Aiming to address issues such as insufficient soil loosening and root-stem accumulation during ginger harvesting using combine harvesters, a collaborative harvesting system was developed. The system configuration and operational mechanisms were thoroughly analyzed, with coordinated optimization implemented for motion parameters, load distribution, geometric characteristics, and spatial relationships of core components. Utilizing the EDEM – RecurDyn coupled simulation algorithm, an interactive model integrating ginger plants, soil, and harvesting equipment was established to identify key factors affecting harvest quality. Macro-micro multi-scale simulations revealed dynamic coupling patterns among ginger plants, soil media, and harvesting devices. Experimental parameters included harvesting speed, excavation blade vibration frequency, and the ratio of excavation blade inclination angle to clamping device angle. Soil cohesion failure threshold and ginger-soil adhesion rate served as evaluation metrics, with a three-factor orthogonal rotation experimental design employed. Optimized parameters yielded as follows: 0.33m/s of forward speed, 5.4Hz of soil vibration frequency, and a 0.96 ratio between excavation blade and clamping device angles. These conditions resulted in 95223 Bond fracture events between soil particles, 9.93% ginger-soil adhesion rate, and optimal harvesting performance. Field trials confirmed the enhanced efficiency of the optimized system, with measured data closely matching regression model predictions. The synergistic interaction between the optimized vibratory separation unit and clamping mechanism ensured effective soil fragmentation and orderly material transport.