UAS Platforms for Autonomous Navigation

Introduction & Objective

This research focuses on the development of Unmanned Aircraft Systems (UAS) platforms for autonomous operation, particularly in challenging environments where traditional GPS-based navigation is unreliable. The objective was to design, build, and test custom hardware and software solutions for robust positioning, navigation, and mapping.

• Goal: To achieve reliable waypoint navigation in GNSS-denied or challenged conditions.
• Approach: Integrating novel positioning solutions and developing custom algorithms for perception and localization.

Key Contributions

UAS Platform Development: Built multi-rotor UAS platforms and tuned the flight controllers for stable and efficient data collection. This involved calibrating sensors and optimizing PID loops to ensure precise flight characteristics required for research.

Custom Hardware & Sensor Integration: Designed custom hardware to seamlessly integrate various sensors (e.g., LiDAR, cameras) with lightweight computing platforms like Raspberry Pi and Odroid, enabling on-board processing and data acquisition.

Alternative Positioning Solutions: Collaborated with NextNav to utilize their proprietary urban and indoor positioning solution on multi-rotor UAS platforms, providing an effective alternative to GPS for waypoint navigation in challenging urban environments.

Advanced Algorithmic Implementation: Implemented a plane estimation algorithm using Singular Value Decomposition (SVD) for rapid and accurate plane extraction from point cloud data, a critical step for localization and mapping algorithms.

Methodology

Dataset Collection: Collected extensive UAV datasets to analyze GNSS positioning under various GNSS-denied and GNSS-challenged conditions. This included flights in urban canyons, under dense tree cover, and indoors.

SLAM Implementation: Adopted and customized open-source SLAM (Simultaneous Localization and Mapping) algorithms for the datasets collected by UAVs, focusing on point cloud-based and visual SLAM techniques.

Ground Truth Generation: Utilized a terrestrial LiDAR scanner to obtain high-precision ground truth point clouds. These were crucial for evaluating and validating the accuracy of the developed localization and mapping algorithms.

Results & Findings

Based on the analysis, significant improvements in positioning accuracy were observed using the NextNav solution compared to traditional GPS in urban settings. The implemented SVD-based algorithm also demonstrated high computational efficiency for real-time plane detection.

Conclusion

This research successfully demonstrated the feasibility and effectiveness of using custom-built UAS platforms and alternative positioning systems for navigation in GNSS-denied environments. The work provides a strong foundation for future advancements in autonomous drone operations, especially for applications like search and rescue, infrastructure inspection, and delivery in complex urban landscapes.