Distributed Control of Heterogeneous Robotic System for Area Exploration

本研究的主要目标是开发一个分布式控制系统，用于多个异构机器人的自主操作，以在不确定环境中协同执行任务。异构机器人包括具有不同传感器和控制能力的地面移动机器人和微型飞行器。本研究开发的功能可以扩展到各种工业应用，例如监视、勘探和绘制人类无法进入的区域（例如废弃矿山或危险区域）、地下采矿、搜索和救援，以及涉及多机器人协调的其他任务。该研究专门开发了解决联合地面空中小组的分布式定位、勘探和控制体系结构的方法。MAV正逐渐成为室内监视和搜索操作的首选。MAV的高度自由度允许在杂乱的室内环境中导航。相对于地面上的替代品，它可以在多层建筑中以3D视角快速搜索。MAV有限的有效载荷引入了与飞行时间、传感器和设备处理能力相关的限制，削弱了其定位能力和整体任务的有效性。另一方面，地面机器人有各种尺寸，它们有能力提供更快的处理、远程传感和高度精确的定位信息。因此，探索任务正逐步向多智能体实现方向发展，以利用相关能力。该建议包括几个主要的研究组成部分；一个更高层次的控制体系结构，用于空中和地面机器人的分布式控制、使用视觉注意力的编队控制和跟踪、不同车辆的共享和相互定位、MAV姿态估计和轨迹控制，以及最后用于定位和映射（SLAM）的传感器融合和数据分析。

The main objective of this research is to develop a distributed control system for autonomous operations of multiple heterogeneous robots to collaboratively perform a task/tasks in an uncertain environment. The heterogeneous robots include both ground mobile robots and micro-aerial vehicles (MAV) having different sensor and control capabilities. The capabilities developed in this research can be extended for a variety of industrial applications, such as surveillance, exploration and mapping of areas that are not accessible for humans (e.g. abandoned mines or hazardous areas), underground mining, search and rescue, and also for other tasks involving multi-robotic coordination. The research specifically develops methods to address distributed localization, exploration and control architecture of combined ground aerial teams. MAV`s are progressively becoming the preferred choice for indoor surveillance and search operations. The high degrees of freedom of the MAV allows navigation in cluttered indoor settings. It can maneuver through multi floor buildings with fast searches in a 3D perspective, relative to its ground based alternatives. The limited payload of MAV`s introduce constraints related to flight time, sensors and processing power of the devices, crippling its localization ability and effectiveness of the overall mission. On the other hand the ground robots are available with a range of sizes and they have the ability to provide faster processing, long range sensing and highly accurate localization information. Therefore, exploration missions are progressively moving towards multi agent implementations to harness the relative capabilities. The proposal includes several main research components; a higher level control architecture for distributed control of aerial and ground robots, formation control and tracking using visual attention, shared and mutual localization of different vehicles, MAV pose estimate and trajectory control, and finally sensor fusion and data analysis for localization and mapping (SLAM).