gitmyhub

NeuralAir

TypeScript ★ 1 updated 2mo ago

A hackathon demo imagining a Solana-coordinated marketplace and dashboard for autonomous delivery drones.

Next.jsReactTypeScriptSolanasetup: moderatecomplexity 3/5

NeuralAir is a hackathon project that imagines Solana, a blockchain network, acting as the coordination layer for a network of autonomous delivery drones. It presents itself as a decentralized aviation network where drone missions like cargo delivery, agriculture, firefighting support, and traffic monitoring can be posted to an open marketplace, and where a fleet of drones, charging stations, and pilots would coordinate through the blockchain rather than one central company.

The app has several main screens. A dashboard shows a live interactive map of drones, their routes, and basic fleet statistics like battery levels and active flights, along with an AI dispatcher that can accept natural language requests, such as asking it to send the nearest drone to a location, and reply with a parsed instruction. A marketplace lets you browse and create drone missions and, on Solana's test network, connect a Phantom wallet to simulate paying into an escrow address. A Sky Charge section models a decentralized network of charging pods for drones, showing live charging sessions and running totals of energy used and money earned. A control page offers a first person, keyboard controlled flight view, and a flight logs page keeps a searchable history of past flights that can be exported as a CSV file.

The README is direct that this is a demo built for hackathon judging rather than a working aviation system: drone movement and much of the mission activity are simulated for a reliable walkthrough, though the wallet connections and interface are real when configured, and the architecture is designed with an eye toward later connecting to real drone hardware using the MAVLink protocol family.

The project is built with Next.js, React, and TypeScript, uses Solana's Devnet test network, and can optionally call the OpenAI API for its natural language dispatcher, falling back to canned replies if no API key is set.

Where it fits