PDF version of this entire document
Exploring the problem of car navigation and in particular implementation thereof depends on tools that already exist. Many commonly used algorithms are properly implemented and are well documented. Having surveyed what is available for mobile platforms, I found about 3-4 distinct packages/libraries, most of them proprietary (non-libre) and non-gratis. Advice from veterans in the area (computer vision for Android and iOS) led me to the BSD-licensed OpenCV, which - although poorly documented for particular platforms - remains one of the popular choices, as evidenced by the thriving development community rallying around it. OpenCV boasts many classes of relevance to my application, enabling rapid development and rich visualisation. A company known as Willow Garage [2] has brought OpenCV [1] to Android, a Linux- and Dalvik-based operating system maintained and rapidly developed almost exclusively by Google (through a consortium called Open Handset Alliance, or OHA for short).
