As a case study for initial cost problem, we realize new manipulation and perception for a heavy material handling agricultural robot with high-cost components reductions, which requires more robust controllers. More precisely, first, we give a new manipulator with a passive joint and a new camera configuration with only one camera. Second, we design robust controllers using gain-scheduling and -synthesis based on some structural properties such as the friction in the presence of uncertainty, nonlinearity, and constraint, while avoiding conservative results by an analytical approach. Finally, we evaluate the validity by experiments in an actual open field, which is the most important part of this paper. The robust controllers reduce the initial cost.