This master’s thesis addresses the design, construction, and experimental validation of an automated robotic system intended for performing endotherapy, i.e., targeted injection of substances into tree trunks in orchards. Traditional plant protection methods face challenges such as spray drift, environmental pollution, and high water consumption, while manual endotherapy is a labor-intensive process.
The developed system integrates the fields of mobile robotics, computer vision, and robotic manipulation with the aim of fully automating the process. The robotic manipulator is designed as a TRT (translational-rotational-translational) mechanism with three degrees of freedom, enabling precise positioning and compensation for the position of the mobile carrier platform. To perform the trunk injection, a dedicated linear mechanism with a hollow injection needle is integrated, adapted to deliver high thrust force of up to 600 N.
The system software is divided into a low-level and a high-level part. The former handles calibration, UART communication, inverse kinematics computation, dynamic regulation of injection pressure and insertion speed, as well as stepper motor control. The high-level controller uses computer vision (RGB-D) to detect trunk geometry and determine the optimal injection point, processes inputs from a gamepad, and controls the mobile platform.
|
