Simulated images from the IDEFIX rover WheelCams

Introduction

The IDEFIX rover has been integrated into the Martian Moons eXploration (MMX) mission spacecraft and is awaiting launch in October 2026. Upon arrival at the Martian moons, the rover will be deployed to the surface of Phobos and will perform the first ever technical demonstration of wheeled mobility on a small Solar System body [1]. The rover is equipped with four scientific instruments for characterizing the surface properties of Phobos: RAX, a Raman spectrometer, the NavCams, a stereo pair of cameras, miniRAD a radiometer, and the WheelCams. The WheelCams, with a pixel resolution of approximately 100 µm, will provide direct images of the surface and will capture the dynamic interaction between the rover wheels and regolith [2]. These cameras will allow for a unique, in-situ characterization of the surface material on Phobos. Still images will provide information regarding the size and morphology of the grains, the bulk mechanical properties of the regolith, and the possible layering within the shallow sub-surface of Phobos [2]. Dynamic images will be used to assess the clumping, shearing, and flow behavior of the regolith and the overall thrust and performance of the rover. The cameras can be used as a stand-alone instrument during nominal driving operations or be paired with the rover NavCams to augment the scientific output from other unplanned, but possible mobility tests, such as pivoting, driving in reverse, extending/retracting the rover legs, or excavating the regolith with a single wheel.

In anticipation of the in-situ rover operations on Phobos, we have developed a preliminary pipeline for simulating and analyzing the eventual images from the IDEFIX WheelCams. We produce representative WheelCam images using 1) optical simulations that combine the 3D rover CAD with the ANSYS programs Zemax and SPEOS and 2) experiments with the ISAE-SUPAERO WheelCam testbed [3].  Currently, the optical simulations are used to estimate the camera field of views and the quality of the images for various rover configurations, and the experimental images are used to characterize surface grain morphology [4,5], trench morphology [6,7], wheel sinkage, and wheel velocity. Together, the simulated and experimental images will provide a robust framework to support the IDEFIX operations and scientific analysis on Phobos.

Simulated images

The IDEFIX rover includes two WheelCams, which are mounted at different angles within the rover chassis. The front WheelCam is tilted at an angle of approximately 42 degrees and points at the undisturbed surface ahead of the front rover wheel. The rear WheelCam is tilted by approximately 45 degrees and shows the trench that is created by the front wheel. When the rover is in the nominal driving configuration, the front and back wheels are within the field of view of each respective camera.

Fig. 1: 3D rover CAD showing the fields of view for the front WheelCam (red) and the rear WheelCam (green)

We use the 3D rover CAD (Fig. 1), in combination with ANSYS OpticStudio to simulate the fields of view, resolution, and focus of the cameras for different rover configurations (e.g., when the rover is high vs. low sinkage configurations or when the rover legs are articulate to some off-nominal rotation angle). The simulations will be used to correct the perspective of the WheelCam images and to assist with operation planning for the WheelCam instrument.

Experimental images

In addition to performing optical simulations, we generate experimental images using the WheelCam testbed at ISAE-SUPAERO [3]. The testbed includes flight-spare LEDs, provided by CNES, and a representative rover wheel, provided by DLR. The testbed cameras are positioned to replicate the fields of view from the nominal driving configuration, and the experiments are performed with comparable exposure times and luminosity conditions as the actual WheelCams. Fig. 2 shows representative images from the front (top row) and rear (bottom row) testbed cameras for experiments with various granular materials.

Fig. 2: Representative views from the IDEFIX rover WheelCams for experiments using (a, d) gravel, (b, e) coarse sand, and (c, f) a Phobos regolith simulant [8]. The images for the front and rear testbed cameras are shown in the top and bottom rows, respectively. The rear testbed images have been rotated by 90 deg.

The image processing pipeline consists of algorithms that characterize both the wheel and surface properties. We begin by correcting the testbed images for distortion and then track points around the rim of the wheel to construct a homography matrix to transform the images based on camera perspective. The object detection is also used to determine the rotational velocity of the wheel, and the transformed images are used to calculate wheel sinkage. Static images are used to characterize the morphology of the grains around the wheel [4,5] (Fig. 3), while a series of images are used to determine local surface topography and trench morphology [6,7].

Fig. 3: Identified grains for the trench morphology analysis from WheelCam testbed experiments in (a) gravel, (b) Lunar simulant, (c) Phobos simulant, and (d) coarse sand [5].

As part of future work, the pipeline will be extended to characterize regolith flow around the wheel, identify and track the motion of clumps, determine rover forward velocity, and extract trench features such as the angle of repose. Together, these observables will be linked to the physical properties of the Phobos surface, providing a robust foundation for the in-situ data analysis and ultimately contributing to the broader MMX science objectives regarding the origin and evolution of the Martian moons.

Acknowledgements

This work was funded by CNES in the context of the MMX IDEFIX rover and the WheelCams.

References

[1] Ulamec, S., et al. Acta Astronautica 210 (2023): 95-101.

[2] Murdoch, N., et al. Progress in Earth and Planetary Science 12.1 (2025): 54.

[3] Sunday, C., et al. Review of Scientific Instruments. In preparation (2026).

[4] Robin, C. Q., et al. Nature communications 15.1 (2024): 6203.

[5] Robin, C. Q. Institut Supérieur de l’Aéronautique et de l’Espace. PhD Thesis (2026).

[6] Amsili, A., et al. EPSC2024-407. Copernicus Meetings (2024).

[7] Amsili, A., et al. Advances in Space Research. Submitted (2026).

[8] Miyamoto, H., et al. Earth, Planets and Space 73.1 (2021): 214.