Performance Analysis and Dexterity Monitoring of Hexapod-Based Simulator
Mohammad Reza Chalak Qazani
1
, Houshyar Asadi
2
, Siamak Pedrammehr
3
, Saeid Nahavandi
4
1,2,3,4
Institute for Intelligent Systems Research and Innovation, IISRI, Deakin University, VIC, Australia
e-mail: m.r.chalakqazani@gmail.com
Abstract—Washout filter is in demand to generate the vehicle
acceleration and angular velocities with high loyalty to make
the best realistic motion. The limitation of the simulator’s
moving platform is an integral part of every motion simulator.
Adaptive and model predictive control washout filters
contribute with this limitation online. It is necessary to have a
real vision to the workspace limitation which is needed for
unpredicted next step of the simulator. In this paper dexterity
analysis is addressed as a novel method to monitor the
performance accuracy of the simulator. The dexterity of both
the classical and optimal washout filters are compared. Finally,
it is shown that the dexterity analysis in optimal washout filter
is necessary to avoid the workspace limitations.
Keywords-parallel-based simulator; dexterity analysis;
classical washout filter; optimal washout filter
I. INTRODUCTION
The regeneration of motion signals from the real vehicle
cannot be modified for the driving simulator not merely
because the real vehicle produces a large acceleration than
the motion platform, but the reachable workspace of the
manipulator is restricted as well. The rotation limitations of
the universal and spherical joints, displacement of prismatic
joints, and the collisions between pods have a direct effect on
the workspace of the platform and reduces the platform
rotation. To overcome this problem, Motion Cueing
Algorithm (MCA) has been announced to produce the
motion signal for the simulator as close as the real vehicle
motion. The emulation of the visual and vestibular
information to the real motion signal is called motion cueing
[1]. MCA is used to prohibit motion sickness and platform
limitation. Colombet et al. [2] showed that the driving
simulator decreases the motion sickness 12 times smaller in
comparison with the static simulator. The standard form of
MCA is known as washout filter and it is sorted into four
categories such as classical, adaptive, optimal and Model
Predictive Control (MPC) washout filters.
Classical washout filter is employed to generate certain
rotational velocities and linear forces for the perception of
the simulator driver that can be sensed in the real car. This is
the common method used in motion cueing algorithms [3].
The first MCA have been proposed by Schmidt and Conrad
as a classical washout filter [4, 5]. Their work has been
improved by Reid and Nahon [6-8]. The human body
perception model is capable of sensing the linear force and
the angular velocity. The outputs of the classical washout
algorithm are sway, surge and heave as linear positions, and
roll, pitch, and yaw as angular positions of the simulator
platform [9]. The tilt coordinate takes advantage of the
somatogravic illusion using the gravity to simulate the low-
frequency acceleration [10-12]. The classical washout filter
consists of the fixed structure without considering of human
perception model such as vestibular system. Kurosaki [13]
used the control techniques to minimize the error. He found
an optimized structure for the washout filter instead of
optimizing the signal. Sivan et al. [14] proposed a new
method considering a vestibular model that was missed in
[13]. Nahon and Reid [6, 15] assumed a center of rotation
using a frame attached to the head of the driver or pilot and
minimized the error based on the cost function. This method
has further been improved by Genetic Algorithm [16, 17]
and considering human sensation model [18-20]. The
significant difference of the optimal washout filter and the
classical one is its higher order transfer function. Since the
generation of the mechanisms, researchers had a struggle
with performance analysis from the design of the mechanism
to the configuration evaluation in a specific workspace area.
Dexterity is of importance as it benefits the performance
analysis of the mechanism [21-25]. Considering the
determinants, the first measurement of dexterity has been
performed by Paul and Stevenson [26]. Yoshikawa [27]
defined the square root of the JJ
T
determinant where J is the
Jacobian matrix of the manipulator.
In this paper, the dexterity analysis is employed to
evaluate the platform configuration. This, moreover, helps to
find the unpredicted next step of the simulator, which would
be useful in the new generations of washout filters. In section
II of the paper, the classical and optimal washout filters are
addressed. The SimMechanic model of the simulator is
generated in section III considering washout filtering, inverse
kinematics, PID controller under MATLAB. The results are
obtained and evaluated through classical and optimal
scenarios in section V. In section VI, the concluding remarks
are presented.
II. CLASSICAL AND OPTIMAL WASHOUT FILTERS
The purpose of classic motion cueing algorithm is the
production of certain linear forces and angular velocity to be
sensed by human perception in the simulator compared to the
personal senses in the real vehicle. Classical washout filter is
composed of four basic strategies. A linear or non-linear
scale factor and limitation which are used to match the signal
with the physical parameters of the simulator and eliminate
the data out of limitation, respectively. The second strategy is
the high-pass filter to remove the low-frequency part of the
linear acceleration signal and transfer the high-frequency as a
226
2018 4th International Conference on Control, Automation and Robotics
978-1-5386-6338-7/18/$31.00 ©2018 IEEE
Authorized licensed use limited to: University of Tasmania. Downloaded on July 10,2024 at 10:12:08 UTC from IEEE Xplore. Restrictions apply.