Accurate online road maps are required in the modern world for many purposes, such as location-based mobile services, disaster relief, and autonomous vehicles. However, making these maps is a task primarily done by humans, and is therefore slow and tedious. To speed up and automate the map making, we will create a system which takes an existing road map and a set of GPS trajectories and suggests new roads to the user based on the trajectories that are not covered in the map. We will adapt existing algorithms such as k-means clustering and RoadRunner for the purpose of updating maps, rather than inferring from scratch. The system will be tested on maps in the United State and in Qatar. We aim that the system would infer most of the roads, avoid inferring any extra roads, and have proper road connections.

“Through SuperUROP, I want to get deep into research and learn skills such as writing papers and giving presentations. Both will prepare me for graduate school and industry. I have taken courses in software development and algorithms, and I am therefore prepared to work on this project of online street maps. I feel excited about implementing algorithms that can detect accurate information about the road network with the given noisy data.”