Future lunar missions will require autonomous rovers capable of traversing tens of kilometers across challenging terrain while maintaining accurate localization and producing globally consistent maps. However, the absence of global positioning systems, extreme illumination, and low-texture regolith make long-range navigation on the Moon particularly difficult, as visual-inertial odometry pipelines accumulate drift over extended traverses. To address this challenge, we present a stereo visual simultaneous localization and mapping (SLAM) system that integrates learned feature detection and matching with global constraints from digital elevation models (DEMs). Our front-end employs learning-based feature extraction and matching to achieve robustness to illumination extremes and repetitive terrain, while the back-end incorporates DEM-derived height and surface-normal factors into a pose graph, providing absolute surface constraints that mitigate long-term drift. We validate our approach using both simulated lunar traverse data generated in Unreal Engine and real Moon/Mars analog data collected from Mt. Etna. Results demonstrate that DEM anchoring consistently reduces absolute trajectory error compared to baseline SLAM methods, lowering drift in long-range navigation even in repetitive or visually aliased terrain.
Figures & tables
Fig. 1: LightGlue feature matching between left and right stereo images from our Unreal Engine lunar simulation. Features are detected and matched reliably despite challenging illumination conditions. A maximum of 200 matches are plotted to reduce clutter.
Fig. 2: Example data from each dataset. Top: sample rover imagery. Bottom: corresponding digital elevation models (DEMs). From left to right: LuSNAR, S3LI, Unreal Engine simulation.
Fig. 3: Opportunity rover model [ 23 ] used in our custom Unreal Engine simulation.
Fig. 4: Trajectories from LuSNAR scene 9. Ground-truth trajectory is shown in green, estimated VO trajectory in red, and DEM-anchored SLAM in blue. DEM anchoring successfully constrains the SLAM trajectory to the surface and mitigates vertical drift.
Scene
VO
DEM-SLAM
ORB-SLAM3
RPE [m] ↓
ATE* ↓
RPE [m] ↓
ATE* ↓
RPE [m] ↓
ATE* ↓
1
0.0020
0.0039
0.0020
0.0010
0.0046
0.193
2
0.0017
0.0048
0.0017
0.0012
0.0019
0.089
3
0.036
0.0042
0.036
0.0036
0.0020
0.146
4
0.0016
0.0057
0.0016
0.0021
0.0024
0.055
5
0.0021
0.0073
0.0021
0.0023
0.0020
0.044
TABLE I: RMSE comparison of different SLAM methods across nine LuSNAR scenes. Best values are in bold.
Fig. 5: S3LI crater loop with loop closure. Black: ground truth. Blue: before loop closure. Red: after loop closure. The loop closure connection (from start to end) is indicated with a straight red line. Left : loop closure without DEM anchoring leads to significant distortion. Right : DEM anchoring regularizes loop closure, mitigating drift and preventing catastrophic deformation. Residual deviation remains due to limited loop-closure constraints and the fact that DEM height and normal cues alone do not fully constrain tangential motion along visually aliased terrain.
Fig. 6: Long-range simulated traverse in Unreal Engine over an approximately 11 km trajectory. Ground truth is shown in black. Left : visual odometry (VO, orange) accumulates severe global drift over the traverse, deviating significantly from the true path. Right : DEM-anchored SLAM (blue) remains closely aligned with the ground truth but terminates early due to backend resource constraints, resulting in a shorter reconstructed trajectory while still exhibiting substantially reduced drift over the estimated portion.
Perturbation
Level
ATE [m] ↓
Drift [%] ↓
No DEM anchoring
–
0.788
0.986
DEM anchoring
–
0.672
0.717
Vertical noise (corr.)
0.5 m RMS
0.658
0.677
Vertical noise (corr.)
1.0 m RMS
0.894
0.898
Vertical noise (corr.)
2.0 m RMS
0.671
0.536
Vertical bias
+0.5 m
0.680
0.750
TABLE II: DEM uncertainty sensitivity for DEM-anchored SLAM on a representative simulated traverse. Lower is better.
We present a modular, full-stack autonomy system for lunar surface navigation and mapping developed for the Lunar Autonomy Challenge. Operating in a GNSS-denied, visually challenging environment, our pipeline integrates semantic segmentation, stereo visual odometry, pose graph SLAM with loop closures, and layered planning and control. We leverage lightweight learning-based perception models for real-time segmentation and feature tracking and use a factor-graph backend to maintain globally consistent localization. High-level waypoint planning is designed to promote mapping coverage while encouraging frequent loop closures, and local motion planning uses arc sampling with geometric obstacle checks for efficient, reactive control. We evaluate our approach in the competition's high-fidelity lunar simulator, demonstrating centimeter-level localization accuracy, high-fidelity map generation, and strong repeatability across random seeds and rock distributions. Our solution achieved first place in the final competition evaluation.
Accurate position estimation is crucial for the successful implementation of future lunar landings using autonomous vehicles, especially in dangerous environments with sparse terrain features. In this paper, we propose a terrain relative navigation (TRN) algorithm combining our deep-learning crater detector, which was designed specifically for the NASA Crater Detection Challenge problem, and an Extended Kalman Filter (EKF). Our detector analyzes crater features from the monocular images acquired from orbit, and their matches with craters from a global database are identified via a Hungarian assignment approach followed by the consensus-based outliers removal method. The estimated measurements are then used to refine an EKF, where spacecraft pose estimation in the Lunar-Centered Lunar-Fixed (LCLF) frame of reference, augmented with altitude aiding information, constrains radial drift. The simulation results indicate that even if the spacecraft is off from its actual location up to 5 km, TRN could recover from this situation, achieving navigation error reduction to a few hundred meters. It should be noted that in order to maintain crater feature correspondences, it is important to match the image resolution and the scales within the scene to the detector training set distribution.
Batu Candan, Simone Servadio
Aerospace Engineering Department · Iowa State University · Ames, Iowa, USA
Navigation and mapping on the lunar surface require robust perception under challenging conditions, including poorly textured environments, high-contrast lighting, and limited computational resources. This paper presents a real-time mapping framework that integrates dense perception models with a 3D Gaussian Splatting (3DGS) representation. We first benchmark several models on synthetic datasets generated with the LuPNT simulator, selecting a stereo dense depth estimation model based on Gated Recurrent Units for its balance of speed and accuracy in depth estimation, and a convolutional neural network for its superior performance in detecting semantic segments. Using ground truth poses to decouple the local scene understanding from the global state estimation, our pipeline reconstructs a 120-meter traverse with a geometric height accuracy of approximately 3 cm, outperforming a traditional point cloud baseline without LiDAR. The resulting 3DGS map enables novel view synthesis and serves as a foundation for a full SLAM system, where its capacity for joint map and pose optimization would offer significant advantages. Our results demonstrate that combining semantic segmentation and dense depth estimation with learned map representations is an effective approach for creating detailed, large-scale maps to support future lunar surface missions.
Guillem Casadesus Vila, Adam Dai, Grace Gao
Department of Aeronautics and Astronautics, Stanford University, Stanford, CA, United States