SkyLogix - Autonomous AI Drone Delivery Agent

Web-based control panel for placing delivery orders and tracking autonomous drone missions in real time
Urban environment with live buildings, delivery hubs, charging station, and dynamic obstacles
User onboarding interface for selecting the delivery location within the smart-city grid

Inspiration

The rapid growth of e-commerce and on-demand services is pushing urban logistics to its limits. Traditional delivery systems are not designed to operate in dynamic environments where obstacles move, weather changes in real time, and energy constraints affect mission safety.

I wanted to explore a key question:

Can a drone behave like an intelligent agent that perceives its surroundings, makes decisions on its own, and safely completes deliveries without human control?

Instead of building a static simulation, we designed a system that demonstrates real AI behaviour — planning, adapting, and acting autonomously in a smart-city environment.

This vision led to the creation of SkyLogix – Autonomous AI Drone Delivery Agent.

What it does

SkyLogix is a goal-driven autonomous AI agent operating inside a real-time 3D urban environment.

It can:

📦 Accept delivery tasks through a web interface or manual controls
🧭 Plan optimal routes using A* pathfinding
🐦 Avoid dynamic aerial obstacles in real time
🔋 Monitor battery levels and trigger emergency return
🌧 Analyse weather conditions and cancel unsafe missions
🔁 Replan routes when the environment changes
🏠 Return to the charging station automatically
📊 Prioritise multiple deliveries using intelligent queue optimisation

This creates a fully automated, end-to-end autonomous logistics workflow.

How we built it

Tech Stack

Python → core logic & AI behaviour
VPython → real-time 3D smart-city simulation
Flask → web-based delivery interface

AI Path Planning

I implemented the A* search algorithm:

$$ f(n) = g(n) + h(n) $$

Where:

g(n) → actual travel cost from the start node
h(n) → heuristic (Euclidean distance to the goal)

This allows the drone to compute optimal and collision-free paths in real time.

Autonomous Agent System

The drone continuously:

Perceives the environment
Evaluates constraints
Makes decisions
Executes actions

This forms a complete AI agent loop.

Dynamic Smart-City Simulation

I simulated:

🏢 Buildings & delivery hubs
🐦 Moving aerial obstacles
🌦 Weather system (clear, rain, wind, storm)
🌙 Day/night cycle
⚡ Charging station

The drone adapts its behaviour in real time based on these conditions.

Challenges we ran into

Real-time path replanning

Handling dynamic obstacles while keeping the drone motion smooth required efficient collision detection, an optimised search space, and smart fallback routing.

Autonomous workflow design

Each mission had multiple stages — pickup, delivery, return, and charging. Designing a reliable state machine for this was complex.

Performance vs realism

Continuous AI computation inside a dense 3D environment required careful optimisation.

Accomplishments that I am proud of

Built a complete goal-based autonomous AI agent
Implemented real-time A* path planning with dynamic rerouting
Created a fully functional smart-city logistics simulation
Designed an end-to-end automated delivery workflow
Integrated a web-based order system
Implemented weather-aware mission control
Developed a battery-based emergency return system

Most importantly:

🚁 The drone is not manually controlled — it makes its own decisions and completes deliveries autonomously.

🧠 How the Project Uses AI

SkyLogix functions as a goal-driven autonomous AI agent that can plan, decide, and act on its own in a dynamic environment.

Intelligent Path Planning: Uses the A* heuristic search algorithm to compute optimal, collision-free delivery routes.
Autonomous Decision Making: Continuously evaluates battery level, weather conditions, obstacles, and delivery goals to choose the safest action (deliver, reroute, return, or charge).
Dynamic Path Replanning: Generates new routes in real time when obstacles or environmental conditions change.
Task Optimisation: Prioritises multiple deliveries using intelligent queue logic.

This combines classical AI planning, real-time reasoning, and autonomous agent architecture, similar to how real-world delivery drones and mobile robots operate.