Hyper-heuristics are a set of approaches aimed at automating the design of heuristic methods to solve hard computational search problems. The term was first used in 2000 to describe heuristics for choosing heuristics in combinatorial optimization. However, the idea of automating heuristic design dates back to the 1960s. Hyper-heuristics operate on a search space of heuristics rather than directly on the problem's solution space. They are categorized into heuristic selection and heuristic generation. The paper discusses the origins, intellectual roots, and current research trends in hyper-heuristics, including their application in scheduling, timetabling, production scheduling, bin packing, workforce scheduling, constraint satisfaction, and vehicle routing. It also covers the use of machine learning, evolutionary computation, and other techniques in hyper-heuristics. The paper highlights the importance of learning mechanisms, feedback, and the distinction between online and offline learning. It discusses various hyper-heuristic approaches, including those based on constructive and perturbative heuristics, and their effectiveness in solving complex problems. The paper also identifies key research directions, such as the integration of hyper-heuristics with standard local search techniques and the exploration of additional domains where constructive heuristics are available. Overall, the paper provides a comprehensive overview of hyper-heuristics, their applications, and future research directions.Hyper-heuristics are a set of approaches aimed at automating the design of heuristic methods to solve hard computational search problems. The term was first used in 2000 to describe heuristics for choosing heuristics in combinatorial optimization. However, the idea of automating heuristic design dates back to the 1960s. Hyper-heuristics operate on a search space of heuristics rather than directly on the problem's solution space. They are categorized into heuristic selection and heuristic generation. The paper discusses the origins, intellectual roots, and current research trends in hyper-heuristics, including their application in scheduling, timetabling, production scheduling, bin packing, workforce scheduling, constraint satisfaction, and vehicle routing. It also covers the use of machine learning, evolutionary computation, and other techniques in hyper-heuristics. The paper highlights the importance of learning mechanisms, feedback, and the distinction between online and offline learning. It discusses various hyper-heuristic approaches, including those based on constructive and perturbative heuristics, and their effectiveness in solving complex problems. The paper also identifies key research directions, such as the integration of hyper-heuristics with standard local search techniques and the exploration of additional domains where constructive heuristics are available. Overall, the paper provides a comprehensive overview of hyper-heuristics, their applications, and future research directions.