AI Explorations
Google Maps Competition
This project, for my AI Explorations class, was to utilize AI and machine learning in order to complete one of several tasks such as navigation, solving Sudoku, or a task of one's choosing. For this, my team and I chose to work on navigational based AI in the area around the university. But just doing navigation wasn't simple enough and so the scope was increased to create comparisons with leading navigational tools such as Google maps and Waze in several key areas such as distance to destination, time of arrival, accuracy, and speed of determination. To do this, the area around the university was converted into a node map, seen below, with data attached to the nodes and links for the major roads included within the map.
Using several tree search-based algorithms (like A*, Dijkstra's, depth first, and breadth first) navigational data was taken from the algorithms and compared to results obtained from Google Maps and Waze. Surprisingly, other than a few small deviations on one particular route, the algorithms performed nearly as well as both navigational apps producing nearly identical routes and travel times. After comparing results, some research was performed in order to determine where differences in results became present. Most of the differences occurred due to construction or road closures that were not taken into account by the static node map created by the team.