Abstract: For automatic transplanters that complete seedling picking and planting operations using robotic arms, Cartesian coordinate, serial and parallel configuration robotic arms are currently widely adopted. However, they still have problems such as redundant degrees of freedom and large weight, making it difficult to meet the lightweight and high flexibility requirements for seedling picking and planting of small vegetable transplanters. To this end, this study designed a small, lightweight and highly integrated special robotic arm system for seedling picking and planting. Based on the 2ZS-Q2 type semi-automatic vegetable transplanter platform, by analyzing the seedling picking and planting operation space and the path requirements of the robotic arm, a small robotic arm with a P-R-R-R-R serial configuration (one translation and four rotations) and a flexible pneumatic seedling clamping system were proposed. A dual-layer control system with Orange Pi and ESP32 as the core was constructed to realize the kinematic solution and multi-axis coordinated control of the robotic arm, and the total weight of the robotic arm is 14 kg. A two-layer control architecture was constructed using an Orange Pi as the upper-level computational unit and an ESP32 as the motion control core, achieving multi-axis coordinated motion and real-time trajectory resolution. The forward kinematic model of the robotic arm was established using the improved Denavit–Hartenberg (D–H) method, and the rationality of the workspace was verified through motion simulation. The analytical solution for inverse kinematics was derived, providing a basis for motion control computation. Furthermore, based on the three-dimensional orthogonal ranging principle, a three-directional ranging and positioning evaluation system was built at the end-effector of the robotic arm. This system integrated VL53L0X miniature laser sensors on three orthogonal planes, transmitting pose data wirelessly. Positioning and trajectory tracking accuracy tests were conducted. Results showed that the average absolute positioning error was 3.587 mm, and the global average positioning error across the seedling tray was 3.613 mm. The average tracking errors for linear and curved trajectories were 3.621 mm and 5.031 mm, respectively. All accuracy indicators met the practical requirements for picking-and-placing operations. Finally, parameter optimization experiments for the robotic arm’s picking-and-placing operation were carried out. Using "Shachen No.4" pepper plug seedlings as the test subject, with picking frequency, path time ratio, and control frequency as factors, and the picking-and-placing success rate as the indicator, response surface methodology experiments were conducted. Tests revealed that both picking frequency and path time ratio had significant interactive effects on the success rate. The success rate decreased when the picking frequency was too high or when the path time ratio deviated from the optimal range. Lower values of control frequency and path time ratio generally led to a decline in the success rate, indicating that excessively low control frequency and path time ratio may cause robotic arm motion stuttering and seedling-gripping claw vibration. The optimal parameter combination was determined as follows: picking frequency of 30 plants/min, path time ratio of 30%, and control frequency of 25 ms. Under these conditions, the average picking-planting success rate was 96.02%, with a standard deviation of 1.21% and a coefficient of variation of 1.26%, which was in good agreement with the predicted value. This research can provide new insights for the development and application of fully automated picking-and-placing equipment.