ESA GNC Conference Papers Repository

Real-time Rasterization Simulation of Thermal Infrared images for spacecraft applications
Deren Vural, Iain M. Martin, Martin Dunstan, Manual Sanchez-Gestido
Presented at:
Sopot 2023
Full paper:

As Thermal/Infrared (TIR) sensors have become more advanced, they are being considered for Guidance, Navigation and Control (GNC) systems, in addition to the existing roster of sensors. A crucial step in the development of GNC systems is the testing of sensors, including analysis methods that utilise the sensor data. To expedite this process VIS image generation software like the Planet and Asteroid Natural Generation Utility (PANGU) [1] and Airbus's SurRender [2] are used to generate synthetic images representative of the kind of data that would be gathered. For spacecraft this was previously limited to the VIS spectral range and LiDAR, but as a result of the development in TIR sensors, methods of generating representative synthetic TIR images of spacecraft are now required to aid in development and testing. PANGU v6 added TIR simulations of planetary surfaces and spacecraft in 2021 [3]. This paper will summarize and review problems involved in generating representative TIR images of spacecraft using traditional rendering techniques and proposes a method of real-time image generation to simulate time-based effects. Traditional thermal analysis is performed using Lumped Parameter (LP) and Finite Element Analysis (FEA) techniques that utilize standard heat transfer equations tailored for spacecraft in custom-made software tools like ESATAN-TMS. Modern engineering CAD software like SolidWorks can simulate thermal systems using modern FEA techniques, though few take into account the orbital nature of spacecraft and the resulting changes in thermal behaviour. For this reason, ESATAN's orbital module ensures it stands as one of the few tools available for directly simulating spacecraft thermal behaviour. To generate TIR images, we can adapt techniques from traditional thermal analysis software to more flexible rendering tools like PANGU, which generates representative images using efficient GPU-based rendering techniques over computationally expensive techniques like FEA. A stateless model was developed for PANGU v6.0, using traditional thermal analysis equations to compute TIR radiance in each simulated image [3]. However, there remain some discrepancies in the thermal profile, and images, generated using this technique [3], possibly because it functions without any consideration of changes that occur over more than a single frame. To more accurately simulate the thermal profile of real-world spacecraft, time and its effects on the thermal behaviour of the simulated spacecraft should be considered. While radiative and conductive properties define the ability of a material to transfer thermal energy, the thermal capacitance (a.k.a. inertia) of an object describes the rate at which it can do so. To assist in the approximation of the capacitance of a mesh CAD model, which describes the boundary surfaces of an object rather than defining the solid object itself, we can adapt traditional thermal analysis techniques. One possible method is to create a thermal mathematical model (TMM) [4] limited to the surface thermal properties, that could be used to simulate the approximate flow (conduction) of heat across the surface using node capacitance. The creation of a TMM of the surface of a model requires mesh stitching, i.e. joining together the meshes that make up a CAD model so that a point, which may be described multiple times in order to describe multiple polygons (e.g. the corner of a cube), may be represented as a single node. While not as accurate as describing the conduction of heat through a solid object, the approximation of surface conduction has the potential to increase the accuracy of generated images where conduction plays a significant effect, i.e. due to a rapid change of incoming radiation e.g. entry and exit of eclipse. This paper reports on the development of a new GPU-based renderer that aims to include the thermal behaviour of spacecraft over time in TIR image generation, in order to improve the representative quality of the images. Two techniques are being developed in the renderer: the first is a capacitance model that involves the creation of a TMM and the assignment of thermal properties to each node, to keep track of the temperature of each. The TMM created will be constructed using a mesh "stitching" technique, that creates a geometrically unique set of nodes that more accurately depicts physical properties. The second utilises the TMM to simulate conductive effects over edge nodes, calculating the conduction from one polygon to another. At each frame (i.e. time steps) the temperature and then radiance of each node is calculated using its thermal properties (i.e. emissivity, absorptivity and conductivity) and the previous temperature. The radiance is calculated in the GPU fragment shaders; ensuring that the temperatures passed to the GPU from the vertex shader are properly interpolated, allowing gradients across surfaces with differing node properties. The full paper will examine the quality of the prototype renderering system described above, and examine how valid and representative the generated images are using a traditional thermal simulation tool, ESATAN-TMS. Using this, we can create equivalent CAD models and use them to generate an overall thermal profile of a simple spacecraft in a known orbit. We can then compare this thermal profile to the surface temperature profile generated using the prototype developed, in order to evaluate how representative the images generated are compared to the stateless model currently in use within PANGU. [1] Martin, I., Dunstan, M. Parkes, S., Sanchez-Gestido, M. and Ortega, G., "Simulating planetary approach and landing to test and verify autonomous navigation and guidance systems", in: 10th International ESA Conference on Guidance, Navigation & Control Systems (ESA GNC) (May 29 -- June 02, 2017), 2017, p.15 [2] Brochard, R., Lebreton, J., Robin, C., Kanani, K., Jonniaux, G., Masson, A., Despré, N. and Berjaoui, A., "Scientific image rendering for space scenes with the SurRender software", in: arXiv preprint, 2018, DOI: 10.48550/arXiv.1810.01423 [3] Martin, I., Dunstan, M., Vural, D. and Sanchez-Gestido, M, "Simulating Thermal Infrared Images of Planets, Asteroids and Spacecraft", in: 8th International Conference on Astrodynamics Tools and Techniques (ICATT) (June 23--25, 2021), Paper 259, Virtual, 2021, p.14 [4] Savage, C.J., "Spacecraft Systems Engineering", in: John Wiley & Sons Ltd., 2003, ch.11: Thermal Control of Spacecraft, pp.355-391