Flying IoT: Sensor Fusion Performance Analysis for UAV Applications in Indoor Spaces

In recent times, the integration of Unmanned Aerial Vehicles (UAVs) with Internet of Things (IoT) platforms has brought a new dimension of mobility, significantly enhancing the capabilities of complex sensor networks. This chapter focuses on the development of an autonomous quadcopter, tailored as a core component of an AI-driven sensing network in indoor settings. We explore the integration of various vision-based and laser sensors with an onboard computer, utilizing the Robot Operating System (ROS) and WiFi technology for real-time wireless data collection and communication. Our system achieves significant milestones in 3D mapping and localization, demonstrating the tracking camera T265 sensor’s superiority in localization and computational efficiency over traditional SLAM algorithms. Through rigorous testing and calibration, including initial flight trials, we highlight the vast potential of these autonomous systems within an IoT framework. Moreover, we delve into the edge computing capabilities of the onboard Jetson Nano processor, optimizing its performance for enhanced functionality. These optimizations led to full utilization of the processor’s capabilities, reducing the mapping time of a 3 m × 3 m area from 4 to 5 min to under 3 min, with improved accuracy and reduced feature drift. Utilizing advanced AI and computer vision techniques, we test and compare SLAM algorithms, further cementing the role of UAVs in revolutionizing IoT applications.