Christoph Hinze

Abstract

Industrial machine tool feed drives are predominantly controlled by cascade control due to their low tuning complexity and inherent robustness. However, the cascaded structure requires the inner cascades to have higher dynamics than the outer cascades, which limits the achievable dynamic accuracy. Direct control approaches, which substitute the position and velocity cascade, offer the potential to utilize the unused potential. A promising approach is model predictive control (MPC), which optimizes the manipulated variable with a plant model along a prediction horizon. However, model uncertainties between the nominal model and the real plant lead to tracking errors. Therefore, this paper presents, a linear MPC (LMPC) and an adaptive MPC (AMPC) with an additional integral action to robustly compensate for model mismatches. Both controllers use a compliant model, are real-time capable with a sample rate of \$\$\\2\\backslash,\backslashtextrm\kHz\\\$\$and consider state and input space constraints. The AMPC accounts for position-varying stiffness and friction. The controllers are experimentally compared with classical P-PI cascaded control on a ball screw drive. They show a tracking error reduction of \$\$\\37\\\backslash,\backslash\%\\\$\$(LMPC) and \$\$\\44\\\backslash,\backslash\%\\\$\$(AMPC) during a high speed motion profile and an increase in bandwidth of \$\$\\180\\\backslash,\backslash\%\\\$\$(LMPC) and \$\$\\184\\\backslash,\backslash\%\\\$\$(AMPC), resulting in significantly improved dynamic accuracy.

Abstract

The drivetrain flexibility of industrial robots limits their accuracy. To open up new areas of application for industrial robots, an increased dynamic path accuracy has to be obtained. Therefore, this paper addresses this issue by a gain-scheduled drive-based damping control for industrial robots with secondary encoders. For this purpose, a linear parameter-varying (LPV) model is derived as well as a system identification method is presented. Based on this, a gain-scheduled drive-based LPV damping control design is proposed, which guarantees stability and performance under variation of the manipulator configuration. The control performance of the approach is experimentally validated for the three base joints of a KUKA KR210-2 industrial robot. The approach realizes a trade-off between ease of implementation and control performance as well as robustness.

Abstract

The precision of machine tools is essentially determined by the feed drive control. To improve the dynamic behavior of elastic feed drives, this paper presents a novel position control concept, substituting the conventional P position controller of the industrial cascade control with a continuous sliding mode controller (SMC) together with a Kalman Filter for state estimation. The novelty of this approach lies in the focus on practical applicability: The underlying PI-cascade structure leads to low commissioning effort and the SMC controller’s parameterization is simplified to a single tuning parameter. A general compliant feed drive model is used as plant, whose parameters usually are already known from commissioning of the velocity controller. The SMC position controller is derived analytically and its stability and damping properties are investigated for the dominating mechanical mode for uncertainties of the model parameters and a wide range of controller gains. The effectiveness of the SMC-PI controller is validated for an industrial ball screw feed drive. Extensive experimental results show a considerable decrease of tracking errors in command tracking, disturbance rejection and for a high speed cutting (HSC) milling experiment compared to P-PI cascade control, while stability margins remain almost independent of the controller gains. The parameterization is discussed on the basis of a theoretical analysis and measuring results. Furthermore, excellent robustness properties regarding model errors are demonstrated.

Abstract

deformable linear objects (DLOs) such as hoses, cables or wires are often a limiting factor in robotic manipulation. For an industrial robot control, their nonlinear dynamics combined with their large variations in material parameters make it challenging to handle. Therefore, DLOs are usually manipulated manually in the industry, offering a huge potential for automation. This paper contributes to the field of robotic handling of DLOs by presenting and evaluating a novel model based controller and observer concept which is based on a multibody simulation. The presented closed-loop framework of a kinematic trajectory following controller with an online point cloud processing observer is first evaluated separately in the simulation and then in an experiment, showing its potential for cable routing scenarios.

Abstract

The handling of deformable linear objects (DLOs), such as cables or hoses, with industrial robots is hardly industrially automated due to the underactuated material behavior. Increasing computing power and decreasing cost of processors over the last decades are now allowing the online computation of more complex handling processes. Within this paper, DLOs are approximated as a multibody chain with a finite number of degrees of freedom. Based on that model, a nonlinear, partially IO-linearizing controller is derived to control an underactuated end point on a trajectory when grasped by a robot arm at the other end. The controller is verified in simulation for high dynamics and shown to significantly reduce the tracking error. A practical evaluation finally shows the qualitative suitability of the multibody approximation and is used to validate the functionality of the controller in an open-loop experiment.

Abstract

Current robot control is a rather static, monolithic application, where the application data has to be programmed, before the actual process starts. For the automation of more dynamical tasks, additional real-time control loops for process control and a more flexible robot control are needed. Therefore, this paper introduces a robot control architecture for real-time and separate asynchronous communication, which is able to build up local real-time control loops as well as including massively scalable cloud components. In a proof-of-concept, a multibody dynamic simulation of a robot and a controller are virtualized as separate components. The real-time suitability of the architecture is evaluated in comparison with non-virtualized and a monolithic variant of the same application.

Abstract

The handling of non-rigid objects, such as cables, with industrial robots is characterized by nonlinear, timedependent and location-dependent equations for the object's behavior. To manipulate them in a desired way, real-time capable simulations for control that are able to run in cycle-time are required as well as highly accurate and detailed simulations for path planning. We introduce a common simulation environment that is wrapped in a container with the advantages of including all dependencies and being easy to set up and orchestrate. Automated code generation is used to set up a standardized communication and virtualization in a Docker container such that only the simulation itself has to be provided by a user. The advantage of containerization, compared to the simulation running natively, is finally demonstrated with a sample robot simulation under heavy CPU load.

Abstract

For increasing the productivity in milling processes an estimation of the milling stability in dependency of critical parameters, such as spindle speed, cutting-depth and cutting-width is needed. In this paper an approach for online estimation of the milling stability classification via Artificial Neural Networks is provided. The network grows with the needs of the network and is trained with an Extended Kalman Filter. Some statistical training parameters are adapted during the training process and a first-order trust criterion is introduced to estimate the prediction quality of the network. The proposed network is capable for continuous online training with sorted input data. With the adaptive enhancements the training works well on analytical benchmark functions and with a 2-DOF milling stability simulation.

Abstract

Abstract The productivity of milling processes is limited by the occurrence of chatter vibrations. The correlation of the maximum stable cutting depth and the spindle speed can be shown in a stability lobe diagram (SLD). The stability is different for different width of cut and can change with the axis positions. Today it is already a great effort to estimate the \SLD\ only for one position. Many experiments are necessary to measure the \SLD\ or derive a detailed mathematical model to calculate the SLD. Moreover not only the cutting depth, but also the cutting width should be represented in the SLD. This paper presents a new approach to assess the process stability based on measured acceleration signals. The multidimensional stability lobe diagram (MSLD) are derived during the production using two new continuously learning algorithms. In this paper the application of a continuous learning support vector machine and a continuous neural network is shown. The support vector machine and the neural network are extended to make them capable for continuous learning and time-variant systems. A new trust criterion is introduced, which gives information about the prediction quality of the output for the selected input region. The learned \MSLDs\ are evaluated against analytically calculated \MSLDs\ and the learning algorithms can reproduce the analytical results very well.