Abstract: Against the backdrop of advancing the “Dual Carbon” goals and the green transformation of the coal industry, intelligent dry separation technology addresses the challenges of high water and energy consumption in traditional wet coal preparation, as well as low efficiency and poor accuracy in manual gangue sorting. It serves as a key pathway to drive the low-carbon transformation of the sector. The research history and application progress in the field of intelligent coal dry separation are reviewed, and the working mechanisms, system compositions, and advantages and disadvantages of three mainstream intelligent dry separation technologies, namely γ-ray recognition, X-ray recognition, and image recognition, are briefly described. The separation effects, industrial application cases, and research status of various technologies are summarized, and the laboratory research results of Laser-Induced Breakdown Spectroscopy (LIBS) are also introduced, which can fully grasp the current development level of intelligent dry separation technology. The relationships, contradictions, and gaps in existing research are analyzed: the fusion of multi-sensors and the optimization of intelligent algorithms can significantly improve the separation accuracy. Among them, the dual-energy X-ray combined with the Relief-PSO-SVM algorithm can eliminate the interference of material thickness, the full-spectrum fitting error of the Partial Least Squares-Monte Carlo Library Least Squares (PLS-MCLLS) hybrid model is less than 3%, and the deep learning model can improve the average accuracy of black coal and gangue recognition to 98.34%. The research contradiction of LIBS technology is identified: although it has achieved multi-element ppm-level detection in the laboratory, it is temporarily not suitable for industrial coal separation scenarios due to the limitations of explosion-proof safety, detection efficiency, and detection representativeness. At the same time, the bottlenecks of current intelligent dry separation technology in terms of complex coal quality adaptability, small particle size (0～25 mm) separation accuracy, and radiation protection cost control are clarified, and the inconsistencies between existing technologies and the needs of industrial large-scale applications are sorted out. It is pointed out that the current intelligent dry separation technology has achieved a leap from single recognition to multi-modal fusion, and has shown significant advantages in replacing manual hand sorting, lump coal separation, and fine coal separation. The core steps and development directions of subsequent research are proposed, namely developing a multi-sensor collaborative online detection system with dual-energy X-ray as the core, constructing a fusion algorithm of “physical mechanism + data-driven”, optimizing the modular design of equipment to reduce investment and operation and maintenance costs, and promoting the standardization of equipment to adapt to complex working conditions such as underground. It is predicted that this technology will support a new model of full-grain size clean utilization of coal and provide important technical support for the low-carbon and efficient transformation of the coal industry.