Implementation of Nonlinear Model Predictive Path-Following Control for an Industrial Robot

Abstract

Many robotic applications, such as milling, gluing, or high precision measurements, require the precise following of a pre-defined geometric path. We investigate the real-time feasible implementation of model predictive path-following control for an industrial robot. We consider constrained output path following with and without reference speed assignment. Finally, we present results of an implementation of the proposed model predictive pathfollowing controller on a KUKA LWR IV robot.

Mid-Level MPC and 6 DOF output path following for robotic manipulators

“…In this article we extend the results of [5] and [9] to 6 DOF paths using a novel method of following 3 points, and argue for the flexibility of output path following systems with kinematic control. We also provide experimental results of the 6 DOF path formulation for a spiral path with MPTTC and MPFC, and exemplify the flexibility by following a 3D Lissajous path with the UR3 robot and the UR5 robot.…”

“…In [4], the MPFC is shown to converge to the path given appropriately chosen terminal constraints and penalties. In [5] the MPFC is used to generate torque inputs for a KUKA LWR IV robot. The OCP is solved using the ACADO framework [6], which uses sequential programming and the qpOASES active set solver.…”

A Comparative Study of LQR and Integral Sliding Mode Control Strategies for Position Tracking Control of Robotic Manipulators

“…In designing the LQRT, the linear dynamic equation as in (7) and 8is used. By introducing the integral action into the system in (7), the augmented system can be formed as: (20) where matrices are used to solve the Riccati matrix equation for the LQRT to obtain a unique solution of the P matrix. Moreover, x ̇e = -Cx+r, and r defines the position reference.…”

“…As many researchers have studied the position tracking problem, various modern controls have been developed to satisfy tracking performance, such as model predictive control (MPC) [20], computed torque control (CTC) [21], adaptive control [22,23], and sliding mode control (SMC), as in [3,24]. In comparison to CTC, MPC offers low sensitivity against model imperfection.…”

On model predictive path following and trajectory tracking for industrial robots

“…In [4], the MPFC is shown to converge to the path given terminal constraints without needing terminal penalties. In [5] the MPFC is implemented on a KUKA LWR IV robot, without end penalty or a terminal constraint. This is done with the ACADO framework [6], which uses a sequential programming method (SQP), iteratively solving quadratic programs approximating the nonlinear program using the qpOASES active set solver.…”