Robust Tracking Control of a 3D Concrete Printer

Abstract

A relatively new and not rather common technique in construction industry is the use of 3D printers to print architectural structures and buildings. Nowadays, parts of the printed object are build in inside environments, where after the parts have to be transported to their desired location, which is very costly and time consuming. A solution could be to bring the printer to the construction site. However, current printers are not designed to be mobile or to be used directly at construction sites. This is what leads to the topic of this thesis: develop a robust tracking mechanism for the 3D concrete printer nozzle that is able to cope with disturbances that arise when having a mobile printer that is used in outside environments. Current printers have placed a mixing unit next to the construction area, pumping the concrete mixture to the nozzle. In this research, a new design concept is proposed which should displace the whole mixing unit together with the nozzle over the construction area while still ensuring high tracking accuracy of the printing trajectory. This is done by means of a dual staged manipulator. Horizontally spanned cables and an industrial crane account for the coarse displacement, whereas a Stewart-Gough platform accounts for the fine control of the nozzle. Both stages are assumed to be decoupled and by using the Euler-Lagrange approach, two separate models are derived with the corresponding equations of motion. This model is implemented in a fully parametrizable simulation environment in order to test the later developed controllers. Subsequently, a control strategy for the coarse stage is developed. Three control strategies are combined in order to guarantee a robust tracking: robustness by an \hinf controller, anticipation by a feed-forward controller, and slight tracking improvement by additional PI control. The disturbances that are taken into account are: wind, misalignment and support structure deflection. It turned out that the control strategy is able to reject these uncertainties in the simulation. The residuary error is compensated by the fine stage by using a PI control approach. The results show that the Stewart-Gough platform significantly reduces the tracking error to within a range of the desired accuracy. A high accuracy is beneficial for the reliability of the concrete structure and savings on finishing processes.