Soft robotic peristaltic sorting tables are a new class of industrial robots that move objects on their surface by producing moving surface deformations. To realise such deformations, the surface layer is made of a soft material with an embedded array of actuators. In industrial automation, the array must be controlled in a way that objects on the surface of the table are transported and realigned in a way that solves a given task. The control of a soft robotic peristaltic sorting table is challenging because of the high dimensionality of the control signals, unrestricted degrees of freedom in terms of the robot shape, a lack of parametric and non-parametric models, possibly complex automation tasks with multiple steps, nonlinear material properties, and incomplete knowledge about the state of the robot and the objects on its surface. In this paper, we analyse the feasibility of approaches from artificial intelligence to solve the control problem. We conclude that an automation task can be efficiently subdivided into a sequence of easier subproblems covering a range of basic, repetitive movement patterns. The input signals resulting in certain movement patterns can be optimised by comparing the actual robot shape with the ideal one. The convexity of the optimisation problem depends on the robot design and the used loss function.
展开▼
