Robotic abrasive belt grinding of complex curved blades based on a novel force control architecture integrating smooth trajectories

Constant force control and smooth trajectory planning are regarded as key strategies to improve the surface quality of blades in robotic abrasive belt grinding. However, some technical challenges still remain: (1) external normal grinding force with time-consuming sensing, large fluctuations, and force overshoot, and (2) lack of global C2 continuous grinding trajectories. In this paper, a unified architecture is proposed to concurrently address the above issues. The main contribution of this paper is to provide a complete solution for robots to automatically accomplish the grinding of complex curved blades. The grinding process is arranged as follows: Firstly, cubic B-spline and C2 continuous quaternion spline curves are used to generate the global C2 continuous trajectory, thereby avoiding vibrations and unexpected slowdowns during grinding. Secondly, the actual grinding force is sensed based on a low-speed Fourier series trajectory and the least square algorithm. Finally, a novel force control scheme consisting of an adaptive hybrid impedance (AHI) control scheme and two nonlinear tracking differentiator (NTD) modules is proposed to overcome the problems of large force fluctuations and force overshoot. Experimental results demonstrate that the proposed solution delivers better performance. Furthermore, the surface roughness Ra of the machined blade is controlled within 0.8 mu m, which meets the requirements of real-world industrial applications.