UAV-Based Agentic AI System for Search and Rescue
London Southbank University
Sep 2024 - Current (1 year 10 months)
Developed a prompt-driven AI framework using UAVs for autonomous emergency response. Key achievements:
- Implemented deep reinforcement learning for real-time UAV navigation.
- Integrated YOLO-based object detection models for human recognition.
- Designed a simulation testbed for validating system performance with real-world data.
- Leveraged ChatGPT and Claude to enhance the development process by brainstorming prompt-driven workflows, debugging Python code, simulating agent conversations, and fine-tuning prompts for decision-making tasks. Their integration significantly accelerated development, tested use cases, and validated logic across multi-agent systems, contributing to the overall success of the project.
- Utilized Microsoft E