Sine cosine particle swarm optimization algorithm for optimizing large scale issues

Smarter Search for Tricky Problems

From planning safe robot routes through crowded warehouses to fine‑tuning power grids and engineering designs, today’s toughest computing tasks often involve thousands of variables and tangled constraints. Traditional search methods can get overwhelmed, wasting time exploring dead ends or settling for so‑so answers. This paper introduces a new computer algorithm that tackles such large, complex problems more reliably and efficiently, and shows how it can guide robots through cluttered, changing environments without collisions.

Why Big Problems Are Hard to Tame

Many modern challenges fall into what researchers call large‑scale optimization: finding the best combination among a huge number of possibilities. Imagine trying to pick the shortest, safest route for a robot when the map is packed with walls, moving obstacles, and narrow passages. Popular methods like Particle Swarm Optimization (PSO) and the Sine Cosine Algorithm (SCA) mimic groups of simple agents searching a landscape. They are easy to use and fast, but they tend to crowd around good‑looking regions too quickly. As a result, they often get stuck in local “good enough” spots instead of discovering the truly best solution, especially when the number of dimensions—independent choices that must be made—becomes very large.

Blending Two Swarms into One Stronger Search

To overcome these limits, the author designs an improved hybrid method called ISCA‑PSO, which fuses an enhanced version of SCA with PSO. Two key ideas power the upgrade. First, a Dynamic Position Correction mechanism continuously measures how far each candidate solution is from the current best one. When a solution is far away, the algorithm nudges it more strongly toward the best region to speed up convergence. When it is already close, the method allows it to keep some independence so the swarm does not collapse too quickly into a single point. Second, an Orthogonal Crossing mechanism borrows tools from experimental design to mix information from different good solutions in a systematic way. Instead of random crossover, it generates a small but carefully chosen set of new candidates that spread across the search space, preserving diversity while avoiding a combinatorial explosion of possibilities. Together, these mechanisms let the hybrid algorithm balance broad exploration with focused refinement.

Putting the Hybrid to the Test

The new algorithm is tested on a suite of demanding benchmark functions that researchers around the world use to compare optimization tools. These functions are designed to be deceptive, with many peaks, valleys, and flat regions that can trap naive search methods. Across several modern test collections and in dimensions up to ten thousand, ISCA‑PSO consistently reaches solutions extremely close to the theoretical best, while keeping variability between runs very low. In contrast, the original SCA and several other advanced methods lose accuracy or become unstable as the dimensionality increases. Statistical tests confirm that the performance gains are not due to chance: the hybrid method generally ranks first among competing algorithms on high‑dimensional problems.

Guiding Robots Safely Through Crowded Worlds

To show that these improvements matter in practice, the study applies ISCA‑PSO to robot path planning. Paths are described as smooth curves built from sequences of waypoints, and safety is enforced by requiring a minimum distance between the robot’s body and any obstacle. In two complex 2D maps filled with barriers, the new algorithm finds shorter paths than several well‑known alternatives, while keeping wider safety margins. In more demanding simulations with many moving obstacles, it plans routes that avoid collisions entirely in repeated trials, while maintaining reasonable computing times. The paths are not only short but also relatively smooth, making them easier for real robots to follow. Even in extreme 2D mazes with hundreds of obstacles and in intricate 3D scenes packed with barriers, ISCA‑PSO achieves the best combination of path length, success rate, and robustness among a range of state‑of‑the‑art methods.

What This Means for Real‑World Systems

For non‑specialists, the key message is that this work delivers a more dependable way to search enormous, messy solution spaces. By carefully steering a virtual swarm—tightening around promising areas without losing the ability to explore—the ISCA‑PSO algorithm finds better answers more often, even as problems scale up dramatically. In concrete terms, that means robots that can weave safely and efficiently through crowded, changing environments, and optimization tools that can be trusted on complex engineering tasks where both accuracy and speed matter. Future extensions that add richer awareness of terrain and environment could make this approach even more broadly useful in real‑world autonomous systems.

Citation: Wang, Y. Sine cosine particle swarm optimization algorithm for optimizing large scale issues. Sci Rep 16, 12303 (2026). https://doi.org/10.1038/s41598-026-41180-4

Keywords: large-scale optimization, metaheuristic algorithms, particle swarm, robot path planning, sine cosine algorithm