In recent decades, there has been rapid development of multi-sensor data fusion because of its versatility and broad application areas, heightening the requirement for multi-sensor calibration. There are many proven solutions, but they cannot thoroughly meet all the evaluation criteria, such as accuracy, automation, and robustness. Therefore, this review aims to contribute to this growing area by reviewing recent research into multi-sensor calibration and proposing possible future research directions. The main part of this literature review comprehensively explains the different characteristics and conditions of various multi-sensor calibration method. In addition to summarizing different standard target-based offline calibrations, this paper also focuses on online calibration methods used for indoor and outdoor SLAM, mobile mapping and autonomous driving. Online calibration includes traditional motion-based calibration as well as feature-based calibration. Features from the environment and mutual information(MI) are two main automatic feature-based calibration approaches discussed in the paper. The review also discusses the coarse-to-fine calibration process. Motion-based calibration can provide initial values to feature-based calibration. Besides, some registration methods can also be used for refinement. Finally, this paper addresses the critical factors for evaluation, offering valuable insights into inspiring future research directions. Considerably more research should focus on the capability of online targetless calibration and simultaneous multi-sensor calibration.