IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART A: SYSTEMS AND HUMANS 1
Two Mitigation Strategies for Motion System
Limits in Driving and Flight Simulators
Chris W. Schwarz, Member, IEEE
Abstract—Limited workspace is a challenge for all motion-
based simulators, whether they are large excursion systems like
the National Advanced Driving Simulator or smaller simulators
utilizing only Stewart Platforms. Two approaches for addressing
this challenge are nonlinear washout scaling and software dis-
placement limiting. This paper presents new algorithms devel-
oped for these approaches. The nonlinear scaling method uses
the cubic Hermite interpolation polynomial to smooth the cor-
ner in the scaled output at the limit. A software displacement
limiting method that generates control signals in the table frame
of reference is introduced. As a result, unwanted acceleration
artifacts caused by unbalanced limiting of actuators are avoided.
The methods are described, and offline simulation results using the
new displacement limiting method are presented.
Index Terms—Displacement control, limiting, motion compen-
sation, motion control.
I. I NTRODUCTION
D
RIVING and flight simulators are commonly used today
for research and training. All motion-based simulators
have in common that they attempt to reproduce vehicle dynam-
ics within a very limited workspace. Thus, onset accelerations
can be reproduced quite accurately, but sustained ones cannot.
Motion limits on position, velocity, and acceleration are typi-
cally enforced to prevent damage to the machine. Preferably,
the motion drive algorithm (MDA), i.e., washout never lets the
motion base reach a limit; however, it does happen in practice.
A software-limiting scheme can mitigate these occurrences by
gently slowing the offending actuator down and holding it under
the stroke limit. Actuator limiting can produce unacceptable
false motion cues to the driver. As a result, simulator operators
often have to be extremely conservative when setting the MDA
parameters so that actuator limiting is never encountered.
The goal of the methods described herein is to provide ways
of mitigating the adverse effects of hitting motion limits. If this
can be done successfully, it may allow the simulator operator to
relax their constraints on the motion, for example, by opening
up the MDA to provide larger onset cues. As a general rule,
motion limits should still be avoided; however, an increase
in performance could be realized if the operator was able to
tune the system for the average driver/pilot rather than the
Manuscript received October 27, 2005; revised September 4, 2006. This
paper was recommended by Associate Editor L. Rothrock.
The author is with the National Advanced Driving Simulator, Iowa City,
IA 52242 USA (e-mail: cschwarz@nads-sc.uiowa.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSMCA.2007.897590
worst case. Then, when the worst case is realized, the resulting
motion commands will be dealt with in ways that do not cause
unacceptable false cues.
The simplest method of constraining the motion commands
in the MDA block is to provide scaling factors for each degree
of freedom (DOF) and clip the signal if a specified acceleration
or velocity limit is reached [1], [3]. Encountering this type
of limit will cause a sharp jerk and wipe out all cues above
the limit. A variation on this approach is to replace the clipping
limit with a linear scale region of reduced slope so that the
scaled output grows more slowly than the input [3]. A newer
polynomial-based nonlinear scaling method was introduced
in [3]. The third-degree polynomial used there is similar to the
Hermite approach, but the parameter set is more complicated,
and there are no guidelines for wisely selecting the parameters
of the polynomial. In the worst case, the scaled output can expe-
rience local minima when the input is monotonic. The proposed
nonlinear scaling method addresses these shortcomings.
The second strategy specifically regards the workspace limits
of the Stewart Platform. These occur if the stroke limit of
an actuator is reached, if two actuator legs collide, or if a
joint reaches a mechanical limit. A well-designed hexapod for
simulator application will reach the first type of limit before
ever reaching the next two. A review of Stewart Platforms was
compiled in [5].
An actuator-limiting method is described in [2] for the Uni-
versity of Toronto Institute for Aerospace Studies (UTIAS)
flight simulator. The method computes a distance from the
actuator limit at which deceleration must begin to remain below
a specified level. The actuator comes to a stop as the virtual ac-
tuator position is tracked beyond the limit and back into the al-
lowable region. Then, the actuator accelerates and again tracks
the position command. This algorithm effectively softens an
encounter with a limit but may still create a noticeable false cue.
Rubio et al. proposed a controller that avoids workspace
limits and singularities with the use of virtual springs [4]. Theirs
is a force–force controller and requires the consideration of the
Stewart Platform dynamics. As well, control occurs completely
in actuator space and therefore could still produce artifacts in
the simulator reference frame.
The kinematic mapping from simulator positions to actuator
positions is quite simple to do; however, the inverse problem is
more complex, which requires the solution to be either a high-
order polynomial or several simultaneous nonlinear equations.
Dieudonne et al. introduced an iterative method to solve the
inverse kinematic equations in real time [6]. We have chosen
here to avoid the standard iterative technique in favor of using
1083-4427/$25.00 © 2007 IEEE