Osprey: Multi-Session Autonomous Aerial Mapping with LiDAR-based SLAM and Next Best View Planning

TL;DR

Osprey is presented, an autonomous aerial mapping system with state-of-the-art multisession light detection and ranging (LiDAR)-based mapping capabilities that enables a nonexpert operator to specify a bounded target area that the aerial platform can then map autonomously over multiple flights.

Abstract

Aerial mapping systems are important for many surveying applications (e.g., industrial inspection or agricultural monitoring). Aerial platforms that can fly GPS-guided preplanned missions semi-autonomously are already widely available but fully autonomous systems can significantly improve efficiency. Autonomously mapping complex 3D structures requires a system that performs online mapping and mission planning. This paper presents Osprey, an autonomous aerial mapping system with state-of-the-art multi-session LiDAR-based mapping capabilities. It enables a non-expert operator to specify a bounded target area that the aerial platform can then map autonomously over multiple flights. Field experiments with Osprey demonstrate that this system can achieve greater map coverage of large industrial sites than manual surveys with a pilot-flown aerial platform or a terrestrial laser scanner (TLS). Three sites, with a total ground coverage of $2528$ m$^2$ and a maximum height of $27$ m, were mapped in separate missions using $112$ minutes of autonomous flight time. True colour maps were created from images captured by Osprey using pointcloud and NeRF reconstruction methods. These maps provide useful data for structural inspection tasks.

Figures (15)

• Figure 1: The Osprey autonomous aerial mapping system surveying a multi-building industrial site.
• Figure 2: The Osprey aerial platform, a DJI M600 drone with a Frontier and Rajant WiFi module mounted underneath it. The Frontier consists of (a) a Hesai QT64 LiDAR, (b) a SevenSense Core Research sensor module with an IMU and three 1.6 megapixel colour fisheye cameras, and (c) an Intel NUC computer. Communication with the operator is provided by a wireless mesh network comprised of (d) a Rajant BreadCrumb ES1 module on Osprey and (e) Rajant BreadCrumb DX2 modules distributed around a target site.
• Figure 3: Flowchart showing an overview of the five key components of the Osprey mapping pipeline: Sensor Payload, Odometry, Mapping, Mission Planning and Motion Planning. The rectangular boxes represent sensor inputs or algorithms and the connections between them denote data pipelines. The diamonds represent boolean decisions, with green arrows denoting true outcomes and red arrows denoting false outcomes.
• Figure 4: An illustration of the factor graph, showing the states, $x_i$, and the factors connecting them from the (yellow) and registration (pink) modules.
• Figure 5: An illustration of a VILENS-SLAM pose graph created from a multi-session mapping mission. Nodes associated with the first session are denoted as $p_i$ and those from a subsequent session are denoted as $q_i$. Odometry factors from VILENS (orange) connect consecutive nodes in the same session. Geometric loop closure factors (light blue) can exist between nodes in the same or different sessions, but only an inter-session loop closure is shown for simplicity. A relocalisation factor (purple) is added between nodes in different sessions after ScanContext identifies a successful match.