Abstract: The disclosure generally relates to a system comprising a memory and a processor configured to access the memory and execute the machine-executable instructions stored on the memory to determine that a target vehicle is engaging in an abnormal driving, determine that the abnormal driving exceeds a notification threshold parameter of an abnormal driving notification system, generate a notification, and monitor the ego vehicle and/or driver to alter notification threshold values based on the reactions of the ego vehicle and/or driver inputs.

Abstract: A server for communication-efficient model aggregation in federated networks for connected vehicle applications is provided. The server includes a controller programmed to: obtain contributions of a plurality of vehicles in a federated learning framework; determine weights for local gradients received from the plurality of vehicles based on the contributions; adjust the weights based on a comparison of potential functions for the plurality of vehicles; and aggregate the local gradients based on the adjusted weights to obtain a global model.

Abstract: A trailer monitoring system for a vehicle includes a processor and a memory communicably coupled to the processor. The memory stores instructions that when executed by the processor cause the processor to receive, from at least one monocular camera mounted to a vehicle towing a trailer, at least one monocular camera image of at least a portion of the trailer. The instructions also cause the processor to generate at least one depth map based on the at least one monocular camera image. The instructions further cause the processor to identify a trailer irregularity based on the at least one depth map.

Abstract: An object detection system for a vehicle includes a processor and a memory communicably coupled to the processor. The memory stores instructions that when executed by the processor cause the processor to receive a priority assignment set by an occupant of a vehicle. The priority assignment corresponds to a customization of at least one warning characteristic. The at least one warning characteristic includes a distance threshold. The instructions also cause the processor to detect an object in an external environment of the vehicle and apply the priority assignment to the object. The instructions further cause the processor to issue a warning according to the at least one warning characteristic. According to the distance threshold, the warning is issued when a distance from the vehicle to the object meets the distance threshold.

Abstract: Systems and methods are provided for detecting when a vehicle engages in anomalous driving behavior and managing nearby vehicles to mitigate risk to the nearby vehicles due to the anomalous driving behavior. According to some embodiments, the methods and systems comprise receiving an indication that an anomalous driving behavior of a first vehicle is detected, and identifying a plurality of vehicles within a geographic area of the first vehicle based on receiving the indication. The methods and systems further include generating an arrangement for the plurality of vehicles based on the detected anomalous driving behavior, determining a target speed for each of the plurality of vehicles based on the determined arrangement and a current speed of each of the plurality of vehicles, and communicating control messages to each of the plurality of vehicles. The control messages comprising the target speed to distribute the plurality of vehicles into the determined arrangement.

Abstract: Systems and methods are provided for programmatically verifying an origin of abnormal driving. Some examples of abnormal driving may include aggressive driving (e.g., tailgating, cut-in lane, etc.), distracted driving (e.g., swerving, delayed reaction, etc.), and reckless driving (e.g., green light running, lane change without signaling, etc.). For example, the systems and methods may receive an identification of a second vehicle performing abnormal driving from an ego vehicle; initiate a verification process of the identification of the abnormal driving; access driving data associated with an origin of the abnormal driving, wherein the driving data includes the ego vehicle and the second vehicle; using the driving data, determine a confirm or deny decision regarding the identification of the second vehicle from the ego vehicle; and provide the confirm or deny decision.

Abstract: A parking assist system for a vehicle includes a processor and a memory communicably coupled to the processor. The memory stores instructions that when executed by the processor cause the processor to detect a parking space for a vehicle. The instructions further cause the processor to identify one or more parking parameters related to parking the vehicle in the parking space. The parking parameters include a point-of-interest associated with the parking space. The instructions further cause the processor to determine a suitability of the parking space based on the one or more parking parameters. Determining the suitability of the parking space includes predicting a point-of-interest-related action to be performed by an occupant after parking the vehicle in the parking space.

Abstract: Compensation can be made for mismatch in abnormal driving behavior detection between an ego vehicle and its human driver. Sensor data of a surrounding driving environment can be acquired. The acquired sensor data can be analyzed to detect abnormal driving behavior by a vehicle in the surrounding driving environment. Responsive to detecting abnormal driving behavior by the vehicle in the surrounding driving environment, the abnormal driving behavior can be classified as to whether it is severe. If the abnormal driving behavior is classified as not being severe, an amplified version of the abnormal driving behavior can be caused to be presented to the human driver of the ego vehicle.

Abstract: Systems include a controller programmed to instruct a sensor to detect an object in an environment of a vehicle in response to an activation of a fall-on-car prevention function, determine a first value of a parameter pertaining to the object at a first time with the sensor, determine a second value of the parameter pertaining to the object at a second time with the sensor, and automatically cause an action to occur based on a comparison of the first value of the parameter and the second value of the parameter.

Abstract: A method for detecting objects that are likely to fall from vehicle cargo includes obtaining first data of a cargo of a first vehicle captured at a first time and second data of the cargo captured by a second vehicle in an environment of the first vehicle at a second time after the first time, determining a first distance between two key points of the first data or a first shape constructed by key points of the first data and a second distance between two key points of the second data or a second shape constructed by key points of the second data, and detecting a movement of the cargo based on a comparison of the first distance and the second distance or a comparison of the first shape and the second shape.

Abstract: Systems, methods, and vehicles for classifying driving behavior of a target vehicle are provided. The systems include a controller programmed to identify a target vehicle associated with a pseudo-id, obtain a first tag representing first driving behavior of the target vehicle during a first period of time along with the pseudo-id based on first motion data of the target vehicle, obtain a second tag representing second driving behavior of the target vehicle during a second period of time along with the pseudo-id based on second motion data of the target vehicle, the second period of time being after the first period of time, and classify driving behavior of the target vehicle based on interpretation of the first tag and the second tag.

Abstract: A method according to some embodiments includes sensing, by a sensor set of the ego vehicle, a remote vehicle to generate sensor data describing driving behavior of the remote vehicle. The method further includes classifying a type of the remote vehicle based on the sensor data. The method further includes retrieving a behavior profile for the type of the remote vehicle that identifies criteria for the type of the remote vehicle and classifications of abnormal behavior and normal behavior based on the criteria. The method further includes comparing the behavior of the remote vehicle to the behavior profile for the type of the remote vehicle. The method further includes determining that the behavior of the remote vehicle is normal based on the comparing.

Abstract: Systems, methods, and other embodiments described herein relate to improving the performance of a device in different geographic locations by using transfer learning to provide a customized learning model for the different locations. In one embodiment, a method includes receiving segments of a model from separate members in a geographic hierarchy and assembling the segments into the model. The segments include at least a first segment, a second segment, and a third segment. The method includes processing sensor data using the model to provide an output for assisting a device.

Abstract: Systems and methods are provided for improving the performance of anomaly detection, including detecting abnormal driving behavior at locations and in driving situations where the accuracy of detecting abnormal driving behavior has decreased or known to cause inaccurate performance responses. For example, the anomaly detection mechanism may detect abnormal driving behavior (e.g., delayed responses, road rage, swerving, etc.) or environmental anomalies (e.g., potholes, fallen trees, etc.), and particular driving situations or locations can have repeatedly shown decreased rates of accuracy with responding to these anomalies, including abnormal driving behavior (e.g., delayed driver responses to left-turns, U-turns, traffic light, railroad crossing, construction zones, etc.) or situations that are known to cause inaccurate performance responses (e.g., potholes, fallen trees, etc.). Embodiments can revise the anomaly detection system to operate the vehicle more consistently with other human drivers.

Abstract: Methods and systems perform incremental learning object detection in images and/or videos without catastrophic forgetting of previously-learned object classes. A two-stage neural network object detector is trained to locate and identify objects pertaining to an additional object class by iteratively updating the two-stage neural network object detector until an overall detection accuracy criterion is met. The updating is performed so as to balance minimizing a loss of an initial ability to locate and identify objects pertaining to the previously-learned object classes and maximizing an ability to additionally locate and identify the objects pertaining to the additional object class.

Abstract: A touch detection system is provided. The touch detection system includes a hardware processor. The touch detection system further includes a touch input surface mounted using a set of torque sensors, where a location of a touch on the touch input surface is calculated by the hardware processor based on collective results from the set of torque sensors, the touch comprising a user input.

Abstract: A touch detection system is provided. The touch detection system includes a hardware processor. The touch detection system further includes a touch input surface mounted using a set of torque sensors, where a location of a touch on the touch input surface is calculated by the hardware processor based on collective results from the set of torque sensors, the touch comprising a user input.

Abstract: Methods and systems perform incremental learning object detection in images and/or videos without catastrophic forgetting of previously-learned object classes. A two-stage neural network object detector is trained to locate and identify objects pertaining to an additional object class by iteratively updating the two-stage neural network object detector until an overall detection accuracy criterion is met. The updating is performed so as to balance minimizing a loss of an initial ability to locate and identify objects pertaining to the previously-learned object classes and maximizing an ability to additionally locate and identify the objects pertaining to the additional object class.

Abstract: Parallel linguistic corpora used to train cognitive translation systems are built from robot plans. A collection of robot plans in various languages is analyzed in candidate pairings to see if any two plans are directed to the same task. If so, verbal instructions in different languages from the two plans (such as plan names) are assumed to have the same meaning, and are stored in association as part of the bilingual corpus. If each plan has multiple steps, the corresponding steps from the two plans can also be stored in association as part of the corpus. Robot plans are modeled as finite state automata having a plurality of links representing sensory states and a plurality of nodes representing motor actions, and the determination of plan similarity is based on a comparison of the two FSAs.

Abstract: The disclosure includes implementations for providing ground adjustment for an in-vehicle augmented reality system. A system may include a three-dimensional heads-up display unit (“3D HUD”) installed in a vehicle. The system may include a memory storing instructions that, when executed, cause the system to: determine a plurality of elevation values for a plurality of points on a road surface, where each elevation value is associated with a point from the plurality of points and describes the elevation of that point; identify a graphic for display on the 3D HUD, where the graphic is associated with at least one point from the plurality of points; determine which location of the 3D HUD is associated with the at least one point associated with the graphic; and display the graphic at the location of the 3D HUD so that the graphic superposes the point when viewed by a driver.