Traditional machine tools used in machining or assembly systems are characterised by a typical serial structure, where each axis pulls the weight of subsequent axes and provides their motion. There are plenty of mechanical structures already consolidated, which have been investigated and analysed in great detail by engineers but the problem is that such structures are hardly flexible; once the machine has been designed, there is no more space for further improvements: only slight modifications can be done. In traditional machine tools it is not uncommon to have problems also in modifying the size of the machine (smaller or larger) because even little changes affect the overall design. As far as the innovation is concerned, it is very difficult to expect serial machine tools to change greatly in the next future.

In assembly systems, the problem appears when high velocity and high force are required, in addition to a very safe behaviour. This is not allowed by traditional cartesian or articulated robots. It can be estimated that new assembly stations with very high forces required (in excess of 25.000 N or more) can be definitely more effective than traditional robots. Critical assembly stations (characterised by high failure rates because of  lack of stiffness or shortage of flexibility) may also take great advantage with new solutions. These new technical solutions could be also very competitive if they use standard components.

In 1994 two American machine tool companies, Giddings & Lewis and Ingersoll, surprised the world with the presentation of a new type of machine tool. These new systems were based on the parallel link kinematics structure, first developed by Steward in England in1965, which represented a further development of his patent from 1964. The two companies called these new machine tools ”Hexapod”. This name originates from the six basic constructional components of the machine - telescopic, numerically controlled drive units which, by changes in their respective lengths, control six degrees of freedom in position and orientation of a platform. After examination of the background of these developments, it quickly becomes apparent that a variety of ideas for the application of parallel kinematics structures to the fields of robotics and machine tools have been developed. Most of the applications were developed after 1995. This implies that the large number of prototypes presented at the EMO ‘97 is a result of research inspired by the American activity in 1994. Also at the latest EMO ’99 in Paris, a small number new parallel kinematics have been shown. One common characteristic of all machines to date, is the fact that they are prototypes or machines with very little real industrial use. The search for a machine tool structure with superior performance in the areas of accuracy, dynamics and price in comparison with conventional designs is clearly underway. A definite direction is, however, not yet clear. It is to be noted, that across Europe intensive research and development is being carried out in the search for innovative ideas for parallel link machines.

However, the question concerning the necessary configuration for real technical progress compared with conventional high-speed machine tools has not been answered. Many questions still remain to be solved: the poor ratio of machine size to working envelope in systems with telescope arms as well as for those with constant arm lengths, the disadvantageous non-linear conditions of motion and force-ratio, the limited slew angle of £ 30° for the moving platform in systems with more than three axes, machine accuracy dependent on thermal properties, static and dynamic behaviour, joint hysteresis, the design-specific cost in comparison with conventional machine tools, and much more.

If scientific works in the field of parallel mechanisms can be traced back in the 50’s and 60’s [McGough, Stewart], the first machine-tool based on parallel kinematic chain has been presented very recently [Giddings_Lewis]. From that date, in a very limited period of time, an incredible number of prototypes have been developed worldwide; this has been a tremendous but quite chaotic effort that we could try to briefly outline by presenting the main categories of machines: machines for high-speed handling, classical hexapod (six extensible legs link the base to the spindle), machines with fixed actuators (all the actuators are fixed on the base, and the legs have a constant lenght), hybrid machines (a parallel mechanism carries a serial mechanism or viceversa). This shows that the main problem addressed by parallel machine-tool designers has been the definition of their kinematics structure independently of their final use. Only very few machines have been designed with a customer in mind: most existing machines are more “concept-machines” than “industrial solutions”. As a matter of fact, most machines have not proved completely to date their competitiveness regarding some key features of machine industry. Moreover, most machines and prototypes still rely on custom-made components (joints, drives), and some of them, on in-house designed equipment (dedicated controllers). Clearly, intensive use of non-standardised components cannot lead to low costs solutions. Considering the manufacturing aspect, only two machines have been produced on a “large” scale : the Demaurex Delta, and the Neos Tricept. Consequently, a lot of efforts must be now dedicated to set up manufacturing and assembly organisations to deal with machines based on identical modules rather than different axis.

Looking at the state-of-the-art it becomes clear that for both arm types a change of orientation is only partially possible and that in particular the non-linear behaviour in the path transmissions have a negative influence on the effective machine volume as well as on the rigidity of the machine tool. It is also now necessary to develop new systems for a better accuracy, if parallel kinematics for conventional machine tools will be designed as an alternative. These deficits have to be abolished.

The development of methods for direct and indirect measurement of the tool centre point (TCP) is inevitable if parallel kinematics are to be as accurate as conventional high-speed machine tools. Thus a part of the mentioned deficiencies (caused by temperature, process and acceleration forces as well as by geometric tolerances of the machine components and control influences) can be compensated. If the clear price advantage of hexapods (high number of identical elements) shall become visible, the development of measuring methods has to be realised at low cost.

A next essential objective is the removal of the orientation limitation of hexapods. Thus it would be able to achieve 5 axis machining without angle limitations and maintaining a very high stiffness, on the contrary to the typical serial machines where an additional 2 axis head is assembled reducing the rigidity of the machine.

The non-linear transmission of force and motion of the drives to the TCP represents a considerable problem for the hexapods. This has a direct influence on e.g. the complexity of the hexapod design and dimensioning, the complexity of the forward and inverse model in the NC control, the collision checking and the driving power to be installed (speed and acceleration). Therefore, it must be examined, which basic possibilities for linearising of the non-linear motions exist and how they can be realised consequently within a machine tool concept. The linearising will mean: nearly equal stiffness over the whole work space, extended work space, and extended orientation possibility. By linearising not only the installed driving power is used optimally, but also the utilisation ratio in regard to the machine volume is optimised.

In this context it is important to note that with the new proposed approaches apparently systematic limitations can be removed and so new chances for future concepts are given to the parallel kinematics.

In Parallel kinematics the geometric tolerances of the machine parts lead to position inaccuracies that cannot be accepted. Here algorithms for calibration (based on the direct and indirect measurement of the TCP) have to be developed and evaluated.

Control algorithms that are able to perform a multi-axis control based on the indirect position information of the single axis and the position information of the direct TCP measurement  have to be developed and evaluated in order to significantly rise, about a 30%, the control bandwidth and thus stiffness and accuracy of the axis movements.

A drawback of the parallel devices is the relatively high complexity of controlling the kinematic chain (since even very simple trajectories need an interpolation of more than 2 axes). Algorithms for kinematic transformation, collision detection calibration and jerk-limited set-value generation have to be developed and evaluated in the NC.

Definition of specific methods for the manufacturing and assembling of parallel machinery for the minimal introduction of sources of error during this processes will be done. Furthermore geometric errors have to be measured during a calibration process and considered in the transformation in the numerical control. Therefore the machine accuracy will be improved in a 25%.

Real-time monitoring and compensating systems will be developed and validated in the machine prototypes of the consortium. This way the parallel machines’ accuracy and repetitivity will be another 25% improved.

The actuators and joints represent the main components of parallel kinematics, which mainly determine the machine-properties like accuracy, dynamic performance and costs. The design of those special actuators has not been tackled in detail until now. From measurements of already realised actuators it could be detected, that especially the optimal integration of linear guideways between the inner and outer tube as well as the integration of the measurement system is significant for the actuators and so for the machine accuracy. Further the behaviour of the strut under rotational acceleration has not been studied until now in detail. Since the actuators are one of the mainly cost-driving components of parallel kinematics they should be designed in close relation to this fact. This objective has not been tackled until now so far. Standard actuators and joints are to be designed as well as new manufacturing and assembling methods for them achieving stiffness improvements of 50% and accuracy profits of 60% with regard to the actual designs and prototypes. The standardisation of these components should have repercussions on their cost that will be decreased in a 50%.

The project will develop and regard new drive concept actuators. Possible new drive technologies would be linear direct drives integrated into struts, hollow shaft drives and solenoid-motors integrated into struts. Improvements of a 50% in the static and dynamic stiffness for the sub-assembly strut-drive is expected, as well as a gain of a 40% in the accuracy of the combination.

The optimisation ideas for already used actuators and new developed actuators should be realised and tested on a strut-test-bench, which has to be build up. The strut test-bench should be able to perform positioning tests, positioning test under load, positioning test for different actuator orientations and under rotational acceleration, frequency response tests and thermal analysis.

Development and evaluation of new sensors and methods for direct and indirect measurement of the TCP will be done, optimising their integration in the machine. This way a significant improvement of the 200% in the machine accuracy and repetitivity is expected.

With at least one demonstrator for industrial process into each sector will demonstrate and quantify the improvements managed regarding the partners’ serial machinery that is to be substituted.

With such a lot of factors, new solutions and components it is not easy to find the right machine. Besides there is another key factor for the success of parallel kinematics which is the process of choosing the right kinematic structure and to configure it properly. Due to this complexity of the hexapod design and dimensioning a design system will be established in order to simplify the design process.

Most of the work, which has been done in the field of computer-aided-design for parallel kinematics, is mainly mathematically oriented and deals with simplified geometry. The machine geometry is most often simplified to points, connected by actuators, neglecting the real dimensions of the components. Also the mechanical properties of the components are simplified by unidirectional springs, neglecting e.g. the bending-effects in the struts etc. So existing calculation methods have to be extended, so that it is possible to give not only qualitative results e.g. for the cartesian stiffness of a machine tool but also reliable quantitative results. Furthermore one has to derive advanced optimisation criteria. These new objective functions of the optimisation process must have a close relation to industrial needs, such as position-dependent weighted performance measures.

The fact, that parallel kinematics are determined by much more design parameters than serial arranged kinematics also offers the opportunity, to configure them closer to the requirements of the special application.

The objective is to get an optimal configuration for a given application with predefined standard parts and so could be used as a configurator and re-configurator for special applications. This difficult design process cicle is expected to be about 5 times shortened.

Special attention will be taken into consideration concerning the use of the machines, how to use them, how to obtain the best way of using for the persons who are involved in the processes of using. A lot of tasks and activities are related with the human problem of the inertia to change the way of working and several tasks will tackle these problems with assistance tools and methodologies.

This project carries risks related to the nature of its objectives. The design of a new parallel kinematics machine is far beyond the state-of-the-art practice and it is therefore to be feared that we will hit on a number of unforeseen limiting factors normally irrelevant in today machines. This risk presents a more difficult challenge because it has to do with factors that are inherently difficult to foresee. The only possible answer to manage it consists on the intermediate tests and validation during the first 6 workpackages, whenever this is meaningful and feasible.

The sequential development of solutions and methodologies through the life’s cycle of the components and machines (design, manufacturing, use and end of life) could introduce a certain risk concerning problems or solutions that could be possible from one stage and could be impossible from other. In this case the project will avoid it with a continuous interchanging of information and results with each deliverable (D1-13). Furthermore the workpackage 7 will monitor the congruity of the solutions and problems between workpackages and tasks.

Other important risk, which is especially taken in care, is to avoid particular solutions for specific problems and machines or processes. In this case the workpackage 7 is continuously monitoring for the standardisation of the results and designs as well as the compiling of similar problems, multipurpose and cross-sectoriality.

Besides, the recognised know-how and background of the partners that are the main actors into the parallel kinematics scenario in Europe will surely minimise the risks. In fact, practically in the totality of the existing industrial prototypes or parallel machines in the EU, at least one of the project’s partners is or has been involved in their design or active development.