Visual breadcrumbs for emergency return of unmanned aerial vehicles

无人机（UAV）技术在过去十年中发展迅速，使自动化，按需遥感和交付成为近期现实，应用范围从精准农业到应急响应行动以及侦察和监视。一个尚未开发的关键构建块是一个解决方案，该解决方案能够实现安全的超视距飞行，并在发生通信和/或GPS信号丢失等故障时防止车辆失控。该提案的主要目标是开发一种基于视觉的导航解决方案，使无人机在失去通信和定位手段时能够安全返回家园（即，GPS）。 基于视觉的导航技术将使无人机能够确定其相对于飞行期间记录的或飞行前获得的航空地图的位置。多伦多大学（UofT）在过去的十年中一直在开发地面车辆的视觉路线跟踪技术，并在自主飞行的控制算法设计方面获得了丰富的经验。该项目将需要进行重大扩展，包括使地面视觉导航框架适应空中应用，以及开发一种基于视觉的飞行控制器，以保证可靠的定位和安全着陆。该项目将与位于加拿大的新兴世界级固定翼无人机供应商PrecisionHawk以及加拿大国防研究与发展公司（DRDC）合作开展，该公司将旋翼无人机用于各种国防应用（例如，城市复杂区域的监测）。基于视觉的导航技术的这一飞跃将使UofT在无人机安全、长期自主领域取得核心科学进步，并使PrecisionHawk和DRDC能够将这些进步融入其未来的产品和解决方案中，从而显著提高无人机的安全性及其应用范围。此外，本文提出的安全解决方案将有助于加快无人机在超线应用中的商业用途的监管框架的发展。

Unmanned Aerial Vehicle (UAV) technology has advanced at a rapid pace in the last decade, making automated, on-demand remote sensing and delivery a near-term reality with applications ranging from precision agriculture to emergency response operations and reconnaissance & surveillance. A key building block that is yet to be developed is a solution that enables safe beyond-line-of-sight flight, and prevents uncontrolled vehicle behavior in the event of failures such as communication and/or GPS signal loss. The main goal of this proposal is to develop a vision-based navigation solution that enables UAVs to safely return home when they lose their means of communication and localization (i.e., GPS). Vision-based navigation techniques will enable the UAV to determine its location relative to an aerial map recorded during flight or obtained prior to flight. The University of Toronto (UofT) has been developing visual route-following techniques for ground vehicles for the past decade and also gained extensive experience in control algorithm design for autonomous flight. This project will require major extensions including the adaptation of the ground-based visual navigation framework to aerial applications and the development of a vision-based flight controller that guarantees robust localization and safe landing. The project will be carried out in partnership with PrecisionHawk, an emerging world-class provider of fixed-wing UAVs located in Canada, as well as Defence Research & Development Canada (DRDC), which uses rotor-based UAVs for a variety of defence applications (e.g., surveillance of complex urban areas). This leap forwards in vision-based navigation technology will allow UofT to make core scientific advances in the area of safe, long-term autonomy of UAVs and will allow PrecisionHawk and DRDC to incorporate these advances in their future products and solutions, thereby significantly increasing the safety of UAVs and their range of applications. Moreover, safety solutions as proposed here will help to accelerate the development of a regulatory framework for the commercial use of UAVs in beyond-line-of-side applications.