Tight Fusion of Odometry, Kinematic Constraints, and UWB Ranging Systems for State Estimation of Integrated Aerial Platforms

Integrated Aerial Platforms (IAPs), consisting of multiple interconnected aircraft, offer a promising framework for aerial manipulation tasks by enhancing localization accuracy and reliability. Unlike aerial swarms, the interconnected nature of IAPs allows for exploiting physical constraints among aircraft to improve positioning and navigation systems. In this paper, we present an advanced decentralized multi-aircraft visual-inertial-range-physical odometry system that considers the position, velocity, and attitude constraints inherent to IAPs. By tightly fusing visual-inertial-range odometry and Ultra-Wideband (UWB) with kinematic constraints, we optimize odometry accuracy through the use of novel constraint-based methods. Our algorithm’s performance is validated on simulated datasets and self-collected datasets, with flight experiments conducted on the IAP, demonstrating a significant improvement with a 28.7% reduction in drift over the baseline on real-world datasets and a 24.5% reduction on simulation datasets. Note to Practitioners—Motivated by the need for accurate localization in Integrated Aerial Platforms (IAPs), this paper introduces a novel approach to multi-aircraft odometry technology through the application of physical constraints. We propose a method that combines the physical constraints of IAPs with ranging data to achieve rapid and efficient unification of coordinate systems and precise estimation of anchor positions. We then develop a theoretical framework to analyze the position, velocity, and attitude constraints induced by the multiple aircraft within the IAP and introduce a decentralized optimization method to enhance odometry accuracy. This framework is referred to as Visual-Inertial-Range-Physical Odometry (VIRPO). Furthermore, using multiple self-collected datasets for IAPs, we conduct a thorough evaluation of the proposed VIRPO algorithm, comparing its performance with baseline methods. Test results show that VIRPO significantly improves localization accuracy. Finally, we implement the IAP platform and integrate the VIRPO algorithm to conduct flight experiments, verifying its effectiveness in real-world flight conditions. This marks the first time the VIRPO algorithm has been run onboard an actual IAP platform. In future research, we plan to explore more complex IAP configurations and additional types of constraints to further enhance the localization performance of IAPs.