Improved Harris Hawks Optimization Algorithm

Abstract

The Harris Hawks Optimization (HHO) algorithm is a nature-inspired metaheuristic that mimics the cooperative hunting behavior of hawks. Despite its success in various optimization tasks, it suffers from several limitations, including low computational accuracy, a tendency to become trapped in local optima, and difficulty in balancing exploration and exploitation. To address these challenges, this paper proposes an enhanced version of HHO, named FL-HHO, which integrates four key improvements: the Halton sequence for enhanced population diversity, a modified Escaping Energy Factor E, an improved Frog-leaping mechanism, and a convergence trend analysis module. FL-HHO is evaluated on seven classical benchmark functions and 30 functions from the CEC2014 benchmark suite. The experimental results demonstrate that FL-HHO exhibits a significant advantage on classical benchmarks, achieving top performance in search precision across nearly all functions and reaching the theoretical optimum on three of them. In terms of computational efficiency, FL-HHO ranks third among all compared algorithms. On the CEC2014 benchmarks, it secures first place on over 50% of the functions, with slightly lower performance observed on certain multimodal functions. Ablation experiments further verify the effectiveness of each proposed component, particularly highlighting the contribution of the modified Frog-leaping mechanism to global exploitation and the Halton sequence to initialization robustness. In practical scenarios, FL-HHO is applied to industrial robot path planning, where it achieves the shortest travel distance among all evaluated methods, confirming its effectiveness in real-world tasks. The implementation code is publicly available at: https://github.com/zhu-cheng/FL-HHO/tree/main.