Human-Robot Collaboration in Unmanned Aerial Vehicle River Patrol Application

Abstract

The autonomy of Unmanned Aerial Vehicles (UAVs) has significantly influenced research and market development, leading to innovative applications across various fields. However, conventional joystick control methods are still prevalent, presenting operators with challenges related to complexity, limited maneuverability, and multitasking capabilities. To develop a user-friendly and intuitive UAV control system, this paper proposes a vision-based human-robot collaboration system. In this system, the UAV serves as a tool for environmental perception and recognition, while operators take on roles of supervision and decision-making. This study aims to validate the usability of the human-robot collaboration system in real-world river environments. Consequently, we designed two experimental scenarios: river patrol with personnel detection, and riverbank construction site inspection tasks. The UAV employs the Global Positioning System (GPS) for large-scale localization, augmented by visual positioning and semantic segmentation modules to enhance the river tracking model for river patrol. Meanwhile, operators can monitor and semi-automatically adjust flight parameters in real time through a visualization platform.