Abstract: Horizontal planting has been used for fresh sweet potatoes in recent years. A large number of sweet potatoes can be produced, even potato block sizes. It is necessary to place potato seedlings after trenching transversely on the potato ridges when using the horizontal transplanting manually. The labor-intensive and inefficient have severely constrained the development of the sweet potato industry. However, there is no horizontal transplanting machine for sweet potatoes. In this study, a horizontal transplanting machine was designed for sweet potatoes, according to the agronomic requirements. Two rows of sweet potatoes were realized in the rotary tillage, horizontal planting, soil covering and ridge repair, and laying drip irrigation belts in one operation. High transplanting quality and transplanting efficiency were achieved in the current process of sweet potato seedling transplantation. The transplanting motion of sweet potato seedlings was theoretically studied using the EDEM-RecurDyn coupling simulation. The main influencing factors on the transplanting quality of sweet potato seedlings were determined as the forward speed of the transplanting machine, chain speed, soil depth, and seedling claw height. A coupling model was constructed to simulate the seedling operation process for the planting device-flexible sweet potato seedling-sweet potato ridge. The simulation determined that the seedling trajectory was a short cycloid, and the seedling claw height was 50 mm. An analysis was made on the movement of sweet potato seedling posture and soil particles during transplantation. The final position and ridge shape of sweet potato seedlings were confirmed, according to the agronomic requirements for horizontal planting. The Box Behnken experimental design was used with the machine forward speed, chain forward speed, and soil depth as experimental factors, while the qualified rate of planting depth, planting spacing, and transplanting efficiency as evaluation indicators. The level was also evaluated for the influence of experimental influencing factors on various experimental indicators. A quadratic regression orthogonal experiment was conducted to determine the optimization model. The validation test results showed that the optimal combination of working parameters was the moving forward speed of 0.4 m/s in the machine, the moving forward speed of 0.2 m/s in the chain, and the soil depth at 46 mm. At this time, both qualified rates of planting depth and spacing were 92% at the transplanting efficiency of 263 plants/min. The relative error between the average value of the field experiment and the optimized value was less than 5%, indicating an accurate regression model. The number of sweet potatoes planted using a horizontal transplanting machine was 3-6, and the size of the sweet potato blocks was uniform, especially suitable for the commodity properties of fresh sweet potato blocks. The transplanting machine performed better planting, fully meeting the agronomic requirements for horizontal planting of sweet potatoes. The findings can provide a strong reference for the design and optimization of sweet potato transplanting machines.