Behavior tree for autonomous drones in smart manufacturing

无人机（UAV）已经在商业和科学上长期成功地应用于许多户外任务：监视，安全，交通，娱乐等。最近，无人机在仓库内操作的焦点在学术界和工业界都有所增加。随着智能自动化趋势的到来，无人机将扮演与自主移动平台、机械手一样的角色，以提高工厂的运营效率。无人机利用空中空间执行必要的任务，如监控、库存管理、检查和维护，以最大限度地提高工厂的生产力。智能工厂中无人机的开发和部署面临三个主要的研究挑战：(1)多无人机系统在信息丰富的网络环境下的自主运行；(2)无人机系统对人员和事故的安全性；(3)无人机系统效率的优化。无人机应该能够根据实际情况改变操作模式并快速处理。当在智能工厂中运行时，系统之间交换的信息量迅速增强。利用这一点使得多无人机系统适合解决这些问题。为了加速智能工厂环境中自主多无人机系统的发展，最大限度地提高生产率，本研究旨在开发一种基于行为树框架的无人机自主系统规划工具。本研究将有助于智能工厂环境下无人机系统控制的课题。在这种环境中，信息量的交换速度很快。这在将无人机系统投入运行时需要考虑到某些优点和缺点。该研究重点解决了3个具体问题，并将有以下预期的科学贡献：(1)智能体（无人机和服务器）系统的规划框架，以自主方式执行任务；(2)将安全元素扩展并应用于多智能体系统的规划工具；(3)为规划框架开发基于安全和自主的强化学习。该项目与华威创新制造和未来材料（特别是工业4.0）的全球研究重点以及WMG的可持续发展战略（包括数字技术，材料和制造以及能源）密切相关。这很适合由主管领导的研究主题，因此，将提高他们在该领域的领导地位。

Unmanned aerial vehicles (UAV) have been successfully applied to many outdoor tasks: surveillance, security, transportation, entertainment, etc. for a long time commercially and scientifically. Recently, the focus of UAVs operating inside warehouses has increased in both academia and industry. With the coming smart automation trend, drones will play the same role as autonomous mobile platforms, manipulators in improving operational productivity in factories. UAVs utilize the aerial space to perform the necessary tasks, such as monitoring, inventory management, inspection and maintenance, to maximize the plant's productivity.There are 3 main research challenges for development and deployment of UAVs in smart factories: (1) The autonomous operation of a a multiple UAV system in an information-rich network environment; (2) The safeness of the UAVs system with respect to people and incidents; (3) The optimization of the efficiency of the UAVs system. The drones should be able to change the operating mode according to real situations and handle them quickly. When operating in a smart factory, the amount of information exchanged between systems is rapidly enhanced. Taking advantage of this makes multi-UAV systems suitable for solving these problems. To accelerate the development of an autonomous multi-drone system in smart factory environments maximising productivity, this research aims to develop a planning tool for autonomous systems of unmanned aerial vehicles based on the behavior tree framework. The proposed research will contribute to the topic of drone system control in a smart factory environment. In this environment, the amount of information is exchanged quickly. This creates certain advantages as well as disadvantages that need to be taken into account when putting the drone system into operation. The study focuses on solving 3 specific problems and will have the following expected scientific contributions: (1) A planning framework for a system of agents (drones and servers) to perform tasks in an autonomous manner; (2) Extending and applying safety elements to the planning tool for a multi-agents system and (3) Developing safety and autonomy-based reinforcement learning for the planning framework.This project is strongly aligned with Warwick's Global Research Priorities on Innovative Manufacturing and Future Materials (especially Industry 4.0) and the sustainable development strategies in WMG, including Digital Technologies, Materials and Manufacturing, and Energy. This fits well within the research themes led by the supervisors, consequently, will boost their leadership in the fields.