UAV Agent

From UAV Agents to Agentic UAVs: Pioneering Human-Like Reasoning in Autonomous Aerial Systems

Agentic AI Autonomous UAVs LLM Reasoning Action Execution
View on GitHub

The Agentic UAVs Vision

The Agentic UAVs project introduces a groundbreaking approach to unmanned aerial vehicle autonomy by integrating cutting-edge Agentic Artificial Intelligence technologies and advanced Large Language Models.

Revolutionary Paradigm

Unlike traditional autonomous systems that depend on predefined rules, our UAVs are empowered with human-like reasoning capabilities to perceive, reason, decide, and act independently.

Four-Layer Architecture

Perception, Reasoning, Action, and Learning layers work together to create truly intelligent aerial systems capable of complex decision-making.

Evolution of Intelligence: From Detection to Action

Our journey showcases the progression from traditional object detection to fully autonomous action execution, demonstrating increasing levels of UAV intelligence and autonomy.

Demo 1
UAV Search and Rescue Mission
Smart Detection & Analysis

Foundation: LLM-Augmented Scene Understanding

The inaugural demonstration where UAVs combine YOLO object detection with Large Language Models for intelligent scene interpretation and contextual analysis.

Core Capabilities
  • Real-time object detection
  • LLM-powered scene understanding
  • Contextual analysis and reporting
  • Intelligent alert generation
Technologies
YOLO Detection Visual LLMs GPS Navigation Real-time Analysis
Demo 2 - Advanced
Advanced UAV Agent with Action Execution
Autonomous Reasoning & Action

Evolution: LLM Agent with Action Execution

The advanced demonstration introduces autonomous decision-making and action execution, marking the transition from UAV Agents to truly Agentic UAVs with human-like reasoning capabilities.

Enhanced Capabilities
  • Autonomous rescue assessment
  • Physical rescue kit deployment
  • Real-time authority notifications
  • GPS coordinate transmission
  • Multi-modal action execution
Advanced Technologies
LLM Agents Action Execution Autonomous Decision-Making Multi-Agent Systems Email Integration

Intelligent Workflow: From Perception to Action

1

Autonomous Patrol & Detection

UAV conducts systematic search missions using GPS navigation and real-time object detection to identify persons or objects of interest.

GPS Navigation YOLO Detection Computer Vision
2

LLM-Powered Scene Analysis

Advanced Visual LLMs analyze the detected scene, understand context, assess situation severity, and determine if rescue intervention is needed.

Visual LLMs Contextual Reasoning Risk Assessment
3

Autonomous Decision Making

Demo 2 Innovation: LLM Agent evaluates if detected pilgrim requires rescue assistance and autonomously decides on appropriate intervention level.

LLM Agents Decision Logic Autonomous Reasoning
4

Multi-Modal Action Execution

Demo 2 Breakthrough: Physical rescue kit deployment, real-time email notifications with images and GPS coordinates, and coordinated emergency response activation.

Physical Deployment Email Integration GPS Coordination Emergency Response

Key Innovations & Impact

Human-Like Reasoning

LLM Agents enable UAVs to think, reason, and make complex decisions like human operators, moving beyond simple rule-based automation.

Action Execution

Physical intervention capabilities including rescue kit deployment and real-time coordination with emergency response teams.

Tiered Autonomy

Scalable autonomy levels from operator-assisted to fully autonomous operations, similar to self-driving vehicle frameworks.

Live Demonstrations

Demo 1: Smart Detection & Analysis
LLM-Augmented Search and Rescue UAV Missions
Demo 2: Autonomous Action Execution
From UAV Agents to Agentic UAVs

Research Team

Prof. Anis Koubaa

Prof. Anis Koubaa

Principal Investigator

Agentic AI & UAV Systems Expert

Khaled Jabr

Khaled Gabr

Co-Investigator

Robotics & Autonomous Systems Expert

The Future of Agentic UAVs

This represents the first step towards a new paradigm where UAVs possess human-like reasoning capabilities. Future developments will explore agentic swarm architectures, multi-agent coordination, and deployment across defense, emergency response, and civilian applications.

Swarm Intelligence
Coordinated multi-UAV operations
Defense Applications
Border surveillance & security
Continuous Learning
Adaptive mission improvement
Collaborate With Us GitHub Repository