Regarding the problems in the seed supply stage of the lateral planter for precut double-bud sugarcane segments, such as complex seed supply mechanisms, low efficiency, low steering accuracy, susceptibility to jamming, and difficulty in meeting lateral precision planting requirements, a chute steering seed supply device was proposed. The working principle of the chute-type lateral seed supply device was expounded, and a theoretical mechanical model for seed steering was established. The influence patterns of key structural and operational parameters on sugarcane segment steering were analyzed, enabling determination of preferable parameters for lateral sugarcane seed supply. Discrete element method (DEM) simulations of sugarcane collisions demonstrated that the lateral rotation of sugarcane segments was primarily governed by the rotational torques generated along the x and y axes during their impact within the chute. These bidirectional rotational torques were influenced by three key parameters: the steering drop height, chute width, and inclination angle of the chute base plate. Based on simulation-guided single-factor experiments, the results indicated that the steering qualification rate was initially increased and then stabilized with the increase of chute base plate angle, plateauing beyond 35°. At a drop height of 100mm, insufficient drop energy resulted in a significant decrease in qualification rate. Furthermore, when the chute width exceeded the sugarcane length, sugarcane tended to undergo longitudinal deflection, resulting in reduced lateral steering qualification rate. Box-Behnken Design response surface methodology experiments and validation tests confirmed that chute base plate inclination, steering drop height, chute width, and their two-way interaction terms significantly influenced the steering qualification rate. The maximum qualification rate of 98.4% was achieved at a base plate inclination of 34.9°, steering drop height of 296.5mm, and chute width of 214.9mm. The mere 0.6 percentage points discrepancy between the predicted value (99%) and experimental result (98.4%) validated the model's reliability.