Parallel mechanisms have several advantages. Depending on the exact configuration the load can often be purely axial, and is always distributed through the legs. The structures have a high degree of inherent rigidity, and as any errors generated in the legs are averaged, these mechanisms tend to be more accurate. Furthermore, the speed of displacement is often greater as the motors are generally positioned on the fixed base. On the other hand, calculation of the available workspace can be difficult and generally results in quite a complex shaped volume (figure 2).

Within this workspace forbidden areas can exist, known as singularities. These are particular positions where the end actuator will either lose its rigidity entirely and hence fall over, or the forces in the legs will increase to infinity and result in breaking part of the robot or locking up. Finally, due to the complex mathematical relationship between the leg length and the position of the end effect, control has been a problem in the past. Modern computer systems have eliminated this disadvantage.

Historically, the design and manufacture of parallel robots have remained within the domain of academia or high-tech companies, who have had a need for a solution to unique problems, such as aircraft simulation. However, as a result of lower cost and high-speed computing availability (easing the control of such mechanisms), industry has started to make in-roads into areas traditionally restricted to standard mechanism configurations, such as high- precision placement required for silicon chip manufacture. This higher volume market means that a number of systems are now commercially available.

It is the variable actuator length Stewart platforms (also known as Hexapods or Gough platform after the original designer) that have had the highest degree of exposure. They now represent the top end of the market both in degrees of freedom (offering 6) and in terms of cost. They are marketed by such companies as Physik Instrumente [1] offering the M-850, Micos [2] building the PAROS (figure 3) and others, who offer machines capable of positioning loads in excess of 50 kg to a resolution better than 2μm and 5μrad.

Figure 3: Stewart platform showing tilt of the top plate.

The majority of requirements will often not need the full six degrees of freedom offered by the above machines, and unfortunately it is this market that is currently under-developed. The large number of possible combinations of degrees of freedom has meant that unless a specific application exists, little has actually been constructed beyond theoretical models, although a number of interesting smaller units are available from such companies as New Focus [3].
What is the ILL's position regarding parallel robots? The ILL is constructing a neutron Strain Imager as part of the Millennium Programme and in collaboration with the University of Manchester (partly funded by an EPSRC grant). Its sample positioning system will consist of a full six degrees of freedom Stewart platform, capable of positioning loads up to 500 kg to better than 50μm.
The initial configuration was designed jointly by the ILL and INRIA Sophia-Antipolis. This is now in the process of being turned into a detailed 3D model by OHE [4] (see figure 4).

The basic features will be six hydraulically driven actuators, each with a stroke of 470 mm and a working pressure of up to 160 bar. A linear transducer with a resolution of 0.1 μm will be positioned within each cylinder and used to determine the leg length. Control will be via a controller-board with a 500 MHz CPU with a cycle-time of 4 kHz. These actuators will be fixed to the top and bottom plates using universal joints. The speed of operation and workspace will be restricted, with two or three possible configurations within the total volume available. These will hopefully be singularity free, although until the final configuration is realised it is not possible to calculate whether this will be the case. If not, then position restrictions will be incorporated into the driver software.
What is the future? There is no danger of the disappearance of traditional serial robots - their simplicity of control and use will always safeguard their place in the market. But certainly the Strain Imager Stewart platform represents a leading edge technology with regard to load capacity and positioning precision. In general, the advantages provided by parallel type architectures will ensure an increase in their use.