Abstract: Spray drift has been one of the major diffuse pollution sources for fertilizers and pesticides in intensive agriculture. Non-target drift loss of pesticides has posed a great risk to the ecological environment in recent years. It is a high demand to improve the utilization rate of pesticides for the reduction of spray drift in an orchard. In this study, a hanging-type, flexible, and targeted spray device was developed to promote the efficiency of spraying with less mechanical damage to the fruit branches and leaves. The spraying device included ultrasonic target detection, a main controller, speed measurement, and a solenoid valve control module. As the core of coordination, the main controller was responsible for the data transmission with the other three modules. A Dongfanghong MS-304 tractor was taken as the carrying platform, with overall dimensions (length × width × height) of 3.6 m × 1.5 m × 1.6 m. A three-point suspension mechanism was also used to carry the spray frame on the rear side of the tractor. The spray frame mainly included a load-bearing frame, a lifting guide rail, a lifting frame, a first-level spread arm, a second-level spread arm, a hanging spray rod, and an inverted Y-shaped nozzle bracket. Among them, the spray arm was folded twice, including the lifting frame, the first-level spread arm, and the second-level spread arm. A double cross universal joint was used for the flexible connection between the hanging spray boom and the spray arm, and a flexible spray hose was between the hanging spray end and the spray head for the flexible spraying. As such, the flexible and bendable spray hose was utilized to reduce the mechanical damage to the irregular branches and leaves. Different address codes were assigned in various modules, due to a large amount of data processing. The data conflicts were avoided to first match the address codes of the main controller as-received data. Three ultrasonic ranging sensors (TCF40-16TR1) were installed on the left and right sides of the tractor, particularly in the upper, middle, and lower order along the vertical direction. The horizontal distance was 4.1m between the ultrasonic sensor holder and the spray rack, in order to detect the target at each height. In addition, a sleeve was designed to strengthen the ultrasonic energy of the sensors in the central axis of the beam, in order to reduce the detection range under the same detection distance for better directionality of the spraying. Different levels of spray nozzles were prevented to trigger the spray ahead of time, due to the driving forward of the tractor. Specifically, the Hall sensor was utilized to detect the magnetic field strength on the tractor rim for the real-time running speed of the tractor. 14 solenoid valves were selected to control the sprinklers, four of which the inverted Y-type sprinkler brackets were controlled by one solenoid valve, and each of the rest was controlled by a solenoid valve individually. A target spray test was performed on the orange trees with a planting spacing of 4.0 m, tree height of 1.6 m, and crown diameter of 1.4 m. Two factors and three levels of orthogonal experiment were designed for the test. The first factor was the travel speed with the three levels of 0.5, 1.0, and 1.5 m/s, and the second factor was the spray pressure with the three levels of 0.4, 0.5, and 0.6 MPa. A full factorial experimental design was also performed as follows. Prior to the test, 10 sampling points were evenly arranged on the upper, middle, and lower planes on the crown of each orange tree. The droplet labeling was also arranged at the sampling points to receive the medicinal solution. The allura red staining solution was then prepared. The tractor throttle was fixed, according to the factors of the spray test and the horizontal relationship. The speed of the tractor was stabilized at the speed required for the test after debugging. The pressure was also adjusted to the required. Finally, an anemometer was used to measure the wind speed of the surrounding environment. The tractor started to travel and spray when the outdoor wind speed was less than 0.2 m/s. The spraying was automatically stopped when the tractor completely passed all the orange trees. The test results show that: 1) the average droplet adhesion rates were 84.7%, 91.7%, and 88.9% at the spray pressures of 0.4, 0.5, and 0.6 MPa, respectively, when the tractor traveled at a speed of 0.5 m/s. There was no outstanding effect of spray pressure on the droplet adhesion rate at this speed. A lower spray pressure was then selected to reduce the loss of medicinal solution. 2) Once the driving speed of the tractor was 1.0 m/s, the average droplet adhesion rates of 0.4, 0.5 and 0.6 MPa were 64.2%, 70.3%, and 75.8%, respectively. The higher the spray pressure was, the higher the droplet adhesion rate was. Therefore, a higher spray pressure was selected to improve the droplet adhesion rate in this case. 3) The average droplet adhesion rate was less than 50% at the speed of 1.5 m/s, indicating the serious spray drift unsuitable for the spraying. Decreasing spraying was observed at the droplet adhesion rate in the upper, middle, and lower layers of the canopy when the spray pressure was the same. Consequently, the greater the speed of the tractor was, the more outstanding the decreasing trend was. Furthermore, the droplet adhesion rate of the leaf back was much lower than that on the leaf front, temporarily unsuitable for spraying with high requirements. It is then necessary to adjust the nozzle angle on the hanging spray bar for a higher droplet adhesion rate on the leaf back in the future.