Tool path optimization using metaheuristic algorithms: A systematic literature review

This study aims to systematically review the post-2021 literature on tool path optimization for multi-hole drilling processes, focusing on the metaheuristic algorithms most frequently applied to this combinatorial problem - namely Ant Colony Optimization (ACO), the Bees Algorithm, the Genetic Algorithm (GA), and the Grey Wolf Optimizer (GWO) - and to comparatively evaluate their relative performance, parameter-tuning burden, and hybridization trends across different problem scales. A systematic search was conducted through Google Scholar using keyword combinations related to tool path optimization, drilling sequence, and the corresponding algorithm families; studies published from 2021 onward that directly addressed hole or tool path optimization, or related Traveling Salesman Problem formulations, were included following a PRISMA-based selection process. A total of 39 application studies met the inclusion criteria and were classified according to the algorithm family employed, then synthesized within a comparative framework alongside nine founding papers used to define the underlying algorithms. Approximately 44% of the reviewed studies employed the Bees Algorithm or ACO, reflecting their maturity across both small-scale benchmarks and large-scale industrial scenarios, while the Grey Wolf Optimizer emerged as a rapidly growing alternative in energy-focused and laser micro-hole applications. No single algorithm consistently outperformed others across all problem scales: Simulated Annealing and the Bees Algorithm performed strongly on small-to-medium regular problems, whereas Ant Colony Optimization and Dhouib-Matrix approaches proved more robust on large, irregular layouts, and hybridization of exploration and exploitation mechanisms emerged as the dominant research direction, with the best-performing hybrids reporting non-productive time reductions of up to 86.65%. Tool path optimization in multi hole drilling remains an active, rapidly evolving field in which metaheuristic algorithms deliver measurable manufacturing efficiency gains; algorithm selection should be scale-dependent, and future research should prioritize physical validation and underexplored methods such as Cuckoo Search, the Shuffled Frog-Leaping Algorithm, and Reinforcement Learning.