∂LITE: Differentiable Lighting-Informed Trajectory Evaluation for On-Orbit Inspection

Jack Naylor1*, Raghav Mishra1,2, Nicholas H. Barbara1, Donald G. Dansereau1,2
1Australian Centre for Robotics and School of Aerospace, Mechanical and Mechatronic Engineering, The University of Sydney
2Australian Robotic Inspection and Asset Management (ARIAM) Hub, The University of Sydney
*Corresponding Author
Paper Slides

∂LITE: Differentiable Lighting-Informed Trajectory Evaluation

An end-to-end differentiable simulation pipeline for on-orbit inspection operations to optimise inspection trajectories for image quality.

∂LITE Overview

System Overview

∂LITE is an end-to-end differentiable simulator for on-orbit inspection which produces non-obvious trajectories based on visual costs. We optimise orbital parameters for passive inspection trajectories that yield superior quality images of a target satellite by minimising specular reflections seen by the sensor.

Motivation

This video of Astroscale's ADRAS-J Mission demonstrates the key visual challenges of inspecting objects on-orbit. The dynamic and intense illumination of the sun saturates images captured by the camera, significantly reducing information such as texture and shape which is relied upon in downstream tasks including 3D reconstruction, change detection and repair planning. ∂LITE optimises trajectories which avoid these direct reflections, significantly increasing the quality of captured imagery.

∂LITE Pipeline

Pipeline Architecture

An overview of the ∂LITE pipeline. Given a TLE for the chaser satellite and a simulation epoch, we compute a time-varying sun vector and satellite position at time t. The scene parameters are updated to produce rendered inspection imagery. From these observations we formulate a visual cost capturing the specularity in the scene to backpropagate gradients through the pipeline.

Trajectory Optimisation Results

Comparisons showing RGB and specular cost visualizations of conventional inspection trajectories (before) compared to our optimised trajectories (after).

Experimental Results

SGP4 Validation

SGP4 Implementation Validation

Comparison of our differentiable SGP4 propagator with existing software, using the Hubble Space Telescope as an example. We achieve similar results, with the benefit of our implementation being fully differentiable via JAX. Similar numerical results were observed for CloudSat and Sentinel-6.

Cost Optimization

Cost Function Optimization

Total cost, specular cost, and distance cost of the orbital parameters over optimisation iterations for the Hubble Inspection. Our optimisation reaches a local minimum yielding substantially reduced specular cost throughout an orbit.

Pixel Saturation Reduction

Image Quality Improvement

Percentage of saturated pixels during Hubble inspection. ∂LITE substantially reduces pixel saturation by minimizing direct sunlight reflections, improving overall image quality.

Specular Cost Comparison

Specular Reflection Reduction

Qualitative comparison of maximum specular costs observed through an orbit. By optimising orbital parameters to observe the target satellite from different locations and angles, we dramatically reduce specular costs observed through an orbit.

Orbit Validation

Trajectory Visualisation

The orbit of the chaser satellite relative to the target satellite (black dot) in the Hubble inspection experiment. The black circle represents the initialised orbit and red curve represents the optimised orbit. The section of the orbit which the parameters were optimised for (t ∈ [0, T ]) are solid, while the rest of the orbit is dashed.

Abstract

Visual inspection of space-borne assets is of increasing interest to spacecraft operators looking to plan maintenance, characterise damage, and extend the life of high-value satellites in orbit. The environment of Low Earth Orbit (LEO) presents unique challenges when planning inspection operations that maximise visibility, information, and data quality.

Specular reflection of sunrays from spacecraft bodies, self-shadowing, and dynamic lighting in LEO significantly impact the quality of data captured throughout an orbit. This is exacerbated by the relative motion between spacecraft, which typically introduces variable imaging distances and attitudes during inspection. Planning inspection trajectories via simulation is a common approach. However, the ability to design and optimise an inspection trajectory specifically for the resulting quality of imaging data in proximity operations remains largely unexplored.

In this work, we present ∂LITE, an end-to-end differentiable simulation pipeline for on-orbit inspection operations. We leverage state-of-the-art differentiable rendering tools and a custom orbit propagator to enable end-to-end optimisation of orbital parameters based on visual measurements. ∂LITE enables us to optimise for non-obvious trajectories, vastly improving the quality and usefulness of attained data. To our knowledge, our differentiable inspection-planning pipeline is the first of its kind and provides new insights into modern computational approaches to spacecraft mission planning.

BibTeX


@article{naylor2025dlite,
  author = {Naylor, Jack and Mishra, Raghav and Barbara, Nicholas and Dansereau, Donald G.},
  title = {$\partial${LITE}: Differentiable Lighting-Informed Trajectory Evaluation for On-Orbit Inspection},
  journal = {IAF 76th International Astronautical Congress (IAC) Space Operations Symposium},
  organization = {International Astronautical Federation},
  year = {2025},
}