In this paper, we propose a persistent coverage control method to safely explore unknown environments using an environmental model learned by the sparse Bayesian approach. A sparse Bayesian classification model is introduced to estimate safety from the obtained partial environmental data by LiDAR sensors. Then, based on the control barrier function method, we propose a control law to cover the unknown environment while guaranteeing the safety of robots using a sparse Bayesian classification model. We also propose an algorithm sequentially updating the sparse Bayesian classification model with new datasets obtained through safe coverage control. Finally, we verify the effectiveness of the proposed algorithm through simulations.
In this paper, we propose a data-driven estimation method of multiple circle parameters which characterize a class of stabilizing controllers in Nyquist plots in a situation where a plant model is unknown. In the previous work, we derived the class of stabilizing controllers with an estimated circle. However, if we can estimate multiple circle parameters, we obtain multiple classes of stabilizing controllers. Therefore, we estimate multiple circle parameters and propose efficient update rules for data-driven methods.
Nonlienar compensator design is an important issue that affects performance of control systems as a whole. Despite its importance, the actual nonlinearity we encounter differs from case to case. Thus systematic design and adjustment of nonlinear compensator is still not easy, and often depend on individual know-hows.
In this paper, we focus on a nonlinear compensator called Data-driven Feedback Modulator (DDFBM), and propose a method to adjust its parameters using the system's data under operation. The DDFBM is supposed to be inserted between the pre-designed feedback controller and the plant to be controlled, which enables us to separate the nonlinear compensator design from the feedback controller design. As for the data-driven adjustment of the DDFBM parameters, we use Bayesian optimization, and show that it is effective against plant parameter perturbations.
Recently digital twins are widely utilized in various manufacturing stages. Robot off-line programming, a typical example of manufacturing digital twin, have been used for over 20 years in actual factory shops, and various methods are proposed to compensate the error between digital model and real world. This paper discusses the error compensation methods for various types of mechanical errors, and proposes a new method which can compensate positional errors of any points based on a certain number of measured points using Delaunay Triangulation. We applied this method to a large-sized forming machine and show that this method can reduce the positional error to less than half.