Goal: Bridging the gap between emergency responders and monitoring sensor systems.
Long-Term Aspiration: Develop an EdgeAI application using Axelera hardware that operates locally to consolidate information from multiple monitoring systems and creates a local LLM “Avatar” that the end-user can interact with in order to “query” the emergency.
Brief Description for the Current Challenge: Simulate a set of sensors generating data → run monitoring algorithms → upon an emergency trigger (manual or monitoring alarm) consolidate key data → launch a session with a locally hosted llm having the summary data as context and an optimized system prompt → run preliminary evaluation tests on the usefulness of the Avatar.
Note: We might want to calibrate the Avatar for operation in a voice interfacing format in addition to the chat interface to allow both office personnel but also frontline-mobile workers to interact with the Avatar (potentially en-route to the emergency).
P.S. Hello everyone! Just found out about the community and the challenge, great to e-meet you all!
