Sampling-based path planning algorithms, such as Rapidly-exploring Random Tree (RRT), are widely used for motion planning in high degree-of-freedom robotic systems due to their efficiency in exploring high-dimensional spaces. However, traditional methods rely on binary collision detection, which only determines whether a sampled configuration is in a collision without quantifying its safety, often resulting in trajectories that are overly close to obstacles and reducing planning success rates, especially in complex environments with narrow passages. To address this issue, we propose Safety-Guided RRT* (SG-RRT*), which integrates a quantitative safety metric based on signed distance functions (SDFs) with a hyperoctant sampling strategy, enabling the planner to prioritize safer configurations and steer tree expansion toward collision-free regions. This approach significantly improves path planning success rates while generating safer trajectories with greater clearance from obstacles. Extensive simulations and real-world experiments demonstrate that SG-RRT* outperforms state-of-the-art methods, including RRT*, Informed-RRT*, TRRT, and Bi-TRRT, by achieving higher success rates and reducing collision risks, with only a slight increase in trajectory length.

This paper presents a Hybrid-feature Geometric Descriptor Registration (HGDR) framework to improve point cloud registration accuracy in challenging architectural environments. Traditional methods relying on single feature types—points, lines, or planes—often encounter limitations in accuracy and robustness, especially in noise-prone or low-overlap settings. In contrast, HGDR simultaneously extracts corner, line, and plane features in one step using a lightweight network and a novel Local Curvature Space Descriptor (LCSD). A two-layer graph matching algorithm leverages global geometric relations to improve correspondence reliability, followed by a two-stage non-iterative transformation estimation that avoids initialization sensitivity. HGDR was rigorously evaluated on indoor and outdoor datasets, outperforming eleven state-of-the-art rotation, translation, and RMSE accuracy methods, including uniform-feature-based and learning-based approaches. It achieved centimeter-level precision generally, with statistically lower error (mean rotation error of 0.79° and translation error of 0.173m) and significantly lower standard deviations than the competing methods. The improved robustness of the hybrid feature method is further validated through ablation experiments conducted under various mixed feature conditions.

This letter proposes Depth-Aided Incremental Image Stitching (DAIIS), an algorithm tailored for low-altitude unmanned aerial vehicle (UAV) images in urban environments. DAIIS integrates single-view depth estimation and plane fitting to identify ground feature points, enabling precise computation of transformation matrices and significantly reducing stitching errors caused by large depth variations. Unlike traditional methods, DAIIS effectively addresses misalignments and artifacts, producing seamless and accurate stitched images that closely resemble the real scene. Evaluated on three distinct datasets, DAIIS demonstrated superior performance, achieving a 70% reduction in mean square error (mse) and notable improvements in peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) compared to other methods. The results validate DAIIS as a robust solution for high-precision image stitching, enhancing accuracy and visual quality in challenging low-altitude scenarios, which offers reliable and high-quality results for various urban applications.

Preserving features of a surface as characteristic local shape properties captured e.g. by curvature, during non-rigid registration is always difficult where finding meaningful correspondences, assuring the robustness and the convergence of the algorithm while maintaining the quality of mesh are often challenges due to the high degrees of freedom and the sensitivity to features of the source surface. In this paper, we present a non-rigid registration method utilizing a newly defined semi-curvature, which is inspired by the definition of the Gaussian curvature. In the procedure of establishing the correspondences, for each point on the source surface, a corresponding point on the target surface is selected using a dynamic weighted criterion defined on the distance and the semi-curvature. We reformulate the cost function as a combination of the semi-curvature, the stiffness, and the distance terms, and ensure to penalize errors of both the distance and the semi-curvature terms in a guaranteed stable region. For a robust and efficient optimization process, we linearize the semi-curvature term, where the region of attraction is defined and the stability of the approach is proven. Experimental results show that features of the local areas on the original surface with higher curvature values are better preserved in comparison with the conventional methods. In comparison with the other methods, this leads to, on average, 75%, 8% and 82% improvement in terms of quality of correspondences selection, quality of surface after registration, and time spent of the convergence process respectively, mainly due to that the semi-curvature term logically increases the constraints and dependency of each point on the neighboring vertices based on the point's degree of curvature.