Abstract Robotic path planning is a fundamental requirement for autonomous navigation, where a robot must reach a target while avoiding obstacles and producing a feasible, smooth, and efficient trajectory. This paper presents SHARP, a hybrid metaheuristic planning framework based on Particle Swarm Optimization, Sine Cosine search, and lightweight Nelder–Mead simplex refinement. The proposed framework introduces two scalarization-based multi-criteria decision layers for robotic path planning: Priority-PSN, which prioritizes path length while penalizing obstacle and boundary violations, and No-Preference-PSN, which selects a balanced solution by minimizing the normalized distance to an ideal point. Cubic-spline interpolation is further applied to convert optimized waypoints into smoother executable trajectories. The approach is validated in static and dynamic two-dimensional environments with different obstacle densities and motion patterns. In six static benchmark environments, PSN consistently produces shorter collision-free paths than PSO, GWO, and SCA. Additional ablation and function-evaluation-normalized experiments show that the full hybrid improves average path quality in cluttered maps, although this improvement is accompanied by higher computational cost. In dynamic replanning experiments, PSN achieves the highest success rate among the evaluated variants, but with increased latency. SHARP provides a practical and adaptable optimization-based framework for intelligent robotic path planning under static and dynamic constraints.
Fakhouri et al. (Sun,) studied this question.