Abstract: The intelligent development of coal mine roadway excavation has been progressing slowly, with the serious issue of mining-excavation imbalance significantly constraining the high-efficiency production of coal mines. Improving excavation efficiency is a crucial issue facing the current stage. As one of the crucial equipment for roadway excavation operations, the boom-type roadheader exhibits advantages in cutting hard rock or complex geological coal seams. The traditional memory cutting method of the boom-type roadheader has shortcomings in dealing with the tunnel cross-section under complex geological variations, failing to meet the requirements of autonomous cutting. In response to this issue, a boom-type roadheader adaptive cutting control method that considers the geological strength index (GSI) is proposed. By guiding the cutting operations of the boom-type roadheader based on geological characteristics, the excavation efficiency and level of intelligence can be enhanced. Firstly, photographs of the excavation roadway cross-sections are taken using machine vision. Cracks are detected using methods such as connected domain extraction and skeleton linking to identify the geometric parameters of the cracks, including quantity, length, width, and density. These geometric parameters of the cracks are used as inputs, and geological strength indicators are used as outputs to establish a neural network model for predicting the geological characteristics of the excavation roadway cross-section. Secondly, a geological strength indicator grid map is established, and an improved ant colony optimization algorithm is employed for cutting trajectory planning on the grid map. By optimizing the objective function through improved initial concentration of pheromones and pheromone evaporation strategies, with the objectives of optimizing time, energy consumption, and geological matching, the optimal cutting trajectory is determined. A model for controlling the rotation speed of the cutting head and the swinging speed of the cutting arm is established to control the cutting speed passing through the current grid, matching the geological strength indicator of the current grid to enhance excavation efficiency. Finally, a stress analysis of the cutting model under different geological strength indicators was conducted in finite element software. A control platform for the boom-type roadheader was set up in the laboratory to compare the performance of cutting trajectory planning and cutting speed control methods with other optimization algorithms. Experimental results indicate that the predicted accuracy of geological intensity index is 95.5%, the consumption time of cutting the complete roadway section is reduced by 6%, and the energy consumption is reduced by 19%. Compared to other methods, this approach can reduce energy consumption and improve cutting efficiency. This method provides a new direction for intelligent excavation and autonomous cutting of boom-type roadheader.