- The paper introduces an EKF that fuses UWB and IMU data to achieve 80Hz 3D localization with reduced latency.
- It significantly improves positional accuracy, lowering maximum error from 0.71m to 0.39m and mean error from 0.30m to 0.16m.
- The approach offers a practical solution for indoor MAV operations where GPS is unavailable, demonstrating robust performance in dynamic, interference-prone settings.
Accurate 3D Localization for MAV Swarms by UWB and IMU Fusion
This paper investigates an extended Kalman filter (EKF) based algorithm to achieve accurate 3D localization for Micro Aerial Vehicle (MAV) swarms through the fusion of Ultra Wideband (UWB) and Inertial Measurement Unit (IMU) data. The motivation for this paper emerges from the inherent limitations of UWB-only localization systems, which typically suffer from high latency and low bandwidth, making them inadequate for real-time control of MAVs in dynamic environments. The proposed solution demonstrates 80Hz 3D localization with reduced delay and significantly improved positional accuracy.
Methodology
The authors implement an EKF that integrates UWB measurements with IMU data to mitigate the latency and bandwidth constraints associated with UWB-only approaches. The EKF is designed to effectively handle individual distance measurements, allowing for a continuous estimation process that avoids the need for simultaneous multiple sensor readings, a common requirement in traditional TOA-based algorithms.
The standard assumption of constant velocity in a traditional EKF is enhanced by incorporating IMU measurements, which provide acceleration data to account for dynamic movements. The state vector is augmented to include acceleration bias, which is estimated as part of the process, thus addressing the bias instability issues often encountered with low-cost IMUs.
Key Results
The effectiveness of the proposed system is validated through experiments in a VICON-equipped environment and actual deployment at Changi Exhibition Centre. Comparative analysis with a vanilla EKF reveals that the fusion EKF achieves superior accuracy and drastically reduced latency, with maximum and mean positioning errors markedly lower than the vanilla implementation—0.39m and 0.16m, respectively, compared to 0.71m and 0.30m.
Practical Implications and Future Directions
This research highlights several promising aspects of UWB and IMU fusion for 3D localization in MAV swarms. Primarily, it presents a viable solution for indoor environments where GPS is unavailable, and visual-tracking systems like VICON may be too costly or impractical. The algorithm's robust performance in varying electromagnetic conditions during a large-scale public exhibition underscores its application potential in real-world environments with potential for high interference.
Future developments could focus on optimizing the algorithm for further integration with additional sensors to enhance reliability under challenging conditions, such as occlusion or high electromagnetic interference. Additionally, investigating the scalability of the system for larger swarms or more complex formations may open new avenues in autonomous drone operation and coordination tasks.
In summary, this paper contributes to the expanding body of research aimed at refining localization techniques for MAV swarms, offering advancements that contribute to both the theoretical understanding and practical execution of multi-agent aerial systems.