Initialization of Monocular Visual Navigation for Autonomous Agents Using Modified Structure from Small Motion

TL;DR

This method, a state-of-the-art factor graph optimization pipeline, extends Structure from Small Motion to robustly initialize a monocular agent in spacecraft inspection trajectories, addressing visual estimation challenges such as weak-perspective projection and center-pointing motion.

Abstract

We propose a standalone monocular visual Simultaneous Localization and Mapping (vSLAM) initialization pipeline for autonomous space robots. Our method, a state-of-the-art factor graph optimization pipeline, extends Structure from Small Motion (SfSM) to robustly initialize a monocular agent in spacecraft inspection trajectories, addressing visual estimation challenges such as weak-perspective projection and center-pointing motion, which exacerbates the bas-relief ambiguity, dominant planar geometry, which causes motion estimation degeneracies in classical Structure from Motion, and dynamic illumination conditions, which reduce the survivability of visual information. We validate our approach on realistic, simulated satellite inspection image sequences with a tumbling spacecraft and demonstrate the method's effectiveness over existing monocular initialization procedures.

Figures (7)

• Figure 1: Landmark parameterization using inverse depth $w$, azimuth $\psi$, and elevation $\phi$.
• Figure 2: Realistic synthetic images of the Hubble Space Telescope produced in our Unreal Engine 5 simulator. As the camera and RSO move, the specular and diffuse reflections and moving shadows simulate the realistic, changing, and harsh illumination conditions encountered in space.
• Figure 3: Proposed
• Figure 4: Ha et al.
• Figure 6: Error values for scale-normalized translations, rotations, and scale-normalized depths for the proposed method (blue) and the USAC_FM_8PTS algorithm Raguram2013 (red) on trajectories spanning from 0 to 6 degrees of parallax.