Institute of Mechatronics and Virtual Engineering Department of Mechanical Engineering Lappeenranta University of Technology Finland

From Editor

FLUID POWER RESEARCH CENTRES WORLD-WIDE

Introduction

Education

Research areas within IMVE

Research in mechatronics and fluid power

Fig. 1:Couplings within a mechatronic machine



Fig. 2:Hardware-in-the-loop simulator

Recently the mechatronic system modeling research has focused in real-time and hardware-in-the loop (HIL) applications that are seen to be the next genera-tion R&D tools in machine building. Figure 2 shows the novel principle for HIL-simulation of hydraulic driven machines proposed by IMVE. In this method the hydraulic circuit is simulated while running the real mechanics of the machine. The virtual and real part in the HIL-simulation is coupled by using hardware-interface consisting of hydraulic high-performance proportional or servo valves. The hydraulic flows are simulated and realized in the real system by using flow feedback control loops. Simultaneously, the cylinder pressures are fed back into the simulator.
In addition to the modeling and simulation research various control methods have been developed for hy-draulic servo systems including structural flexibility and flexible hydraulic manipulators. The developed methods rank from robust controllers such as sliding mode to intelligent adaptive controllers such as fuzzy and neural network controllers. Also applications of genetic algorithms and differential evolution in tuning of controllers and training neural networks have been investigated. Figure 3 shows a machine vision based in-telligent calibration system for a flexible hydraulic ma-nipulator. A CCD-camera and laser pointer is attached with the boom. The accurate position of the manipula-tor tip is detected from the fixed spots and moving laser point in X-Y board by a pattern recognition algorithm. The inverse kinematic algorithm of the manipulator is equipped by a NN trained by using the measured posi-tions. By the proposed method the static accuracy the boom in robotic positioning stayed within 10 mm's.


Fig. 3:Automatic calibration of a hydraulic robot by pat-tern recognition and NNs.

One of the most intensive research areas in IMVE has been the parallel robotics. The mecharonics and fluid power group has developed two novel parallel ro-bot concepts, namely, MULTIPOD and PentaWH. The former, shown in Fig. 4, developed for mining applica-tions provides a large workspace while the latter, shown in Fig. 5 provides a small workspace and is de-veloped for welding and machining purposes in ITER fusion reactor. ITER is an international program for developing next generation fusion power plant experi-ment. The both 5-DOF parallel machines provide a su-perior stiffness and payload capacity compared with the serial link robots. The both machines are developed by using virtual prototypes. MULTIPOD is developed in the national projects while PentaWH is the results of intensive international collaboration within European Fusion Program.
PentaWH is a part of a key robotic system that is going to be used in the final assembly of vacuum vessel sectors in ITER. The prototype robot shown in Fig. 5 is oil-hydraulic driven. Oil is not allowed in ITER and thus the next version of the robot, which is under con-struction, will be water hydraulic driven.



Fig. 4:Parallel kinematic drilling boom MULTIPOD



Fig. 5:Intersector Welding Robot PentaWH

Research in modelling of structural flexibility

The multi-body simulation approach has become a state-of-the-art procedure when complex machine systems need to be analysed. This approach is based on the concept of replacing the actual system by an equivalent mathematical model made up from discrete members. The multi-body simulation is powerful, particularly when machine members undergo large rotational motion.
The multi-body simulation approach has proven to be useful method when machine members are assumed to be rigid. The research in IMVE is focused to extend the multi-body simulation to cases where a body exhibits large deformation. This important generalization makes it possible to build a more accurate and sophisticated simulation models. The non-linear deformation should be accounted when for example tires and a slender arm of robot are under investigation. IMVE has made important contributions in the development of the absolute nodal coordinate formulation. The absolute nodal coordinate formulation is a recently discovered approach that is especially designed for the large deformation multibody applications. This formulation, which utilizes global displacements and slope coordinates as nodal variables, makes it possible to avoid the difficulties that arise when a rotation is interpolated in three-dimensional applications. The absolute nodal coordinate formulation uses a displacement field that is linear in the nodal coordinates. Therefore, the formulation leads to a constant mass matrix in two- and three-dimensional analysis. In the formulation, the shape function matrix together with the nodal coordinates is able to describe an arbitrary rigid body motion. Due to this feature of the formulation, the mass matrix remains constant even in the case of large rotations and deformations. The constant mass matrix simplifies the non-linear equations of motion since the centrifugal and Coriolis inertia forces are equal to zero. In IMVE, the absolute nodal coordinate formulation has been employed in the analysis of leaf springs, cables and belt-drives.

Research in dynamics of rigid bodies

Research of modelling methods of rigid body dynamics is motivated by the requirement of real-time simulation in order to form training simulators and other human-in-the-loop simulators. The real-time simulators, Fig.6, are complicated machine systems, and the efficient numerical solution of dynamics of the machine under investigation is only one problem to be solved. The efficiency, reliability and stability of the numerical solution are anyhow critical points in order to achieve decent behaviour of the system. The dynamics solution is used to control other functions of real-time simulators, such as visualization, motion platform, controls feedback etc.


Fig. 6:Components of real-time simulation environment
The research of modelling of rigid body dynamics is focused on the use of efficient algorithms and possi-ble numerical stabilisation methods, required to solve models consisting of several connected bodies. The connection between bodies is traditionally utilized pre-senting constraint equations that must be differentiated twice to use ordinary differential equation solution methods. This causes problems, since the position con-straints are not necessarily satisfied any more and stabi-lization methods must be used to ensure the correctness of solution, which of course increases the amount of required computational capacity. Modern modelling methods are based on methods firstly introduced in robot controllers. So-called recur-sive or semi-recursive methods have shown good com-putational efficiency, especially when models consist of several parts and several constraints. So large sys-tems with only a few degrees of freedom can be solved more efficiently utilizing these methods.
There are a couple of real-time simulators under construction in IMVE. Figure 7 presents some details of a gantry crane operator training. The gantry crane simulator is based on commercial environments, such as Simulink and Opal-RT, combined with visualization using WorldUp.



Fig. 7:a) and b) View from simulator physical cockpit, c) and d) some details of virtual harbour

The straddle carrier simulator and gantry crane simulator version 2, Fig. 8, are based on simulator environment designed and constructed in IMVE.
The simulator environment consists of separate numerical solver that can be called from visualization software, also written in IMVE. The model is sent to solver and required coordinates and other information is transferred back to visualization after a certain amount of solution steps. Unlike commercial solvers, the solver enables quick testing and comparison of several different modelling and solution methods, since the input file format is similar for all methods. Only the pre-processing of model data is done depending of solution method to be used.
Use of self-made solver also enables the connection of user into simulation utilizing for example commercial game controllers or controllers from existing machine. The control information required by motion platform is also easily available. The long reach aim of ongoing research is to form complete real-time simulation environment, which includes all components presented in Fig.6 and enables an effective method of building dynamics models suitable for real-time numerical solution. Furthermore the research of taking advantage of real-time simulation more in design processes and other applications are under consideration.





Fig. 8:A view of straddle carrier simulators and gantry crane user interfaces (under construction)

International Collaboration

 IMVE has large variety if international contacts both in the mechatronics and multi-body system research areas. The key research groups with which practical research projects have recently been carried out are:

Professor Ahmed Shabana's group in University of Illinois in Chicago

Professor Takao Nishiumi's group in National Defence Academy, Japan

Professor Peder Pedersen's group in Aalborg University, Denmark

Professor Jose Escalona's group in University of Seville, Spain

Dr Lawrence Jones's Group in Close Support Unit of European Fusion development Agreement (EFDA)

IMVE has joint publications with all these groups.
 

BACK

---

©2005 Fluid Power Net All rights reserved. Disclaimer