Time Optimal Path Tracking for Industrial Robots with Low Cost Computational Unit using Model Predictive Control. Issue 1 (17th December 2018)

• Record Type: Journal Article
• Authors: Jörgl, Matthias; Gattringer, Hubert; Müller, Andreas
• Other Names: Müller G. guestEditor.; Ulbrich M. guestEditor.
• Is Part Of: Proceedings in applied mathematics and mechanics. Volume 18:Issue 1(2018)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2018-12-17
• Subjects: Applied mathematics -- Periodicals; Engineering mathematics -- Periodicals; Mathematical physics -- Periodicals; 519
• DOI: 10.1002/pamm.201800379
• Languages: English
• ISSNs: 1617-7061

Abstract

Abstract: This paper deals with time optimal path tracking along prescribed geometric paths for robotic systems, like computerized numerical control (CNC) machines. These machines are typically controlled by a low cost microcontroller computational unit (MCU). Such MCUs are commonly used in industrial applications but are limited in view of computational performance. In many cases, these robotic systems have not only a MCU on which e.g. a motion controller is implemented, but also an associated desktop computer that serves as input device and transmits raw data to the MCU. Therefore, the calculation of the time‐optimal path following can be divided into two steps, where the paths are in our case defined as B‐Splines. On the desktop computer, an approximation of the solution is recursively calculated using a log‐barrier method, which solves a convex reformulated path tracking problem with velocity and torque constraints. This log‐barrier solution serves then as a reference generator for a model predictive controller (MPC), which is implemented as a trajectory tracking controller on the MCU. Due to the special formulation of the MPC, the continuity of the solution can be chosen arbitrarily. Thus the method is suitable for control of elastic systems, where C4 continuity is required to avoid vibrations. Assuming that the MPC works well as tracking controller, it is possible to consider only input constraints for the MPC and thus to minimize the computational effort. At last, a validation by means of simulation results is shown, in which the ACADO Toolkit is used for the MPC implementation.