Mathematical Optimization of Resolution Improvement in Structured Light data by Periodic Scanning Motion: Application for Feedback during Lunar Landing

Tarek A. Elsharhawy, Shuo Liu, Luc SaikaliarXiv 2024 · DOI · arXiv

Introduction

Landing a spacecraft on the lunar surface is a formidable challenge, particularly when navigating scientifically significant but hazardous terrain. Current obstacle avoidance systems largely depend on LiDAR. While effective, LiDAR can suffer from resolution and precision limitations in low-light or high-noise environments, such as deep craters or during the lunar night. This paper proposes a novel "super-vision" system for lunar landers by integrating Structured Illumination Microscopy (SIM) with Iterative Learning Control (ILC). This approach aims to surpass the physical resolution ceilings of conventional Structured Light Systems (SLS), providing high-precision terrain mapping without requiring prior knowledge of the landing site.

Methodology

The proposed system fuses three distinct technological domains to enhance visual feedback for autonomous landing:

  1. Structured Illumination Microscopy (SIM): Traditionally a biological imaging technique, SIM uses moiré interference patterns to recover high-frequency spatial information that typically falls outside the diffraction limit of standard optics.
  2. Structured Light Systems (SLS): This serves as the hardware foundation, projecting known patterns onto the lunar surface to derive depth and topographical data.
  3. Iterative Learning Control (ILC): Acting as the system's optimization engine, ILC refines the periodic scanning motion of the SLS hardware. By learning from previous iterations, it minimizes errors induced by the harsh, high-vibration environment of space.

To evaluate this hybrid approach, the researchers developed SIMworldmodel17, a MATLAB-based simulation environment. This tool allows for the interactive tuning of parameters—such as illumination angles, Gaussian smoothing (sigma), and Wiener filtering—to optimize the reconstruction of high-resolution images from simulated lunar surface data.

Original simulated lunar surface image used as the baseline for the SIMworldmodel17 processing.

Results and Discussion

Preliminary results generated via the SIMworldmodel17 interface demonstrate significant improvements in image quality. By fine-tuning the number of illumination angles and applying advanced filtering, the system achieves substantial resolution gains over conventional SLS.

The SIMworldmodel17 GUI displays the original lunar surface image alongside the Fourier spectrum comparison, the high-resolution reconstructed SIM image, and quantitative improvement metrics.

As shown in the simulation dashboard, the processed "High-Resolution SIM Image" exhibits sharper features compared to the "Original Image." The Fourier Spectrum Comparison confirms that the processed image retains significantly more high-frequency detail. Quantitative data indicates a theoretical improvement factor of 2.00, reducing the resolution from 0.96 µm to 0.48 µm. Furthermore, edge detection comparisons highlight the system's ability to maintain structural integrity in noisy conditions.

Key insights from this work include:

  • Cross-Domain Innovation: Applying microscopic imaging techniques to macroscopic spacecraft sensors effectively breaks traditional resolution barriers.
  • Hardware Resilience: Utilizing ILC allows the system to maintain peak performance despite environmental interference or hardware imperfections.
  • Multi-Modal Synergy: The enhanced SLS is intended to complement LiDAR, filling resolution gaps to create a more robust navigation suite.

Conclusion and Future Work

This research establishes a foundational framework for high-resolution, zero-knowledge lunar landing. While simulation results are promising, future work will focus on transitioning from MATLAB models to physical experiments that replicate the chaotic conditions of lunar descent, including extreme lighting shifts and dust interference. Additionally, addressing the computational overhead of running these algorithms on flight-hardened processors will be critical for achieving true real-time feedback during landing.

Summary updated

BibTeX
@misc{elsharhawy2024mathematical,
  author = {Tarek A. Elsharhawy and Shuo Liu and Luc Saikali},
  title = {{Mathematical Optimization of Resolution Improvement in Structured Light data by Periodic Scanning Motion: Application for Feedback during Lunar Landing}},
  year = {2024},
  eprint = {2408.06628},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2408.06628},
}
Figure