Abstract: A mobile device detects a crash event using one or more sensors of a mobile device. The mobile device records a first set of data from the one or more sensors of the mobile device. The mobile device generates a first feature vector including the first set of data and available values for one or more additional data types. The mobile device executes a first machine-learning model selected from a plurality of machine-learning models based on the one or more additional data types for which there are available values to generate a first confidence of a total loss event.

Abstract: A mobile device detects a crash event using one or more sensors of a mobile device. The mobile device records a first set of data from the one or more sensors of the mobile device. The mobile device generates a first feature vector including the first set of data and vehicle data that includes an identifier of a vehicle. The mobile device generates a second feature vector using the first set of data and additional data types. The mobile device predicts a confidence of a total loss event by generating a first confidence value from a first machine-learning model using the first feature vector and a second confidence value from a second machine-learning model using the second feature vector.

Abstract: Techniques described herein provide for methods and systems for vehicle crash prediction using multi-vehicle data. In one example, a method includes receiving event data for events identified by devices disposed in a plurality of vehicles from movements of the vehicles collected by the devices. For each pair of events, a degree of alignment is determined and used to identify candidate pairs of events. For each candidate pair of events, single event detection features are extracted and input into a crash prediction model to determine a crash correlation probability for whether both events correspond to a same crash event. Based on the probability, an indication of the same crash event is transmitted to a remote server associated with either or both vehicles involved in the crash event.

Abstract: Techniques are disclosed for detecting a point-of-impact of a vehicle collision. A mobile device receives a set of sensor measurements from sensors of the mobile device. The mobile device generates a crash feature vector from the sensor measurements and one or more point-of-impact features. The mobile device executes a set of classifiers using the crash feature vector and the one or more point-of-impact features. Each classifier of the set of classifiers is configured to generate an output that is partially indicative of a point-of-impact of the collision. The mobile device generates, from the output of each classifier of the set of classifiers, a prediction of a point of impact of the vehicle collision. The mobile device then transmits an indication of the point of impact to a remote device.

Abstract: Techniques are disclosed for detecting a point-of-impact of a vehicle collision. A mobile device receives a set of sensor measurements from sensors of the mobile device. The mobile device generates a crash feature vector from the sensor measurements and one or more point-of-impact features. The mobile device executes a set of classifiers using the crash feature vector and the one or more point-of-impact features. Each classifier of the set of classifiers is configured to generate an output that is partially indicative of a point-of-impact of the collision. The mobile device generates, from the output of each classifier of the set of classifiers, a prediction of a point of impact of the vehicle collision. The mobile device then transmits an indication of the point of impact to a remote device.

Abstract: Techniques described herein provide for methods and systems for vehicle crash prediction using multi-vehicle data. In one example, a method includes receiving event data for events identified by devices disposed in a plurality of vehicles from movements of the vehicles collected by the devices. For each pair of events, a degree of alignment is determined and used to identify candidate pairs of events. For each candidate pair of events, single event detection features are extracted and input into a crash prediction model to determine a crash correlation probability for whether both events correspond to a same crash event. Based on the probability, an indication of the same crash event is transmitted to a remote server associated with either or both vehicles involved in the crash event.

Abstract: A method includes receiving, from a first mobile device associated with a first vehicle, first data related to a first event involving the first vehicle, receiving, from a second mobile device associated with a second vehicle, second data related to a second event involving the second vehicle, and determining, based on a comparison between the first data and the second data, a degree of alignment between the first event and the second event. The method also includes determining, based on the degree of alignment between the first event and the second event, that the first event and the second event correspond to a same crash event involving the first vehicle and the second vehicle. The method further includes performing one or more actions based on determining that the first event and the second event correspond to the same crash event.

Abstract: Techniques are disclosed for detecting a point-of-impact of a vehicle collision. A mobile device receives a set of sensor measurements from sensors of the mobile device. The mobile device generates a crash feature vector from the sensor measurements and one or more point-of-impact features. The mobile device executes a set of classifiers using the crash feature vector and the one or more point-of-impact features. Each classifier of the set of classifiers is configured to generate an output that is partially indicative of a point-of-impact of the collision. The mobile device generates, from the output of each classifier of the set of classifiers, a prediction of a point of impact of the vehicle collision. The mobile device then transmits an indication of the point of impact to a remote device.

Abstract: Techniques are disclosed for detecting a point-of-impact of a vehicle collision. A mobile device receives a set of sensor measurements from sensors of the mobile device. The mobile device generates a crash feature vector from the sensor measurements and one or more point-of-impact features. The mobile device executes a set of classifiers using the crash feature vector and the one or more point-of-impact features. Each classifier of the set of classifiers is configured to generate an output that is partially indicative of a point-of-impact of the collision. The mobile device generates, from the output of each classifier of the set of classifiers, a prediction of a point of impact of the vehicle collision. The mobile device then transmits an indication of the point of impact to a remote device.

Abstract: A method includes receiving, from a first mobile device associated with a first vehicle, first data related to a first event involving the first vehicle, receiving, from a second mobile device associated with a second vehicle, second data related to a second event involving the second vehicle, and determining, based on a comparison between the first data and the second data, a degree of alignment between the first event and the second event. The method also includes determining, based on the degree of alignment between the first event and the second event, that the first event and the second event correspond to a same crash event involving the first vehicle and the second vehicle. The method further includes performing one or more actions based on determining that the first event and the second event correspond to the same crash event.

Abstract: A method includes receiving, from a first mobile device associated with a first vehicle, first data related to a first event involving the first vehicle and receiving, from a second mobile device associated with a second vehicle, second data related to a second event involving the second vehicle. The method also includes determining, based on the first data and the second data, whether the first event and the second event correspond to a same crash event and performing one or more actions based on whether the first event and the second event correspond to the same crash event.

Abstract: Techniques are disclosed for predicting a confidence of a total loss event. A mobile device detects a crash event using one or more sensors of a mobile device. The mobile device records a first set of data from the one or more sensors of the mobile device. The mobile device generates a first feature vector including the first set of data and vehicle data that includes an identifier of a vehicle. The mobile device generates a second feature vector using the first set of data and additional data types. The mobile device predicts a confidence of a total loss event by generating a first confidence value from a first machine-learning model using the first feature vector and a second confidence value from a second machine-learning model using the second feature vector.