The INS tilt error is caused by the accelerometer error, for example, a milli-g accelerometer bias will induce 1 mrad error for roll and pitch estimates.
The curves in Figure 5 show that the horizontal accelerometer biases are hardly estimable for that they cannot be fully separated from the tilt errors. However, the down accelerometer bias is more readily estimated since the INS is almost leveled (within [+ or -]2[degrees]) relative to the local horizontal.
Figure 4 shows the three instrument errors: [[epsilon].sub.[rho]] = distance error, [[epsilon].sub.[alpha]] = orientation error and [[epsilon].sub.[gamma]] = tilt error
. The neighbourhood of such errors is a portion of the sphere, highlighted in black in Figure 4, which for simplicity can be approximated to a parallelepiped.
* A high stiffness and load capacity to minimize axial, radial, and tilt error
Linear and tilt errors are obtained by this method.
Tilt errors are not measured by gage bore probing, only linear errors
latencies to make the direction judgments, the relative and absolute rotation errors, and the relative and absolute tilt errors) by averaging in each case to provide a single score for: two-floors-up, one-floor-up, the same-floor, one-floor-down and two-floors-down, which comprised the mean of four observations for two-floors-up and -down, the mean of eight observations for one-floor-up and -down, and the mean of 12 observations for the same-floor).
The mean relative tilt errors are illustrated in Fig 4.
With flexible coordinate transformation, sensor and actuator axes need not be parallel to each other and rotation can be added to a single module XY-motion system, or tilt errors
can be reduced for motion accuracies in the sub-nanometer range.