The following sub-tasks were distinguished: (1) taking the two images by the vision system; (2) image analysis; (3) motion planning and execution of the motion to the position in front of the cucumber; (4) slow approach to the cucumber; and finally (5) gripping and cutting of the stalk, storage of the fruit and the return motion of the manipulator to the home position. Unfortunately, it was not possible to automatically break down this last sub-task. Therefore, estimates of the execution times of the motion planning, gripping, cutting and return motion were determined individually after the greenhouse experiments as well. Figure 3 shows the harvest robot in the greenhouse at the IMAG research site during the field test.