Versatile Multipurpose Crashproof UAV: Machine Learning and IoT approach

Abstract

Technological advancements in the drone sector posed an arduous challenge of enhancing the collisiontolerance competence of an Unmanned Aerial Vehicle(UAV). This paper presents the design and system integration of an IoT-enabled UAV which comprises a 3D designed and printed spherical frame wound across the UAV consisting of a high definition camera that demonstrates processing of the video feed captured using machine learning algorithm. A truncated icosahedron shaped protective frame is designed such that it can bounce and roll in the near proximity of objects as well as humans, thereby proving to be crash-resistant. Hence, it offers close scrutinization, surveying and inspection of various structures and analyses them using a machine learning model. Another novel feature of this UAV is the sensor module composed of various detachable sensors used for applicationspecific purposes like gas-leakage detection, air-quality monitoring, temperature, humidity etc. in confined and complex environments. These features of the UAV, on collaborating with various indoor and outdoor applications contribute towards the versatility of this drone. The UAV is integrated with LoRa modules and is used for seamless connectivity and networking over astonishingly great distances. The final prototype of this design was successfully flight tested numerous times and was found to be efficient, robust and stable.