Abstract
The design and implementation of an Artificial Intelligence (AI)- based, auto-adaptive, modular drone for surveillance applications is presented in this paper. The novelty of the proposed drone includes interchangeable modules (thermal imaging, night vision, and gas sensors) that can be swapped within 30 seconds to support real-time adaptability. Further, the payload configuration is optimized based on environmental conditions using an AI-driven module selection algorithm, improving surveillance efficiency by 27%. A hybrid energy management system extends the flight time by 22% (up to 110–120 minutes), while the AI-based flight controller improves stability, reducing drift by 15%. Real-time anomaly detection is achieved with 98.5% accuracy using an edge processing unit. The designed system supports hot-swappable payloads and intelligent power optimization, allowing fast adaptation in different test cases or environments, unlike traditional ones. The proposed drone system is a combination of edge AI, modular design and hybrid energy architecture, making it highly suitable for real-world tasks like disaster response, urban monitoring and industrial safety. Test results show how the drone quickly adapts and performs tasks efficiently in areas such as urban security and disaster response, highlighting it as a scalable and cost-effective solution for surveillance applications.