Second-Order Coverage Control for Multi-Agent UAV Photogrammetry

TL;DR

This paper introduces a multi-agent control strategy for UAVs to efficiently capture images for 3D photogrammetry, improving the quality and scalability of 3D mapping compared to traditional methods.

Contribution

It presents a novel distributed second-order coverage control method that optimizes multi-UAV image capture for photogrammetry, focusing on sharing feature-rich areas.

Findings

The approach produces higher quality 3D reconstructions.

It demonstrates improved efficiency over traditional single-vehicle methods.

Hardware experiments validate the effectiveness of the control strategy.

Abstract

Unmanned Aerial Vehicles equipped with cameras can be used to automate image capture for generating 3D models via photogrammetry. Current methods rely on a single vehicle to capture images sequentially, or use pre-planned and heuristic imaging configurations. We seek to provide a multi-agent control approach to capturing the images required to 3D map a region. A photogrammetry cost function is formulated that captures the importance of sharing feature-dense areas across multiple images for successful photogrammetry reconstruction. A distributed second-order coverage controller is used to minimise this cost and move agents to an imaging configuration. This approach prioritises high quality images that are simultaneously captured, leading to efficient and scalable 3D mapping of a region. We demonstrate our approach with a hardware experiment, generating and comparing 3D reconstructions from image sets captured using our approach to those captured using traditional methods.