The previous two sections analysed the pros and cons of isometric versus isotonic devices and position versus rate control and reviewed relevant literature. In relation to 6 DOF control, however, there exists a number of reasons that make it impossible to conclude anything definitively about the four control techniques that were introduced in section 2.1 merely on the basis of the analysis and review, due to the controversy in the literature and to the particular tasks that have been used in the studies. Given that most of the earlier studies have been carried out for 1 or 2 DOF tasks, this difficulty to generalise is especially true for 6 DOF control. Human manipulation of 6 DOF is much more complex and under most circumstances may be relatively slower than 1 and 2 DOF manipulation. This means that the advantages of higher response speeds that one obtains with isometric devices may not be as useful in 6 DOF tasks. Another shortfall of the literature is that most of the studies draw conclusions based on one dimension of interest at a time, while interactions among dimensions are often overlooked.
As discussed in the preceding section, rate control requires paired reversal actions in step tracking. The user has to go through a cycle of start - speedup - maintain velocity - slow down - stop. With an isotonic device, the latter half of the cycle, slow down and stop, has to be executed in such a way that when the cursor is approaching a target, the user has to return the isotonic device to its null position with correct timing. This may be very difficult to do. In pilot studies carried out by the author, it was found that it was very difficult to return precisely to the 6 DOF null position with an isotonic device. It was even more difficult to return to the null position with correct timing. When returned to the null position too early, the cursor would not hit the target (undershoot). When returned to the null position too late, the cursor would overshoot. An improvement to this problem was made by employing a clutch; once the clutch is released, the input becomes zero (null).
With an isometric device, on the other hand, the self-centring scheme will automatically bring the control action to zero once the human releases muscular tension. This means that part of the control task in rate control with isometric devices is performed automatically by the device itself.
In position control mode, the self-centring effect with isometric devices does not work as an advantage, since position control normally requires control movement in only one direction. Instead, for such cases, the user has to overcome the self-centring force with isometric devices to maintain position. This may not only make it very difficult for the user to maintain output accurately, but can also cause fatigue. In fact, a pilot study by the author showed that it was almost impossible to overcome the self-returning force and perform steady control with a 6 DOF isometric device. In order to have a practical isometric position control technique, a clutch had to be added to engage and disengage the control actions so that the user could do the position control in steps.
What the above analysis indicates is that one should expect an interaction between device resistance (isotonic versus isometric) and transfer function (position versus rate). Isometric devices, in other words, are more compatible with rate control and isotonic devices are more compatible with position control.
With the two compatible modes, i.e. isotonic position and isometric rate control, the former should be easier to learn, due to the presumed simpler mental processing in position control, which should simply be a 1-to-1 (or 1-to-K) mapping in forming control actions. The latter may thus impose a higher mental load on the user in forming the rate control actions, even though part of the work (returning to zero) is facilitated by the self-centring force of isometric devices.
To test these hypotheses, a 6 DOF docking experiment*
was conducted with two modes of device resistance (isotonic vs.
isometric device) and two modes of transfer function (position
vs. rate control). All four of these input techniques, namely
isotonic position, isotonic rate, isometric position and isometric
rate, had 6 degrees of freedom.