifies an entity that owns the encrypted proprietary information, transmits the misbehavior report including the encrypted proprietary information to the entity that owns the encrypted proprietary information, and receives a confirmation report from the entity that owns the encrypted proprietary information indicating whether the misbehavior report is accurate.

Abstract: Certain aspects of the present disclosure provide techniques and apparatus for autonomous driving. In an example method, sensor data associated with an external environment of a vehicle is obtained. A set of prompts for input to a machine learning model trained to assist with autonomous driving for the vehicle is selected via a manager, based at least in part on the sensor data. A set of output tokens corresponding to the set of prompts is generated using the trained machine learning model and the set of prompts. The set of output tokens is associated with at least one of (i) one or more adversarial objects in the environment or (ii) one or more safety events in the environment. Information associated with the set of output tokens is output via the manager for a user of the vehicle.

Abstract: A user equipment (UE) for vehicle-to-everything (V2X) transmission is provided. The UE includes: at least one transceiver; at least one memory; and at least one processor, coupled with the at least one transceiver and the at least one memory, the at least one processor configured to: detect a second position of a second UE with respect to a first position of the UE; generate an alert time window based on the first position of the UE and the second position of the second UE; and schedule at least one transmission from the UE to the second UE based on the alert time window.

Abstract: Systems and techniques are described herein for attack detection. For instance, a process can include determining, from a received query, a first key and value pair for an application programming interface (API); analyzing the first key and value pair to determine whether the first key and value pair are associated with an attack; and based on a determination that the first key and value pair are not associated with an attack, calling the API using the first key and value pair.

Abstract: A stereo camera device includes: a storage unit that stores correlation information of a parallax error in a horizontal direction with respect to a vertical shift of a plurality of images respectively captured by a plurality of cameras through a refractile body that refracts light; a matching unit that matches the plurality of images obtained by the plurality of cameras capturing a same subject to obtain a vertical shift of the plurality of images; a parallax error estimation unit that estimates a parallax error of the plurality of images having been matched based on the vertical shift and the correlation information; and a distance calculation unit that corrects, using the parallax error, a parallax of the image having been matched.

Abstract: Systems and techniques are described herein for attack mitigation. For instance, a process can include allocating a first masked gadget, of a plurality of masked gadgets, for executing a first machine learning (ML) model; allocating a second masked gadget, of the plurality of masked gadgets, for executing a second ML model; and using the first masked gadget with the first ML model concurrently with the second masked gadget with the second ML model.

Abstract: The present disclosure provide techniques for object detection. A method may include extracting a plurality of vehicle-to-everything (V2X) features from a first plurality of V2X messages, wherein the first plurality of V2X messages comprise information about one or more first objects in a scene during a first time period; generating a bird's eye view (BEV) representation of the scene based on, at least, projecting the plurality of V2X features from a V2X space to a BEV space; and detecting, based on the BEV representation, at least a first object among the one or more first objects in the scene.

Abstract: In a power control method in a wireless local area network, a beamformer sends a null data packet (NDP) via a plurality of spatial streams to beamformees. After receiving the NDP, a beamformee determines a first parameter for power control based on RSSIs of a plurality of receive antennas when the beamformee receives the NDP, or based on SNRs of a plurality of spatial streams when the beamformee receives the NDP. The beamformee then generates a beamforming report comprising the first parameter, and sends the beamforming report to the beamformer.

Abstract: Systems and techniques are described herein for attack detection. For instance, a process can include determining, from a received query, a first key and value pair for an application programming interface (API); analyzing the first key and value pair to determine whether the first key and value pair are associated with an attack; and based on a determination that the first key and value pair are not associated with an attack, calling the API using the first key and value pair.

Abstract: A method for selecting a clock source includes a first network device that obtains synchronization offset data between the first network device and a second network device through a first port in a clock synchronization failed state. The first network device determines, based on the synchronization offset data, whether to refer to, during clock source selection, clock information received by the first port.

Abstract: The present application is applicable to the field of video processing, and provides a method for generating a rotation direction of a gyroscope. The method comprises: obtaining an acceleration value and an angular velocity value of an IMU in real time, taking the acceleration value as a first acceleration value, and estimating an attitude from the IMU to a world coordinate system; converting the first acceleration value from an IMU coordinate system to a world coordinate system to obtain a second acceleration value; in the world coordinate system, filtering a second acceleration, and filtering out a gravitational acceleration to obtain a third acceleration; converting the third acceleration into the IMU coordinate system to obtain acceleration components in the X axis, the Y axis and the Z axis of a fourth acceleration; and determining the rotation direction of the gyroscope according to the acceleration components.

Abstract: A photosensitive assembly for camera module includes a support substrate, a photosensitive chip electrically connected to the support substrate, and a molded part injection-molded on the support substrate, wherein the molded part includes a molded base and a first protrusion, a peripheral side of the molded base has a spacing distance from an edge of the support substrate, the first protrusion is extended outward from the peripheral side of the molded base to the edge of the support substrate, and a volume proportion of silicon powder in the molded part is 75%˜85%. By optimizing the composition of the molding liquid of the molded part, the occurrence of mold sticking is reduced.

Abstract: A device for separating and sequestrating carbon dioxide coupled with cold storage in mixed gas via hydrate method, which belongs to the technical field of application of natural gas hydrates includes a gas compression device, a refrigeration cycle device, a hydrate formation/decomposition device, a hydrate cold storage device, a water circulation device and a sensing and monitoring device; taking the separation and sequestration of biogas as an example, the refrigeration cycle device enables the cooling of biogas, decomposition of gas at all levels, hydrate, and circulating water to provide the low-temperature conditions required for hydrate formation; the hydrate cold energy storage device can fully use the latent heat of hydrate phase change to provide the required cooling capacity on the user side; the water circulation device can realize the recycling of decomposition water to ensure the continuous formation of hydrate.

Abstract: A computer-implemented method can be used to generate a composite image. Image data can be received, and a low-resolution composite image can be generated. A first machine-learning model can be applied to the low-resolution image and to a mask of the image data to generate a second image. A second machine-learning model can be applied to the image data to perform color mapping from an optimal patch to a masked foreground area of the image data. The color mapping can be used to generate a third image. A composite image can be generated using the second image and the third image.