Abstract: Nectarines are an important economic tree species in northern China and play a vital role in regional agricultural production and orchard-based economies. However, nectarine harvesting faces challenges such as a short optimal harvesting period, high labor intensity, and continuously rising labor costs, which have become major constraints on the sustainable development of the industry. Consequently, the development and application of efficient mechanized and automated harvesting technologies are urgently required. Nectarines are an important economic tree species in northern China and play a vital role in regional agricultural production and orchard-based economies. However, nectarine harvesting faces challenges such as a short optimal harvesting period, high labor intensity, and continuously rising labor costs, which have become major constraints on the sustainable development of the industry. Consequently, the development and application of efficient mechanized and automated harvesting technologies are urgently required. As the core end-effector of harvesting robots, soft robotic grippers offer inherent compliance and reduced risk of fruit damage, yet conventional designs often exhibit insufficient structural rigidity, leading to low output force and limited posture control stability in complex agricultural environments, which ultimately limits harvesting success rates. To address these challenges, this paper proposes a novel airbag-type soft robotic gripper aimed at enhancing grasping force and adaptive capability through structural innovation, thereby enabling efficient and low-damage nectarine harvesting. The gripper consists of three airbag-type soft actuators and a variable-diameter palm. Under pneumatic actuation, the airbag-type soft actuators can generate large and controllable deformation, providing output forces several times greater than those of traditional finger-type soft actuators while maintaining good flexibility; at the same time, their inherent buffering characteristics effectively reduce contact impact and prevent mechanical damage to the fruit surface. The variable-diameter palm dynamically adjusts the initial relative positions of the three actuators based on fruit size information obtained from a visual system, ensuring a stable enveloping grasp for nectarines of different sizes and significantly improving adaptability and grasping stability. The performance of the proposed actuator and gripper system was systematically verified through a combination of mathematical modeling, finite element simulation analysis, and physical tests A mathematical model was first established to describe the relationship between actuator output force, airbag geometric parameters, and input air pressure, providing a theoretical basis for parametric design and structural optimization. Subsequently, a finite element simulation model was developed, and the results indicate that increasing the input air pressure intensifies airbag wall deformation and continuously enlarges the contact area with the fruit, which is beneficial for stable grasping and stress distribution. A dedicated output force test platform was then constructed, and test results show that the airbag-type soft actuator can generate an output force of 20.16 N at an input pressure of 40 kPa. Based on finite element simulations and expansion deformation test, the expansion displacement characteristics of the actuator were analyzed, demonstrating that the expansion displacement reaches 19.3 mm at 40 kPa. To ensure damage-free harvesting, the operating air pressure was determined to be 35 kPa by comprehensively considering nectarine damage thresholds and actuator performance, and the gripping safety under this condition was further verified through physical damage test. Finally, field harvesting test were conducted, and the results reveal that the airbag-type soft robotic arm achieved an overall harvesting success rate of 86.8% with a comprehensive fruit damage rate of only 4%, demonstrating its effectiveness and application potential. The methodology and findings presented in this study provide a valuable reference for the design and development of high-performance soft robotic end-effectors and automated harvesting equipment for fruit and vegetable production.