Abstract: Abstract: A high-position platform has gradually been utilized to realize the heavy tasks in an orchard, such as thinning flowers and fruit, bagging, and picking in modern mechanized agriculture. However, traditional high-position platforms cannot adjust adaptively in current orchards that are mostly concentrated in hilly and mountainous areas. Particularly, it is easy to cause stress, even the staff falling down from high place when working. Therefore, it is highly urgent to improve the automatic leveling control performance of high-position platform for higher efficiency and safety in hilly areas. In this study, an automatic control system was proposed for the self-developed leveling mechanism of high-position platform using feedforward PID control. A systematic dynamic analysis was also conducted via the electromagnet, proportional valve-controlled hydraulic cylinder, and leveling mechanism. A mathematical model was then established for the feedforward PID control in the automatic leveling system. Three parts were selected to design the model, including the current PI controller, angle PID controller, and feedforward compensator. Specifically, the current PI controller was used to reduce the internal error of the system, whereas, the feedforward compensator was used to increase the response speed with a low steady-state error. Furthermore, the angle information was first transmitted from the inclination sensor to the controller. After processing the received angle information, the feedforward PID controller output the corresponding current for the proportional valve, further to drive the pitch cylinder for the extension or retraction, and finally to tailor the angle of the platform for the standard movement. As such, the simulation of leveling control system demonstrated that the feedforward PID control presented a better performance than PID control. Firstly, the rise time of feedforward PID control was 1.26 s, while the regulation time was 2.05 s, respectively, compared with PID control. Secondly, the steady-state error was 0.020, which was lower than that of PID control. At the same time, a systematic test was also carried out to verify the high-position platform model. Correspondingly, it was found that the experimental and simulated values of rising time, adjustment time, and steady-state error differed by 0.19 s, 0.37 s, and 0.04°, respectively, whereas, those of waveforms were almost the same. It infers that the mathematical model was feasible for the leveling control system of the platform in an orchard. Finally, an automatic leveling control system was built for the high-position platform to conduct static and dynamic tests. The test results showed that the leveling performance of feedforward PID control was better than that of traditional PID control. In the static leveling, the high-position platform was leveled at an angle of -4.9°, -7.4°, and -9.6° relative to the ground. The rise time of feedforward PID control was 1.57, 1.35, and 1.47 s, while the leveling time was 3.15, 2.35, and 2.62 s, excluding the system response time. More importantly, the rise time, adjustment time, and steady-state error were shortened by 20%, 30%, and 0.6%, compared with the PID control. In the dynamic leveling, the high-position platform traveled on undulating roads at a speed of 2 km/h. The maximum error of pitch angle was -3.0°, the average absolute error was 0.79°, the mean square error was 0.58°, and the inclination angle was stable at ±3° for the workbench. Consequently, the automatic leveling control system can fully meet the operating requirements of high-position platform in hilly and mountainous areas.