Abstract: A method for reducing the number of neurons in a trained deep neural network (DNN) includes classifying layer types in a plurality of hidden layers; evaluating the accuracy of the DNN using a validation set of data; and generating a layer specific ranking of neurons, wherein the generating includes: analyzing, using the validation set of data for one or more of the plurality of hidden layers, the activation function for each neuron in the analyzed layers to determine an activation score for each neuron; and ranking, on a layer type basis, each neuron in the analyzed layers based on the neuron's activation score to generate a layer specific ranking of neurons. The method further includes removing a number of lower ranked neurons from the DNN that does not result in the DNN after the removal of selected lower ranked neurons to fall outside of an accuracy threshold limit.

Abstract: Systems and method are provided for monitoring an operator of a vehicle. In one embodiment, a method includes: receiving, by a processor, data generated by the vehicle; determining, by the processor, causal time series event data based on the received data; computing, by the processor, a score for at least one of safety and quality based on a first machine learning model and the causal time series event data; computing, by the processor, at least one explanation for the score based on a second machine learning model; and generating, by the processor, display data to display at least one of the causal time series event data, the score, and the explanation to an end user.

Abstract: A system and method for predicting a collision between a host vehicle and a target vehicle operating on a multi-lane roadway may include determining the host and target vehicles are converging from respective first and second lanes to a third lane intermediate the first and second lanes. A predetermined set of conditions is evaluated including relationships among the host and target vehicle separations and speeds, and a collision predicted based upon the evaluation.

Abstract: Methods and systems to implement sensor fusion to determine collision potential for a vehicle include identifying a specific intersection that the vehicle is approaching, and identifying collision potential scenarios associated with one or more paths through the specific intersection. Each collision potential scenario defines a risk of a collision between the vehicle and an object in a specified area. A weight with which one or more information sources of the vehicle are considered is adjusted for each collision potential scenario such that a highest weight is given to one or more of the one or more information sources that provide most relevant and reliable information about the specified area. Sensor fusion is implemented based on the adjusting the weight of the one or more information sources and performing detection based on the sensor fusion, and an alert is provided or actions are implemented according to the detection.

Abstract: Systems and methods for vehicle-target localization, identification and indication. The system includes a vehicle module operating onboard a vehicle and configured to receive a request for a tether and send the tether responsive thereto, and a target module operating on a target external to the vehicle. The target module can generate target profile data (TPD) comprising location information for the target, such as, a first component that is a function of a detected first WiFi signal having a first signal strength. The request for the tether including the TPD is transmitted to the vehicle module; a tether of secured communications is established between the vehicle module and target module. Real-time TPD is exchanged during the tether and the vehicle module performs the V-T localization as a function of the RT TPD, and commands the vehicle based on V-T localization data.

Abstract: A system and method for determining the presence of a hidden hazard may include identification of an operational scene for a host vehicle, and identification of an operational situation for the host vehicle. Information from a plurality of proximity sensors is collected and classified. A plurality of hidden hazard presence probabilities corresponding to the information from each of the plurality of proximity sensors, the operational scene, the operational situation, and at least one of a comparative process and a dynamic neural network process are estimated. A fusion process may be performed upon the plurality of hidden hazard presence probabilities to determine the presence of a hidden hazard.

Abstract: A system and method for predicting a collision between a host vehicle and a target vehicle operating on a multi-lane roadway may include determining the host and target vehicles are converging from respective first and second lanes to a third lane intermediate the first and second lanes. A predetermined set of conditions is evaluated including relationships among the host and target vehicle separations and speeds, and a collision predicted based upon the evaluation.

Abstract: A system and method for determining the presence of a hidden hazard may include identification of an operational scene for a host vehicle, and identification of an operational situation for the host vehicle. Information from a plurality of proximity sensors is collected and classified. A plurality of hidden hazard presence probabilities corresponding to the information from each of the plurality of proximity sensors, the operational scene, the operational situation, and at least one of a comparative process and a dynamic neural network process are estimated. A fusion process may be performed upon the plurality of hidden hazard presence probabilities to determine the presence of a hidden hazard.

Abstract: Methods and systems to implement sensor fusion to determine collision potential for a vehicle include identifying a specific intersection that the vehicle is approaching, and identifying collision potential scenarios associated with one or more paths through the specific intersection. Each collision potential scenario defines a risk of a collision between the vehicle and an object in a specified area. A weight with which one or more information sources of the vehicle are considered is adjusted for each collision potential scenario such that a highest weight is given to one or more of the one or more information sources that provide most relevant and reliable information about the specified area. Sensor fusion is implemented based on the adjusting the weight of the one or more information sources and performing detection based on the sensor fusion, and an alert is provided or actions are implemented according to the detection.

Abstract: Systems and methods of determining collision potential for a vehicle involve identifying a specific intersection that the vehicle is approaching, and determining an intention of a driver of the vehicle to traverse a specific path through the specific intersection. A method includes identifying an obstructed portion of a relevant area for the specific path of the vehicle through the specific intersection. An object travelling within the relevant area will intersect with the specific path of the vehicle and one or more sensors of the vehicle are blocked from detections in the obstructed portion of the relevant area. An alert is provided or actions are implemented based on the obstructed portion of the relevant area.

Abstract: A vehicle cleaning system for cleaning a vehicle including a processor and a memory including computer-executable instructions that, when executed by the processor, cause the processor to perform operations. The operations include determining that a passenger compartment of the vehicle requires cleaning and identifying a type of cleaning required. The operations also include that at least one of a fragrance system, a dry particulate removal system, and a wet particulate removal system is activated in response to the type of cleaning required.

Abstract: A method for reducing the number of neurons in a trained deep neural network (DNN) includes classifying layer types in a plurality of hidden layers; evaluating the accuracy of the DNN using a validation set of data; and generating a layer specific ranking of neurons, wherein the generating includes: analyzing, using the validation set of data for one or more of the plurality of hidden layers, the activation function for each neuron in the analyzed layers to determine an activation score for each neuron; and ranking, on a layer type basis, each neuron in the analyzed layers based on the neuron's activation score to generate a layer specific ranking of neurons. The method further includes removing a number of lower ranked neurons from the DNN that does not result in the DNN after the removal of selected lower ranked neurons to fall outside of an accuracy threshold limit.

Abstract: Technical solutions are described herein for providing driver notification in a vehicle. An example system includes one or more sensors that measure one or more attributes of a remote object in a predetermined vicinity of the vehicle. The system further includes an output device that provides a notification to a driver. The system further includes a remote object monitoring system that generates a driver notification to be provided via the output device based on the attributes of the remote object. Generating the driver notification includes determining a recklessness score for the remote object based on the attributes of the remote object. Generating the driver notification further includes, in response to the recklessness score exceeding a predetermined threshold, generating the driver notification that comprises a directional information that provides a spatial awareness of a location of the remote object in relation to the vehicle.

Abstract: Technical solutions are described herein for adjusting a notification system of a vehicle. An example method includes receiving, by a controller, an occupant profile of an occupant of the vehicle. The method further includes adjusting, by the controller, a haptic array according to the occupant profile, the haptic array comprising a plurality of haptic actuators, the adjusting comprising activating a first subset of the haptic actuators and deactivating a second subset of the haptic actuators. Further, the method includes providing, by the haptic array, a notification to the occupant of the vehicle by providing a haptic feedback using the activated haptic actuators.

Abstract: A system and method for extracting a relevant phrase from text. The system and method may build a plurality of n-gram phrases using a seed from a seed list as a start, a middle, or an end of each n-gram phrase. The seed list may be directed to a specific vehicle system and each seed may indicate a symptom, part, or action to extract relevant phrases from vehicle information verbatims. The plurality of n-gram phrases may be filtered to obtain one or more relevant phrases. The filtering process may include calculating an external relevance factor, an internal relevance factor, or a context pattern relevance factor.

Abstract: A system and method for extracting a relevant phrase from text. The system and method may build a plurality of n-gram phrases using a seed from a seed list as a start, a middle, or an end of each n-gram phrase. The seed list may be directed to a specific vehicle system and each seed may indicate a symptom, part, or action to extract relevant phrases from vehicle information verbatims. The plurality of n-gram phrases may be filtered to obtain one or more relevant phrases. The filtering process may include calculating an external relevance factor, an internal relevance factor, or a context pattern relevance factor.

Abstract: A system and method of modifying a vehicle service database includes: accessing a database containing previously-received symptom text that has been associated with a vehicle identifier and one or more vehicular service solutions for the previously-received symptom text; determining a statistical likelihood that one or more additional vehicular service solutions apply to previously-received symptom text based on a correlation between the previously-received symptom text and additional vehicular service solutions; determining that the statistical likelihood is above a predetermined threshold; and associating the previously-received symptom text with the additional vehicular service solutions.

Abstract: A system and method of identifying relevant service terms within service records includes: receiving service terms included in one or more service records at computer processing equipment; classifying the service terms into a group of likely relevant service terms and a group of likely irrelevant service terms using the computer processing equipment; and identifying the relevant service terms from the group of likely relevant service terms and ignoring the likely irrelevant service terms using the computer processing equipment.

Abstract: A system and method of modifying a vehicle service database includes: accessing a database containing previously-received symptom text that has been associated with a vehicle identifier and one or more vehicular service solutions for the previously-received symptom text; determining a statistical likelihood that one or more additional vehicular service solutions apply to previously-received symptom text based on a correlation between the previously-received symptom text and additional vehicular service solutions; determining that the statistical likelihood is above a predetermined threshold; and associating the previously-received symptom text with the additional vehicular service solutions.

Abstract: A method allows for testing software under test (SUT) with respect to a partial design model (PDM) having a boundary which differs from a boundary of the SUT. The method includes recording input information including the SUT, the PDM, and coverage criteria defining a required number of the test cases. Variables in the SUT are identified that correspond to boundary signals for the PDM. Test cases are extracted meeting the coverage criteria. The method may include generating additional test cases at the PDM level and mapping the additional cases with corresponding constraint functions to the boundary of the SUT using a forward/backward propagation and/or heuristics guided technique. A system for testing the SUT includes a host machine and memory. The host machine executes process instructions from memory to identify variables in the SUT that correspond to boundary signals for the PDM, and extracts test cases meeting the coverage criteria.