Cambridge Mobile Telematics Inc.

Methods and systems for detecting vehicle crash events using real-time sensor data are disclosed. In some embodiments, an application executing on a computing device disposed in a vehicle receives a stream of time-series data from one or more motion sensors during a drive. For each contiguous interval of a first predefined length, the application generates a crash prediction indicative of whether a crash occurred during that interval. Generating the crash prediction includes executing a crash detection neural network on sensor data collected over a time period that comprises the interval and has a second, greater predefined length. For each crash prediction, the application determines whether the prediction meets one or more predefined threshold criteria. In response to meeting the criteria, the application generates a crash analysis using the corresponding sensor data.

Methods and systems for providing alerts for unsafe driving activities are disclosed. In some embodiments, an application executing on a mobile device receives a plurality of geographic coordinates corresponding to points of interest within a geographic region. During a drive, the application receives measurements from vehicle sensors and determines vehicle locations in real time. The application generates a subset of coordinates within a predefined proximity to the vehicle at a first time during the drive, detects when the vehicle's deceleration rate exceeds a threshold associated with hard braking events, and calculates distances between the event location and each of the subset of coordinates. A shortest distance is compared to a threshold alerting distance, and the application outputs a hard braking alert in response to determining that the nearest point of interest is outside of the threshold alerting distance.

A device positionable in a passenger compartment of a vehicle can include a processor, an accelerometer, and a wireless transmitter. The processor can receive, while in a hibernation mode, an interrupt signal from the accelerometer based on the accelerometer detecting an acceleration of the vehicle that exceeds a predefined acceleration threshold. In response to receiving the interrupt signal from the accelerometer, the processor can switch from the hibernation mode to a sleep mode. While in the sleep mode, the processor can detect a trip start event based on one or more accelerometer values from the accelerometer. In response to detecting the trip start event, the processor can then switch from the sleep mode to an active mode. While in the active mode, the processor can receive additional accelerometer values from the accelerometer and store them for subsequent transmission to a remote server system using the wireless transmitter.

A method including receiving, from sensor(s) of a mobile device in a driven vehicle, current speed of the vehicle, current travel direction of the vehicle, and current location of the vehicle with respect to a road network. The method including receiving contextual data indicating a plurality of triggering conditions located along road segment(s) present in the road network. The method including identifying a road segment of interest based on the current location and current travel direction of the vehicle. The method including identifying a triggering condition in a vicinity of the road segment of interest. The method including generating an alert related to the triggering condition, wherein generating the alert includes: determining an alert modality for the alert and a set of alert characteristics for the alert modality. The method including outputting the alert with the alert modality and set of alert characteristics to the user using the mobile device.

A computer-implemented method for determining personalized driving risk includes collecting, as collected data, location data and driving data associated with a one or more trips associated with a driver. Based on the collected data, one or more risky behaviors of the driver is identified. The collected data is combined with mapping data associated with the location data. Based on the combined data, driver behavior data associated with the driver is determined. Behavioral information associated with the driver and driving suggestion information is provided on a computer display of a mobile computing device.

Techniques for detecting, based at least in part on a first image obtained at a first time, a first indication of a first object included in the first image obtained using a vehicle sensor. The techniques can further include generating, based at least in part on the first indication, a first feature space representation of the first object included in the first image. The techniques can further include comparing the first feature space representation of the first object with a set of stored feature space representations. Responsive to the comparing, the techniques can further include assigning a first unique identifier to the first feature space representation; generating, based on the first indication, a first distance of the first object from the vehicle sensor; and associating the first distance, the first unique identifier, and the first time in memory.

Methods and systems for reconstructing a route traveled by a vehicle using motion sensor data are provided. In some embodiments, a method comprises receiving motion measurements from a plurality of sensors in a vehicle during a drive. An estimated sequence of turns and distances traveled before and after each turn is extracted from the motion measurements. A start location and an end location of the drive are determined. Candidate routes between the start and end locations are selected from historical routes, each comprising a sequence of turns and distances along road segments within a road network. The candidate routes are compared with the estimated sequence to determine mismatch scores. Based on the mismatch scores, the route traveled by the vehicle is identified. An interactive graphical user interface on a mobile device receives a request to display the drive on a map and displays the identified route in response.

A method includes executing a data collection application over a time interval, and requesting, by the data collection application, movements measured by a sensor arrangement of a mobile phone from an operating system of the mobile phone. The movements are measured before execution of the data collection application. The method also includes receiving, by the data collection application and from the operating system, a set of movements and transmitting, by the data collection application, the movements to a server during the time interval.

Techniques for using current and historical driving data to detect crashes are provided. In some examples, a crash detection application executing on a computing device receives movement measurements indicating motion of the device during a drive in a vehicle from a sensor arrangement coupled to the device and the application analyzes the movement measurements to detect motion of the vehicle. The application receives crash detection criteria from a crash detection management server system that were generated using a historical driving performance of the vehicle, of a driver associated with the computing device, or both. By applying the crash detection criteria to the motion of the vehicle, the application determines that the vehicle was involved in a crash.

Techniques for optimizing mobile device resource consumption while detecting driving activities are provided. In examples, an application executing on a mobile device determines that the device is disposed in a vehicle during a drive and receives movement measurements indicating motion of the device caused by a user or the vehicle from a sensor arrangement of the mobile device. The movement measurements are analyzed to detect an occurrence of a driving event during the drive and a notification of the driving event is generated. Audio routes available to the operating system of the device are received by the application and a determination is made that the user is driving the vehicle based on the available audio routes. In response, the notification is provided to the user.