Lane-change detection

点击翻译按钮获取中文摘要

Since the end of 2013, Intact Insurance has offered usage-based insurance programs designed to leverage**data-driven technologies and improve insurance products, by collecting driving data from customers using**either a GPS dongle connected to the vehicle's on-board computer or a smartphone application. In this context,**vehicle position and other sensor data, such as acceleration, wheel rotation, and magnetic azimuth, are recorded**when available. This data is used to characterize driving behavior and, in particular, to detect risky events such**as speeding, harsh breaking, hard acceleration, sharp turns or abrupt lane changes. Along with other contextual**information such as historical risk of a given road, road conditions, other cars' behaviour and speed limits,**these data have the potential to provide better customer segmentation along with risk and cost prediction,**which in turn could help Intact Insurance provide the most competitive pricing to their customers. In order to**build customer-specific road safety indicators and risk factors based on i) where and when the client drives**(contextual information), and ii) how the client drives (driving profile), complex automated procedures must be**developed. The main objective of this research is to tackle issues related to driver's behaviour profiling through**the development and validation of a measure for lane change detection.**Specifically, we propose in this project to investigate how the use of accelerometers and gyroscopes data**collected through the Intact smartphone applications "my Driving Discount" or "automerit" can improve lane**change detection. We also propose to evaluate the minimum data frequency required to achieve a reasonable**detection.