A Ground-Based Near Infrared Camera Array System for UAV Auto-Landing in GPS-Denied Environment

Tao Yang, Guangpo Li, Jing Li, Yanning Zhang, Xiaoqiang Zhang, Zhuoyue Zhang, Zhi Li2016 · DOI

TL;DR

This paper presents a novel ground-based near-infrared (NIR) camera array system designed for autonomous UAV landings in GPS-denied environments. By combining a NIR laser marker on the UAV with a wide-baseline NIR camera array, the system achieves long-range detection and high-precision localization, enabling safe and accurate landings even when GPS signals are unavailable.

What Problem Does This Paper Address?

Safe and accurate landing is a critical yet challenging aspect of Unmanned Aerial Vehicle (UAV) operations. Traditional navigation systems, such as GPS, often suffer from signal blockage in certain environments (e.g., urban canyons, mountainous terrain) or can be susceptible to interference, making them unreliable for autonomous landings in GPS-denied scenarios. While vision-based systems offer a promising alternative, existing methods often struggle with limited detection range, susceptibility to varying illumination conditions, and temporary occlusions, especially at longer distances. This paper aims to bridge this gap by developing a robust, long-range vision-based system specifically for autonomous UAV landings in environments where GPS is unavailable.

Methods

The proposed system comprises three key novel components:

  • NIR Laser Lamp and Camera Array Imaging Module: To overcome the limitations of passive vision systems at long distances, a near-infrared (NIR) laser lamp is mounted on the UAV's nose. This acts as a bright, distinct marker, simplifying detection and improving robustness against background clutter and varying lighting. Two NIR cameras with a wide baseline are positioned on either side of the runway to capture images of this laser marker. A NIR interference bandpass filter is used with the cameras to enhance anti-interference capabilities and ensure the system is only sensitive to the laser's wavelength, allowing for all-weather, day-and-night operation.
  • Wide Baseline Outdoor Camera Array Calibration Module: Achieving high localization accuracy requires precise camera parameters. For large-scale outdoor environments, a novel calibration method is proposed. It utilizes a chessboard pattern for intrinsic camera calibration and an electronic total station for extrinsic calibration. The electronic total station, with its high accuracy over long distances, is crucial for establishing precise world coordinates of control points. This method aims to achieve high calibration accuracy even in challenging outdoor settings.
  • Laser Marker Detection and 3D Tracking Module: The system employs a robust detection algorithm to identify the NIR laser marker in the camera images. This is followed by a multi-camera collaborative detection approach using epipolar geometry constraints to associate targets between the two cameras and remove false positives. Subsequently, a projective invariant binocular location method is used to estimate the 3D position of the marker. Finally, a target tracking algorithm, utilizing Euclidean distance and the Hungarian algorithm, maintains the UAV's trajectory and speed estimation.

Datasets

Not discussed in the paper. The experiments were conducted using a middle-sized fixed-wing UAV and real-time flight experiments.

Key Insights and Ideas

  • NIR Laser as a Robust Marker: Using a NIR laser as the primary target marker significantly improves detection range and robustness compared to relying solely on passive visual cues or visible light sources, especially in challenging lighting and weather conditions.
  • Wide Baseline for Long-Range Accuracy: The strategic placement of cameras with a wide baseline is crucial for achieving accurate 3D localization at extended distances, a common challenge in UAV landing systems.
  • Electronic Total Station for Outdoor Calibration: The integration of an electronic total station for extrinsic camera calibration in outdoor environments is a key enabler for achieving the required precision for autonomous landing.
  • All-Weather, Day-and-Night Operation: The combination of NIR imaging and bandpass filters allows the system to function reliably under various environmental conditions, overcoming limitations of visible light-based systems.

Key Results

  • Detection Range: The system demonstrated a detection range of over 1000 meters, significantly exceeding many existing vision-based systems.
  • Localization Accuracy: Experimental results showed that the localization accuracy achieved centimeter-level precision within 200 meters and maintained an error of less than 0.5 meters in X-axis and centimeter-level in Y and Z axes within the last 100 meters of landing.
  • Robustness: The NIR laser marker was robustly detected from 0 to 45 degrees, while a strong light flashlight was only clearly visible up to 10 degrees, highlighting the superiority of the NIR approach.
  • GPS-Denied Landing: The system was successfully validated in a GPS-denied environment, guiding a UAV from 820 meters to a safe landing.

Limitations and Future Work

  • Limitations:
    • The paper acknowledges that the errors in the X-axis (along the runway) are larger than in Y and Z, although they decrease with distance.
  • While the system is robust to lighting and weather, extreme conditions like heavy fog or dense smoke might still pose challenges.
  • The current system is primarily designed for fixed-wing UAVs with high flight speeds and landing heights.
  • Future Work:
    • The paper suggests that the system's ability to guide multiple UAVs simultaneously (multi-object tracking) is an important area for future development.
  • Further exploration of advanced detection and tracking algorithms could enhance performance in even more complex scenarios.
  • Adapting the system for different types of UAVs (e.g., multirotors) could broaden its applicability.

Summary updated

BibTeX
@misc{yang2016ground,
  author = {Tao Yang and Guangpo Li and Jing Li and Yanning Zhang and Xiaoqiang Zhang and Zhuoyue Zhang and Zhi Li},
  title = {{A Ground-Based Near Infrared Camera Array System for UAV Auto-Landing in GPS-Denied Environment}},
  year = {2016},
  doi = {10.3390/s16091393},
}

Loading related content...

Figure