Integrated Satellite-Navigation for Challenging Environments

定位和导航信息正变得无处不在，因为汽车、移动电话甚至衣服都配备了导航系统/传感器。大多数导航系统至少部分依赖于全球导航卫星系统（GNSS）技术，如美国全球定位系统（GPS）。不幸的是，到达地球的GNSS信号功率极低，在卫星信号衰减或被树叶和建筑物等遮挡的地区，系统性能会下降。在其他情况下，特别是在城市地区，卫星信号在被接收之前被反射，这限制了接收器测量到卫星的直接（理想）距离的能力。在这两种情况下，GNSS接收器都无法可靠地跟踪卫星信号。这个问题可以通过两种方式解决：提高GNSS接收机的能力，以及将GNSS与其他传感器集成。近年来，低成本传感器的数量和可用性急剧增加，特别是在智能手机和汽车上，包括WiFi收音机（以及它们可以访问的在线数据）、相机、光传感器、加速度计、陀螺仪和磁力计。然而，挑战在于确定如何最好地将这些传感器与GNSS集成，以产生更准确、更可靠的定位解决方案，特别是在GNSS性能不佳的地区。考虑到这一点，该项目将使用不同的方法改善退化环境中基于GNSS的导航：1)通过使用来自所有卫星的所有信息来计算位置的新架构来增强当前GNSS接收器的性能。这与传统架构形成对比，传统架构只使用来自每个卫星的数据子集来计算（次优）位置。结果将在更多的位置提供更精确的定位解决方案，从而实现更多的应用。2)结合越来越广泛使用的3D建筑模型（例如b谷歌Earth）。GNSS接收器的先进信号处理方法将与3D建筑模型数据相结合，以识别和纠正到达接收器的反射信号。通过这种方式，而不是仅仅试图最小化反射信号的误差（传统上是这样做的），这种方法以建设性的方式使用反射信号来提高性能。3)使用一个朝上的摄像头，例如智能手机正面的摄像头，计算位置信息，并将这些信息与GNSS合并。这项研究将利用朝上的摄像头设置（相对于通常用于导航的朝前摄像头）的好处来改善基于视觉的导航。此外，城市峡谷中的天际线变化——即在特定位置的建筑物之间可见的天空轮廓——将首次被用作确定用户绝对位置的手段。项目的后期阶段将合并这些方法，以便在更广泛的环境中产生最佳性能。作为该项目一部分开发的算法和软件将有利于GNSS接收器制造商和将GNSS与上述传感器集成在一起的导航系统开发商。预计在2014年至2020年期间，gnss支持的市场价值将超过1万亿美元（加元），这项工作将帮助加拿大在这一快速增长的领域保持领先地位。更重要的是，本研究培养的学生将具备相关的知识和技能，在毕业后为航运业作出积极的贡献。

Positioning and navigation information is becoming ubiquitous as automobiles, mobile telephones and even clothing are being equipped with navigation systems/sensors. Most navigation systems rely, at least in part, on Global Navigation Satellite System (GNSS) technology such as the U.S. Global Positioning System (GPS). Unfortunately, GNSS signals that reach the Earth have extremely low power, and system performance degrades in areas where satellite signals are attenuated or obstructed by such things as foliage and buildings. In other scenarios, particularly in urban areas, the satellite signals are reflected prior to being received which limits a receiver’s ability to measure the direct (ideal) distance to the satellite. In both cases, the GNSS receiver cannot track the satellite signals reliably, if at all. This problem can be tackled in two ways: by improving the capability of GNSS receivers, and by integrating GNSS with other sensors. The number and availability of low-cost sensors has increased dramatically in recent years, especially in smart phones and automobiles, and include WiFi radios (and the online data they can access), cameras, light sensors, accelerometers, gyroscopes and magnetometers. The challenge, however, is to determine how to best integrate these sensors with GNSS in order to yield a more accurate and reliable position solution, especially in areas where GNSS is not performing optimally. With this in mind, this project will improve GNSS-based navigation in degraded environments using different approaches by: 1) Enhancing current GNSS receiver performance by using a new architecture where all information from all satellites is used to compute the position. This contrasts with traditional architectures where only a subset of data from each satellite is used to compute a (sub-optimal) position. The result will be more accurate position solutions in more locations, thus enabling more applications. 2) Incorporating 3D building models that are becoming more widely available (e.g., Google Earth). Advanced signal processing methods with GNSS receivers will be integrated with 3D building model data to identify and correct reflected signals that reach the receiver. In this way, instead of only trying to minimize the error from the reflected signal (as is traditionally done), this approach uses the reflected signal in a constructive manner to improve performance. 3) Using an upward-facing camera, such as those found on the front of smart phones, to compute position information and merge this information with GNSS. The research will exploit benefits of an upward-facing camera setup (relative to forward-facing cameras which are more typically used for navigation) to improve vision-based navigation. Also, for the first time, the skyline variability in urban canyons – that is, the outline of the sky that is visible between buildings at a particular location – will be used as a means to determine the absolute position of the user. Later stages of the project will merge these approaches in order to yield optimal performance in a wider range of environments. The algorithms and software developed as part of this project will be of benefit to GNSS receiver manufacturers and to navigation system developers that integrate GNSS with the sensors listed above. With the GNSS-enabled market expected to be worth over $1T (CAD) between 2014 and 2020 this work will help Canada remain a leader in this rapidly growing field. More importantly, the students trained during this research will possess the relevant knowledge and skill to contribute positively to the navigation industry upon graduation.