Unmanned Aerial Vehicles (UAVs) have been widely used in rice direct seeding in recent years, due to the flexibility and high efficiency suitable for the terrain. Among them, UAV broadcast sowing has been one of the most UAV rice direct seeding, particularly with better seeding uniformity and work efficiency, compared with manual seeding. The broadcast sowing device can also be divided into the centrifugal disc and pneumatic types in China at present. However, the UAV broadcast sowing is easily affected by the rotor wind field, leading to uneven seeding. At the same time, the effect of seeding in the rows and holes can also result in air permeability and occurrence of diseases during the growth of rice in field management. In this study, a control system was designed for the rice shot seeding device in a flight controller order, in order to improve the uniformity and the accuracy of the seeding rate during UAV rice direct seeding. A UAV ground station function was also established to develop the experimental prototype. A closed-loop control was realized in the speed of the stepping motor using the Proportion Integral Derivative (PID). The calibration was then conducted to evaluate the excitation force of the vibration motor and the speed of the friction wheel motor. Finally, the seeding control program was designed to control the whole process of rice shot seeding using a Finite State Machine. The control functions included operation route planning, seeding rate calibration, parameter setting, seed remaining quantity display, and automatic seeding, in order to more easily realize the automatic operation of rice direct seeding. Taking three-fold pelleted rice seeds as the seeding objects, the seeding performance of the prototype was verified from three aspects: the accuracy of the seeding rate, the row effect, and seeding uniformity. The results showed that the average relative error of the seeding rate was less than 4% when the prototype flew at the speed of 1.0-2.5 m/s under the simulation. An excellent performance was achieved in the dynamic adjustment for the seeding control system, particularly with the relatively accurate seeding rate. Specifically, the average probability of seeds was 75.18% within the seed row width of 15 cm, when the prototype was seeding at the height of 2.0 m. By contrast, the average probability of seeds was higher than 80% within the seed row width of 12 cm, when the prototype seeding at the height of 1.0 and 1.5 m, indicating the better performance of the seeding row. Correspondingly, the working height of 1.5 m was preferred, in terms of safety. The average Coefficient of Variation (CV) of seeding uniformity was 20.51%-35.52% when the prototype worked at the height of 1.5 m with a speed of 0.5-2.0 m/s, and the seeding rate of three-fold pelleted rice seed of 90-150 kg/hm2 (corresponding to the seeding rate of naked seeds was 22.5-37.5 kg/hm2 ). It infers that the working speed greatly contributed to the seeding uniformity. Two field experiments were carried out, according to the preferred seeding parameters, where the relative errors of the seeding rate were 2.47% and 4.12%, respectively, the seeding uniformity CV values were 22.17% and 21.82%, respectively, and the seed breakage rates were 0.34% and 0.18%, respectively. The seeding control system fully met the control accuracy requirements of UAV rice direct seeding, according to the standard Technical specification of quality evaluation for the aerial broadcast seeder by remote control (standard NY/T 3881-2021). This finding can provide a strong reference for the UAV rice direct seeding. [ABSTRACT FROM AUTHOR]