Deep Learning on Multi Sensor Data for Counter UAV Applications-A Systematic Review.

Area of Science:

• Computer Science
• Artificial Intelligence
• Robotics

Background:

• Unmanned Aerial Vehicles (UAVs) are increasingly used in consumer applications, leading to potential misuse by malicious actors.
• Protecting critical infrastructure and events from UAV threats necessitates advanced counter UAV (c-UAV) applications.
• Current c-UAV systems utilize multi-sensor fusion (electro-optical, thermal, radar, etc.) for threat identification, but real-time surveillance remains challenging.

Purpose of the Study:

• To provide a comprehensive overview of deep learning technologies applied to c-UAV tasks.
• To examine deep learning advancements in processing multi-sensor data for UAV detection and classification.
• To explore the integration of multi-sensor information fusion within deep learning frameworks for c-UAV applications.

Main Methods:

• Review of existing literature on deep learning methodologies for object detection, classification, and tracking.
• Analysis of deep learning applications on diverse sensor data (electro-optical, thermal, radar, RF).
• Investigation of multi-sensor information fusion techniques enhanced by deep learning for c-UAV scenarios.

Main Results:

• Deep learning shows significant promise in addressing complex c-UAV tasks like object detection and classification.
• Application of deep learning to multi-sensor data fusion enhances the accuracy and confidence of threat identification.
• Novel deep learning approaches are emerging for UAV detection and classification, offering improved performance.

Conclusions:

• Deep learning is a novel and powerful tool for advancing c-UAV capabilities.
• Further research in deep learning for multi-sensor data fusion is crucial for robust c-UAV systems.
• This survey aims to guide future improvements and recommendations for c-UAV applications.