ABSTRACT. Sustainable lunar exploration requires rovers to navigate challenging terrains characterized by weak, deformable surficial regolith, which often leads to immobilization due to excessive sinkage. To mitigate these risks, an innovative Deployable Claw-Wheel (DCW) mechanism is proposed. The design incorporates six internal claws deployed via a central gear system to provide active soil engagement and enhanced trafficability. The research focuses on the geometric optimization of claw tips to maximize tractive performance. Three critical design variables are investigated: claw tip width b (3–5 mm), inner arc radius r (44–52 mm), and the tangent of the wedge angle tan (0–2). A finite element model (FEM) is established using LS-DYNA, incorporating element erosion criteria to simulate soil failure and rut formation. An L25 orthogonal experimental design is implemented under a constant slip ratio of 25% to analyze mechanical interactions between the claws and the lunar regolith. Range analysis reveals that b is the dominant factor influencing both Drawbar Pull (Ft) and Sinkage (z). To map the non-linear response surfaces, a Kriging surrogate model is constructed and validated with a mean relative error within ±10%. Subsequently, the Multi-Objective Particle Swarm Optimization (MOPSO) algorithm is applied to identify the Pareto optimal set for three operational scenarios: tractive priority, stability priority, and balanced performance. Optimization results demonstrate that the refined configuration significantly outperforms the initial design. These findings offer a quantifiable design framework for future study. Specifically, for the balanced performance scenario (Case 3), the finalized optimal parameters are identified as b = 5.0 mm, r = 47.75 mm, tan = 0.69.

ABSTRACT. For future exploration missions targeting lunar craters and lava tubes, traditional rigid wheels are difficult to meet the development demands for higher traction performance and stronger environmental adaptability. Inspired by the locomotion mechanism of the desert camel’s feet, this paper designs a camel-feet-inspired elastic walking wheel, which consists of a rigid hub, eight bionic feet, and eight sets of elastic units, and conducts research on its pressure-sinkage model. The bionic elastic walking wheel enhances its traction performance through the synergistic effect of increasing the contact area and forming a fixed contour. During the load-bearing process, the elastic units are compressed, and the side plates gradually contact the ground to increase the contact area. As the elastic units are further compressed, the side plates retract inward and the bottom plate concaves to form a fixed contour, enhancing the wheel’s grip and traction performance. Taking the density of simulated planetary soil and elastic coefficient as the experimental factors and the wheel sinkage as the test index, the static pressure-bearing test was carried out. Based on the theory of terramechanics, comprehensively considering the influence of the mechanical parameters of simulated planetary soil and the elastic coefficient, a static pressure-sinkage model suitable for bionic elastic walking wheels was established on the basis of the rigid wheels’static pressure-sinkage model. The measured sinkage was compared with the model calculation results to verify the correctness of the model. This model can provide a reference for predicting the sinkage of elastic wheels.

ABSTRACT. Accurate traversability assessment is crucial for ensuring the safety and mobility of Mars rovers in complex extraterrestrial environments. Although vision-based terrain classification and segmentation methods have made significant progress in planetary surface perception, they are still insufficient for precise traversability evaluation because they emphasize semantic recognition while overlooking the bearing capacity of Martian soil. This paper proposes an intelligent Mars surface traversability assessment method that integrates terramechanics theory with deep learning-based scene understanding. Rather than treating terrain segmentation as the final goal, the proposed method uses it as an intermediate step to extract geometric features of surface objects and establish physically interpretable traversability criteria. First, an instance segmentation network is employed to accurately detect independent obstacles, especially rocks, and extract their contours and size-related features. These features are then combined with a previously developed wheel-soil interaction model, the inverted range of Martian soil parameters, and the structural parameters of the rover to construct traversability criteria from both geometric and support constraints. Finally, segmented terrain features are quantitatively mapped into traversability scores, and an intelligent traversability risk assessment framework for the Utopia Planitia region is developed. The proposed method tightly couples visual perception with physical modeling, providing a more interpretable and reliable solution for Mars rover traversability analysis and supporting autonomous navigation, path planning, and mission safety assessment for future planetary exploration.