Abstract: Honeysuckle (Lonicera japonica), commonly known as honeysuckle or renshou in Chinese, is one of the traditional, commonly used, and bulk medicinal materials in China. Modern pharmacological research has demonstrated that honeysuckle and its active components possess significant medicinal value, while the plant also boasts considerable economic benefits in industrial applications. Currently, the picking of honeysuckle is primarily reliant on manual labor. However, with the expanding labor shortage and rising labor costs across China, the honeysuckle picking industry is confronting sustained operational pressure, and the situation is becoming increasingly severe. To address the challenge of difficult picking, several relevant studies on honeysuckle picking machinery have been conducted, such as handheld or backpack-mounted small-scale harvesters designed to assist workers in picking operations. Although these devices have improved picking efficiency to a certain extent, they still require substantial labor input and impose high physical intensity on workers, making them unsuitable for large-scale honeysuckle cultivation. Consequently, mechanized picking has emerged as the primary approach to alleviate this pressure. Nevertheless, the mechanized picking of honeysuckle is plagued by two critical issues: low picking rate and high damage rate. To enhance the picking performance of the equipment, this study established a test bench for a honeysuckle picking device. As the core component of the picking device, the picking tooth is the part that directly interacts with the honeysuckle plants. Based on the main structure and working principle of the device, theoretical analysis identified three key factors influencing the honeysuckle picking effect: tooth sleeve material, steel tooth diameter, and V-shaped groove angle. Through mechanical analysis of bud picking using composite teeth, the surface structure of the tooth sleeve was designed as a V-shaped groove contact surface, and the practical value ranges of each factor were determined by integrating actual operating conditions and machining feasibility. Firstly, a discrete element flexible model of the honeysuckle vine-bud system was developed using EDEM software. The model consists of vines and buds, with the Bonding V2 and Hertz-Mindlin (no-slip) contact models selected to simulate the mechanical behavior of the plant components. A simplified model of the picking mechanism was constructed in SolidWorks software and imported into RecurDyn software in STEP format to ensure compatibility between the structural model and the simulation platform. Secondly, a co-simulation of the picking mechanism was carried out using the EDEM-RecurDyn coupling method. Taking the tooth sleeve material, steel tooth diameter, and V-shaped groove angle as the experimental factors, and the honeysuckle picking rate and damage rate as the evaluation indicators, a three-factor, three-level orthogonal experiment was designed using the Box-Behnken design function in Design-Expert 13 software. Subsequently, response surface methodology (RSM) was employed for analysis of variance (ANOVA) to investigate the interactive effects of the experimental factors on the evaluation indicators. Considering the practical operating requirements and machining feasibility, the structural parameters were optimized, and the optimal parameter combination was determined as follows: polyurethane tooth sleeve, V-shaped groove angle of 45°, and steel tooth diameter of 6 mm. Finally, bench tests were conducted to verify the optimization results. The results indicated that under the optimal parameter combination, the honeysuckle picking rate reached 89.37% and the damage rate was 5.35%. The relative errors between the average experimental values and the simulated predicted values were 3.91 percentage points and 4.67 percentage points, respectively, which are within a reasonable range. This confirms the rationality of the optimized model values and validates the accuracy and reliability of the structural model. Field test results showed that the picking device equipped with composite teeth achieved a picking rate of 86.42% and a damage rate of 5.59%. The relative errors between the field test results and the bench test results were 3.41 percentage points and 4.29 percentage points, respectively, demonstrating that the established bench device can accurately simulate the actual field operating conditions. These results confirm the rationality and feasibility of the designed bench device and validate the appropriateness of the simplification of the discrete element model for honeysuckle plants. Comparative tests revealed that compared with the picking device equipped with steel teeth, the composite tooth-equipped device increased the picking rate by 8.48 percentage points and reduced the damage rate by 32.22 percentage points. In comparison with the device equipped with standard picking teeth, it improved the picking rate by 6.86 percentage points and decreased the damage rate by 10.88 percentage points. The picking effect of the composite teeth is remarkably superior, and the overall picking performance outperforms both the steel-tooth and standard-tooth honeysuckle picking devices. The findings of this research provide valuable references for the design and optimization of high-efficiency and low-damage honeysuckle picking devices.