Fast and secure distributed multi-agent coverage control with an application to infrastructure inspection and reconstruction

In this paper, we aim to present a sophisticated solution to real industrial applications using a multi-agent system. More specifically, we aim to develop a fast and secure distributed coverage control algorithm to enable a multi-agent system to cooperatively inspect and reconstruct infrastructure. The coverage control for mobile sensor networks puts forward a requirement of improving efficiency and guaranteed safety in unknown and obstacle-rich environments. Our algorithm focuses on exploring partially known environments with limited sensing capabilities, while simultaneously avoiding collisions with obstacles and other agents. To collect necessary data, a density map is reactively updated to measure the distribution of viewpoints. Based on this map, each agent will be dynamically driven to the location of interest. A collision-free region is then constructed in real-time for each agent to guarantee safety during the coverage process. Finally, a spatial load-balanced deployment is generated to provide locally optimal sensor coverage of target infrastructure. The overall system with a team of 3 unmanned aerial vehicles (UAVs) has been successfully adopted to physically inspect a large-scale industrial building in Shatin, Hong Kong. The result proves that our proposed method not only guarantees flight safety but also efficiently completes overall tasks in cluttered environments. It greatly reduces the required operating time and thus minimizes potential hazards in using unmanned aerial systems for inspection.