We present A.I.M, a end-to-end mapping application built using a C++ backend, GTK GUI, and a LLM powered LangChain Agent for Semantic Search and Navigation. A.I.M is built on the concept that most modern mapping applications do not consolidate all required information in one place and also lack proper semantic search and navigation capabilities, hindering their usability and purpose. A.I.M uses state of the art LLMs powered with LangChain for large context windows and fast information retrieval, so that users will never have to hop across applications to find necessary information needed to make their navigation easier. Besides this, A.I.M also incorporates several special features, based on research for user needs and current map capabilities, such as Live Weather, Light/Dark Mode, refined search capabilities for Points of Interest (POIs), and a Zoom Scale.
Check out our final product below:
A.I.M supports both typed search and drag-and-drop routing for quick path creation and editing. You can select points on the map, reorder stops, and obtain detailed navigation directions on the application.
Our LLM chat agent allows users to further refine POI search capabilities for semantic information retrieval, while also planning efficient navigation routes between them for the user if needed.
We used state-of-the-art graph traversal algorithms such as A and Multi-threaded Djikshtra* for fast and efficient route planning. Our algorithm placed Top 6 on ECE297 Leaderboard with 70+ teams. To learn more about it, check here.
Shoutout to my teammates: Luthira Abeykoon (@Luthiraa) and Grace Hao (@gracehao15).