Abstract: Abstract: In recent years, garlic planting area has become larger and larger in China, however, the level of mechanization of domestic garlic cultivation is low, at present, so all the garlic cultivation still mainly relies on manpower to complete, with low working efficiency and high labor intensity, moreover, the randomness of planting parameters is not conducive to the development of garlic mechanization. Aiming at the problem of low single seed rate in garlic mechanized planting, a claw type circulation single seed taking device was designed with the design idea of "taking more and keeping one". The dynamic model of garlic seed filling and clearing was established with discrete element technology. The internal mechanism of single seed picking was clarified through single factor simulation test. In the seed-picking process, the arc radius, center angle and lateral spacing of the middle plate of the seed taking claw were taken as the test factors, and the target rate was taken as the test index. Based on the box Behnken test design principle, the multi factor simulation test was carried out, and the parameters that affect the target rate were successively the lateral spacing of the side plate of the seed taking claw, the center angle of the middle plate and the arc radius of the middle plate. In the seed cleaning process, the inclination angle of the seed cleaning grid plate was taken as the test factors, taking the qualified rate and the missed seeding rate as the test indexes. Through the one factor test design principle, the seed cleaning performance test was carried out, and the response curve was obtained. Response curve was drawn using Design Expert 8.0.6 to investigate the effect of grid tilt angle on the response value of target rate, missing rate and replay rate, and optimize the parameters of seed selection. The response curve showed that the target rate increased first and then decreased with the increase of grid tilt angle, and the target rate reached the maximum value when the grid tilt angle was about 7°. The missing rate decreased first and then increased with the increase of grid tilt angle; the replay rate increased with the increase of grid tilt angle. The increase of gate inclination decreased first and then increased, but the increase amplitude was small. The results showed that the optimal values of the parameters were the arc radius of the middle plate was 48.52 mm, the center angle of the middle plate was 72.59%, the lateral spacing of the side plate was 25.11 mm, and the inclination angle of the grid plate was 7.41°. The predicted target rate of the model was 90.64%, the qualified rate was 92.52%, and the leakage rate was 3.30%. Laboratory and field experiments were carried out, and the experimental data were consistent with the optimization results, which provided a reference for the study of garlic mechanized seeding single seed picking technology.