Automotive Visual-inertial Navigation

点击翻译按钮获取中文摘要

An autonomous vehicle (AV) must use a suite of sensors, such as cameras and inertial sensors, to not only localize itself in the world, but also the positions of other vehicles, pedestrians, cyclists, and obstacles. The goal of the proposed research is to fuse measurements from various sensors, such as an inertial measurements unit (consisting of a rate gyroscope and an accelerometer), wheel odometer, as well as a monocular camera to design a simultaneous localization and mapping (SLAM) solution for AVs using automotive-grade hardware. The navigation solution must be able to estimate the position of static landmarks, such as road barriers and sign posts, and dynamic landmarks, such as moving vehicles, relative to the host vehicle. The navigation solution designed will be used as an input to a path-planning algorithm to plan paths for the AV around other vehicles and obstacles. As such, the development of a robust and reliable navigation solution is critical to the safety of the AV. The deployment of AVs is quickly becoming a reality in Canada and will have an enormous impact on the development of our economy, environment, and society. It is estimated that the cumulative potential benefits of AVs in Canada totals $65 billion dollars per year (expressed in 2013 Canadian dollars), including collision avoidance, fuel cost savings, and congestion avoidance. The proposed research will contribute to a rapidly expanding network of AV research in Canada. By contributing to research regarding the navigation solution for AVs, it is expected that the outcome of this research will place Canada closer to deploying AVs in Canadian cities.