Path planning is a fundamental challenge in autonomous mobile robotics, where the raw outputs of classical grid-based planners such as \(A^*\) contain sharp turns and redundant waypoints that are unsuitable for direct execution. Tracking such unsmoothed paths with a Model Predictive Controller (MPC) on a \(100\,\text {m}\) grid produces positional errors exceeding \(10\,\text {m}\) and highly oscillatory acceleration commands, confirming the critical need for post-hoc trajectory smoothing. This paper proposes a three-stage hybrid framework: (i) an Improved \(A^*\) search with Dynamic Programming (DP) generates a globally feasible path and prunes redundant waypoints; (ii) an Adaptive Random Search Optimization (ARSO) algorithm adaptively places one control point per waypoint triplet by minimizing the deviation between the smoothed spline arc-length and the original segment length; and (iii) cubic spline interpolation enforces \(C^2\)-continuous trajectories via natural boundary conditions, without requiring gradient computation, convex decomposition, or neural network training. Validated on two benchmark occupancy-grid environments (a \(100\!\times \!100\) cell map at \(2.5\,\text {m/s}\) and a more demanding \(160\!\times \!160\) cell map at \(3.0\,\text {m/s}\)), the smoothed reference reduces MPC tracking error by up to 100-fold: lateral deviation falls from \(3.0\,\text {m}\) on the raw pruned path to \(1.5\,\text {cm}\) on the ARSO-smoothed path, with residual errors converging to near-zero within \(10\,\text {s}\) in both scenarios. The framework is gradient-free, training-free, and directly compatible with standard occupancy-grid navigation stacks, making it practical for autonomous ground robots and aerial vehicles operating in obstacle-dense, real-time environments.

This study investigates the factors influencing customer engagement in AI-powered fitness applications and the subsequent impact on behavioral outcomes such as word of mouth and sustainable consumption. A quantitative research design was employed, with data collected through a structured survey from users of AI fitness applications, and analyzed using Structural Equation Modeling (SEM). Drawing on the Information Systems Success Model and engagement theory, the research examines the roles of service quality, system quality, information quality, health consciousness, anthropomorphism, and personal innovativeness. Findings reveal that higher perceived quality across service, system, and information dimensions, coupled with health consciousness and human-like features, significantly enhances user engagement. Engagement, in turn, drives both advocacy and sustainable behaviors, while personal innovativeness selectively amplifies the effect of system quality. The study advances theoretical understanding by adapting and extending established models to the context of AI-driven health technologies, while also providing practical insights for the development of intelligent, user-centric fitness applications that promote sustained engagement and responsible use.

Vehicle theft and vandalism remain significant urban security challenges commonly addressed through reactive, post-incident forensic measures. This paper proposes a proactive, real-time computer vision system designed to detect potentially suspicious behavior around parked vehicles, with a specific focus on unauthorized proximity and loitering. The proposed architecture integrates state-of-the-art object detection using YOLOv11 (You Only Look Once version 11), multi-object tracking via a lightweight custom association tracker inspired by the ByteTrack/StrongSORT/OC-SORT paradigm, and monocular depth estimation based on the Intel DPT-Large framework.A key contribution is the identification and mitigation of the Perspective Challenge: the two-dimensional (2D) scale ambiguity that causes distant background pedestrians to appear falsely proximate to foreground vehicles in monocular camera feeds. To address this, three spatial analysis strategies are implemented and evaluated: (A) fixed Euclidean thresholding, (B) adaptive perspective thresholding, and (C) three-dimensional (3D) depth injection. Experimental results on real-world urban surveillance footage (27,000 annotated frames across two datasets) demonstrate that Strategy C achieves the highest precision (0.95) with an F1-score of 0.92, while Strategy B provides the best balance between accuracy (precision 0.88, recall 0.91, F1 0.89) and computational efficiency (32.7 frames per second, FPS). Compared to naive 2D thresholding (Strategy A), Strategy B reduces false alarms by approximately 80%, while Strategy C further improves precision to 0.95 through depth-plane verification. The system maintains real-time performance exceeding 30 FPS under Strategy B, making it a strong candidate for practical urban vehicle monitoring, subject to further large-scale validation across diverse environments.