Design of a new end effector for a robotic arm for grapes harvesting

The constant advances in agricultural robotics aiming to overcome the challenges imposed by the needs of environmental preservation wastes and costs reduction have given growing importance to the concept 4.0 agriculture, the newest field of the farming sector. This thesis focuses on the automation part included in the wider concept of smart agriculture: in particular, it explores what an end effector is and how it has to interact with the environment in order to carry out the task it is meant to undertake. The manipulator that has been developed wants to be an effective device, able to detect, grab, cut and move bunches of grapes in a smart vineyard. The starting point has been a robotic arm that was built by other colleagues and which also had an end effector needing improvements: thanks to inspirations from state of the art robot that are currently available for purchase, many prototypes were designed to explore different solutions and identify the best fit for the final device. Even though the structure was initially thought with the idea of using aluminium for the custom parts, in the definitive end effector they were realised with 3D printed plastic, to meet the needs of a reduced weight and to allow modification on the run due to needs of the team or to house new electronic parts (such as sensors ad camera). To actuate the motion of the grasping fingers, an unused stepper motor was repurposed and used to drive a worm gear, thus converting horizontal rotation into a rotation around the vertical axes of the two main shafts. For the motor control part, the project relied on a custom built board called Virtlab which also controlled all the other motors on board of the robot: the board communicated with a different driver for each motor and required a knowledge of C for microcontrollers in order to be programmed correctly. The end effector needed to come equipped with a cutting device, in order to free the field of view for the grasping action and to detach the bunch from the plant: the solution that made it to reality is a portable electric chainsaw, adequately modified to fit the arm and to not weigh excessively. The ON/OFF control of the chainsaw was implemented thanks to a Nucleo expansion shield whose job is to drive CC motors; the procedure requires some programming that was done in the STM32 Cube IDE, exploiting ready-to-use libraries. As stated before, the end effector needed to accommodate a camera and a distance sensor, key elements for the task of detection and tracking of the bunches of grapes: for the camera an apposite holder was designed, while the sensor was placed in the back of the structure, held in place by plastic support with its laser beam passing through the back plate and between the two sets of fingers. The final device is a new working end effector that, when mounted on the robotic arm and guided by the AI algorithms of tracking and detection, can autonomously spot a bunch of grapes, trim the surrounding leaves and harvest it. Even if the definitive end effector is far from being perfect and could use some tweaking and modifications, it still proves to be concrete solution to reach the goal it was designed for and can constitute a starting point for future studies and technologies in the field of vineyard grapes harvesting.