Interactive Drone Control with Gesture and Posture on the Edge Computing Platform

This paper designs a system that combines sensor and multiple neural network models to achieve the function of interactive drones, such as user tracking, gesture and posture control of the drone, and emergency safety protection. To run on the edge computing platform, Jetson Nano, and control the drone, the neural network models such as object detection, face recognition, and gesture recognition were optimized. The drone could be controlled by the user's gestures and postures successfully.

I. Introduction

Unmanned Air Vehicles(UAVs), also known as drones, have been attracting wide attention in recent decades. Drones are widely used in many applications, such as military use, agriculture, aerial photography, Disaster Management, wildlife monitoring, geographic mapping, rescue, and entertainment [1]. The rapidly increased use of drones also brings safety issues to individuals and society. Due to the difficulty of flying drones, especially for inexperienced beginners, there are often mistakes in control because of inexperience or excessive nervousness. Sometimes not only causes damage to the drone but also endangers personal safety and property.