Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

HUMANCONTROLOFTHESKATEBOARD*

MOW HKRBARD
Department of Mechanical Engineering. University of California, Davis, CA 95616, U.S.A

Abstract Human control of the skateboard is investigated by modeling the rider as a single. rigid bed)
pinned 10 rhe board along the roll axis. Human input is taken to be a torque applied at the ankles. The
equations of the nonholonomic rider-board system are presented and a simple tracking task is established
11s dynamics are augmented by those of the skateboard and rider to describe the complete system in the
tracking mode Several control schemes are discussed. Under certain conditions, simple proportional
feedbach conrrol of rider tilt angle can stabilize roll motion of the vehicle, but in the most general case full
stale feedbach 1s required. In the complete state feedback case, a performance index for the tracking task IS
defined and the minimizing feedback gains determined. Time simulations of the tracking task using the
optimal feedback gains are shown. Experimental results are presented which tend to validate the theory.

I\THODL Cl.lOS made by the USCPSC of many other factors related to


skateboard injuries, one conclusion drawn by Ruther-
The explosive rebirth of the skateboard industry ford, et al. (1978) was, ‘that balance or the ability to
during the last several years has madeitselffelt in many balance in a constantly changing environment at
areas. The economic impact of more than 20 million various speeds is the most critical task for the user’.
riders in the United States alone has been substantial. This so-called balance (or sometimes dynamic bal-
More important, probably, has been the role of the ance) is actually a time varying control applied to the
skateboard in the development of the skateboard vehicle by the rider and is the subject of this paper.
subculture, the rise of skateboard art with perfor- Many references to the human as a controller in a
mances akin to dance and gymnastics, and, inevitably, loop containing a dynamic system have appeared.
the increase in the number of injuries as a result of Originating with the work of James, Nichols and
skateboard accidents. Phillips (1947). Tustin (1947) and Wiener (1950 t. and
Undeniably much of the appeal of the vehicle among continuing mostly in the area of modeling human pilot
youthful riders comes from the risk associated with its dynamics (see, for example, McRuer, er al. (1965)).
use as documented by Hayes (1974). Yet other more quasi-linear transfer functions of the human operator
positive factors must also play a role. A sense of were developed. More recently, with the maturation of
freedom is experienced by the rider who is in complete the vector-space. time domain approach to control
control of a dynamic, evolving trajectory which re- theory, a quantitative model for the response charac-
quires substantiall! less power flow from the human teristics of the human operator tracking a dynamical
than running, but which takes place at speeds high system was developed by Kleinman, Baron and Le-
enough to evoke feelmgs near those of flight. The vison (1970) based on optimal control theory.
challenge lies in the nontrivial task of continuously Other investigators have focused on the human
controlling an inherently unstable vehicle, whose
unstable modes have time constants considerabl)
smaller than those of. say, the bicycle.
200
r
Unfortunately these high speeds and small unstable
time constants can cause catastrophic accidents when
control is lost for a short period. The growth of
skateboard related accidents and their medical ramifi-
cations ha\e been documented in the literature (Shu-
man. 1967; Atienza and Sia. 1976: Jacobs and Keller,
1977 ; Rutherford. YI (II.. 1978).
According to the U S. Consumer Prodact Safety
Commission (USCPSC), skateboard injuries in-
creased by a factor of 38 from 1973 to 1977. Shown in
13 74 7s 76 ii 78
FIN. 1 are injury statistics for the period 1973 through
1978. Although a detailed qualitative analysis was
Fig. I. USCPSC estimated number of skateboard Injuries
treated in hospital emergency rooms in the U S (note
logarithmic scale)

745
746 MOhT HUBBARD

operator in control tasks more closely related to that


studied in what follows. Donaldson (1964) assumed
various transfer function forms and experimentally
determined their parameters using error decorrelation
techniques while the human operator balanced a stick
pinned to a wheeled cart. van Lunteren and Stassen
(1969) performed a similar experimental investigation
of human operator characteristics while stabilizing a
laboratory bicycle model. Morawski (1973) described
the control exercised by a skier in a turn, modeling the
skier’s body as an inverted pendulum and assuming
the time varying sideward control force resulting from
ski side-slip was determined by a negative feedback
mechanism. Hemami and Wyman (1979) have recently
investigated human bipedal locomotion as a feedback
control process.
One previous study has been made by Hubbard
(1979) of the natural lateral behavior of the skateboard
in the absence of rider control. Two simple models
were examined. One major finding was that the speed
of the vehicle is an important parameter and that the
passive vehicle-rider system has certain critical speeds.
These critical speeds divide the speed axis into regions Fig. 2. Side and top views of skateboard showing board
where the uncontrolled vehicle is stable or unstable. A geometry and xyr coordinate system.

second conclusion was that all real vehicles require


some degree of rider control for stability.
In this paper the dynamical equations describing the respective nonhorizontal pivot axes (see Fig. 2), thus
rider controlled skateboard system in a tracking task causing a steering angle of the wheels whenever the
are presented. An integral quadratic performance axles are not parallel to the plane of the board.
index for the task is assumed and the optimal (mi- Since a tilt angle y of the board generates a positive
nimizing) feedback gains are computed from the front steering angle 6, and a negative rear steering
solution of a steady-state matrix Riccati equation. angle 6,, there is a point P on the line between the two
Time simulations of the system using the optimal axle centers which has zero lateral velocity and hence
feedback gains are presented and compared to expe- only forward speed u (see Fig. 2). When the board
rimentally measured time histories of an actual rider moves on a horizontal plane, it can be shown that this
on a skateboard. speed u must be constant (see Hubbard, 1979). Point P
is a distance a6,/(6/ + a,) from the front axle. The
equations of motion are written using an xyz coor-
SYSTEM MODEL

The typical skateboard consists of three com-


ponents: the board, a set of two trucks, and four
wheels. Schematic side and top views of the vehicle are
shown in Fig. 2. The board is generally from 0.5 to
2.0 m long, 0.15 to 0.25 m wide and 1 to 2 cm thick, the
bending stiffness varying considerably depending on
the application. When more maneuverability is desired
a flexible board is used, while stability at high speeds
calls for more stiffness.
The wheels, usually made from urethane which
tracks well over rough pavement, are mounted on the
axles on ball bearings to minimize friction. They are
stiff enough that the deformation due to lateral loads is
small, and hence the slip angle (angle between the
velocity vector of the wheel and the plane of the wheel)
is negligible.
The distinctive components of the skateboard which
allow it to be steered are the trucks which connect the
axles to the board. Angular motion of both the front
and rear axles is constrained to be about their Fig. 3. Rear view of skateboard-rider system
Human control of the skateboard 747

dinate system whose origin is at P. The x axis is in the


direction of motion, the y axis is to the right in the Y

horizontal plane, and z is down, as shown in Fig. 2.


Shown in Fig. 3 is a rear view of a skateboard and
rider combination. The board and rider are assumed to
have separate roll degrees of freedom, which are
specified by the two coordinates, y and tp, respectively.
The rider, modeled as a single rigid body, is assumed to
be connected to the board with a pin along the x
(forward) axis and to interact with the board via an
internally generated torque at the ankles, T. In
addition there is a torsional spring which exerts a
restoring torque between the wheelset and the board
proportional to the tilt y of the board relative to the
wheelset.
The vehicle is steered by making use of the static --
X
relationship between steering angle and tilt of the
board. /
y=c,b, (I)
where c, = cot I.,, and 1., is the angle which the front
pivot axis makes with the horizontal (see Figs. 2 and 3).
Similarly c, = cot i.,.
Now consider an inertial XY coordinate system in
Fig. 4. Top view of skateboard moving relative to an
the ground plane as shown in Fig. 4. Three more
inertialty fixed X Y coordinate system m the horizontal plane.
generalized coordinates (X. Y. and B), which specify
both the position of point P and the orientation of the
line connecting the axle centers relative to the Y axis, After writing the exact nonlinear equations, assum-
are required to completely describe the motion. The ing small angle approximations for both y and 4, and
system is recognized as nonholonomic since the nonin- neglecting second and higher order terms, the three
tegrable constraint equation linear differential equations of motion for the system
become
X = Y tan 6 (2)
Y = 0. (4)
must be satisfied. With thegiven steering angles 6, and
6, the vehicle will execute a constant radius turn with (mh’ + Ib)ji + (m,a + m,B)huY
the yaw angular velocity of the vehicle centerline + (k, - mgh + mhul/oc)? + m,h@ = T, (5)
relative to an inertially fixed line given by
and
m,hfj;+ m,alu$ + (m,lu*/ac)r
(3)
+ (m,l* + I,)4 - m,gl$ = - T (6)
where
where the total mass m = mb + tn, and the two
c = c/c,I(c, + c,) dimensionless parameters a = (a - d) ‘acl - d;ac, and
We assume that the board center of mass (cm.) is /3 = (c, - cl)/2c,c,. Since equation (4) is of little
located midway between the two trucks, but for interest (it merely confirms that the forward speed is
generality the cm. of the rider is assumed not to be constant) it will not be dealt with further here. Equa-
located above the board c.m., but instead over the tions (5) and (6) are two coupled linear second order
centerline of the axles a distance d behind the front differential equations in the variables y and C#Jwith
axle. constant coefficients which depend only on board and
Having stated assumptions and defined coor- rider parameters, geometry and speed u. Although (5)
dinates, the equations of motion can now be written. and (6) describe small motions exactly, it is possible to
Since the two constraint equations (2) and (3) allow the considerably simplify the equations before proceeding
derivatives X and 4 to be solved for explicitly in terms further. The board mass and moment of inertia are
of the generalized coordinates, the system is termed generally considerably smaller than those of the rider.
simple nonholonomic and the genera1 procedure set The terms involving the board height h in (5) and (6)
forth in Kane (1968) can be used to obtain the motion arise due to the fact that the point at which the rider is
equations for the three remaining generalized coor- pinned to the board translates a small amount due to
dinates (Y, y and 4). Details of the calculations are the rotation y, Since h is generally much smaller than
omitted here for brevity. The reader can refer to the other two characteristic lengths in the model (a and
Hubbard (1979) for more detail. 0, the terms involving hare small and may be neglected
as well. When h, mb and I, are set to zero (5) and (15) which the human may employ is to make the angles 4
become and y small (near zero) on the average, that is to
stabilize the vehicle.
k k,y = T
Further performance goals may be desired as well.
(7)
For example, a more quantitative characterization of
the point-to-point transportation task might be: go
from point B toward point E, all the while remaining
tnrx/u$+ $y + (ml’ + I,)4 - mgl4 = - T. (8)
reasonably close to the line connecting them. Without
loss of generality the initial point B may be chosen to
Thus, when board height, mass and moment of be the origin of the X Y coordinate system in Fig. 4,
inertia are neglected, equation (5) (a dr$erenrti/ equa- with point E on the positive Y axis. Thus the differen-
tion where T was viewed as an input and y an output) tial equation which describes the angle between the
becomes algebraic in T and y. This algebraic equation vehicle centerline and the desired reference line is
simply states that in the absence of inertia forces and equation (3) with the desired angle Bd = 0. The
damping, the torque k,y produced by the angular position deviation of the vehicle c.m. is then given by
deflection of the board restoring spring must be large the solution to
enough to exactly balance the applied ankle torque T.
X = u sin 0. (IO)
This means that T and y are no longer independent
variables. Given one, the other is immediately de- with the desired deviation X, = 0. Whenever 0 K I (as
termined from the constant k,. Hence, either can be it will be in a successful tracking situation) equation
viewed as the control quantity in this case. Henceforth, (10) can be linearized to
we shall refer to y as the control. Equation (7) is then
X = uI3. (11)
used to eliminate T from (8) which describes the
remaining dynamics. The uncontrolled system is im- The complete system model may now be concisely
mediately recognized as unstable since 4 and 4 have written in first order vector-matrix form by including
coefficients of opposite sign in (8). equations (3), (8), (11) and the identity $I = 4. These
After using (7) in (8) and taking the Laplace become
transform of the result, the transfer function relating 4
to y is readily calculated to be $x=Fx+Gy, (12)

where the four-dimensional state vector is given by


4(s) malus + mlu’/ac + k,
-= - (9)
(ml* + I&* - mgl 2 = [4, & 6, Xl’
Y(S)
and the constant matrices F and G are
The dimensionless parameter a is a measure of
longitudinal position of the rider, being negative when r O 1 0 0’

the rider is behind point P (see Fig. 2) and zero at point md


____ 0 0 0
P In the former case (a < 0) examination of (9) reveals F = (mP+I,)
that the simple linear feedback control law 7 = C&
fails to stabilize the vehicle, whereas full state feedback
of the form 3 = C&J + C&J succeeds. This motivates
1 0
0 0 0
u 0

and
the full state feedback approach to control law for-
mulation taken in what follows. Although not re- mlu’+ack; T
quired, the value of a is henceforth taken for simplicity
to be zero, and the rider is assumed to be positioned
G=
F O-
(ml* + I,)oc
u/at 0
I
at point P.

HUMAN CONTROL

A SIMPLE TRACKING TASK


Any feedback control system has four components:
the system or plant to be controlled, an array of sensors
The skateboard is frequently used in wildly gyrating to provide information about the outputs of the plant,
motions, which require a high level ofcontrol expertise a control law or logic which calculates the correct
on the part of the rider. For the beginner however, and inputs based on the plant outputs, and an actuator
even for the expert when simple transportation is which translates the calculated plant inputs into
desired, the main object is to keep the body upright physical torques, forces, etc. The equations of the plant
and traveling more or less forward. are summarized in (12).
The dynamical equations for the vehicle and rider For the human in the tracking task defined above,
without control have been shown to be unstable. the sensor array consists of the vestibular mechanism,
Hence one essential feature of any control scheme containing in each ear three semicircular canals and
a
Human control of the skateboard 749

otolith organs, and the visual system. It has been shown E

(see, for example, Young and Meiry, 1965) that the


vestibular system is a rate information channel. The FIMI
Relerence ’
semicircular canals measure input angular velocities Line 2 9, 1

over a wide range of input frequencies, and similar


results have been obtained for the otoliths vis a vis
linear velocities. The visual system derives position
information and perhaps some rate information from B
the visual field. Grunwald and Merhav (1976) have
presented a basic model for this process formulated in
an optimal control framework. lnlllai
Reference
Kleinman, et al. (1970), showed that this portion of Lme 1
the human operator can be modeled well as an optimal
estimator-predictor which yields the statistically best
estimate of the current state, taking into account the Fig. 5. Top view of a turn which can be viewed as an initial
condition problem for the closed loop system at point B
observation noise and sensory time delays. This model
was found to be appropriate for even a relatively
complex control task, a fourth order plant which was
slightly unstable (see Baron, Kleinman and Levison,
1970). Thus it will be assumed in what follows that the
rider has available complete and exact knowledge of
the state. The assumption of exactness (as opposed to
+ (X/X,)’ + (vir.Jzldr (17)
an estimate with uncertainty) is not particularly
significant since, as shown in Wonham (1968), the
and the feedback gains are completely specified by the
separation theorem ensures that the optimal control in
choices for the five constants #J,, 4,. 6,, X,, y., Using
the noisy case is the same as that for the case when the
this formulation, changes in the optimal closed-loop
knowledge is exact. In addition, actuator dynamics
system as a function of changes in the performance
are neglected in what follows.
index weighting parameters are easily studied.
Let a scalar performance index for the regulation
Since the control law (14) is linear in the state, the
task be defined as
model for the closed-loop system becomes linear as
well. The response of the controlled system to initial
J = Z(ZT AZ + by’)dr, conditions, disturbances, etc., can be studied by solv-
(13)
c
-0 ing the linear homogeneous equation

an integral of quadratic forms in the state and the


control, where A is a positive semi-definite matrix and J! = (F + GC):, X(0) = X0. (18)
b is a positive scalar. This measure of performance As an example, turn negotiation from one straight
penalizes (roughly speaking) a weighted sum of the line segment to another can be viewed as an initial
mean square values of the elements of the state and the condition problem. This is illustrated in Fig. 5. Assume
control. the vehicle is exactly balanced and tracking the initial
It is a well-known result in control theory (see, for reference line 1. On reaching the corner at B. the final
example, Bryson and Ho, 1969) that the optimal reference line 2 becomes the new desired trajectory and
control (that which minimizes the performance index the initial condition on the state vector abruptly shifts
(13)) is given by from (O,O,O,O)’ to (0,0,f7,,0)T.

r(t) = C%(t) (14)

where the constant feedback gain matrix C is calcu- A NUMERICAL EXAVPLE

lated from
For purposes of illustration. a particular rider-
C= -GTSfb (15) skateboard system was analyzed. Shown in Table I are
the inertia and geometric parameters of [he vehicle and
and where S is the only positive definite solution to the
matrix Riccati equation
Table I. Skateboard and ruder parameter\ for numertcal
SF + FTS + A - SGB- ‘G’S = 0. (16) stud!
~_____.
It is often reasonable to choose A diagonal with m 77 bg a 0.57 m
elements aii = l/x$ where xai is the maximum I 0.925 m C 0 75
1, 12.2 kgm’ h 50 N m rdd
allowable value ofthe ith element of the state vector. In
a 0.0 I‘ 2 0 m vx
this case (13) becomes -__-..
750
Moryr l-iUBBARD

Fig. 6. Locus of closed-loop system roots as a function of the allowable control ya

rider, which were chosen to be near those of an actual From the preceeding it is clear that the control task
test subject. can be subdivided into two portions; tracking (0 and _r)
Equation (16) was solved by the method ofeigenvec- and balancing (4 and 4). Ahhough neither is neglected
tor decomposition using a computer program adapted entirely by a controller designed using linear-
from the one described in Bryson and Hall (1971). quadratic methods, it is of interest to see the variation
Table 2 gives six different sets of state and control in the optimal closed-loop system as the emphasis
weightings and the corresponding feedback gains and changes from one to the other. Specifically, the perfor-
closed loop system eigenvalues. The effect is shown of mance index can be written as
decreasing the allowable position deviation X,. All the
feedback gains gradually become larger with the
largest increase in Cx (from 0.50 to 3.16). In addition
the dominant (smaller) closed loop system eigenvalue
J=
Iom{(#/dd’ -I- (&d,.,’+ cCCW~.J’

is increased by nearly a factor of three indicating a + WXJ2] + (hJ*W (19)


faster closed loop response. The closed loop roots and the parameter Eallowed to be greater or less than 1
eventually become two complex conjugate pairs. as the tracking task is given more or less emphasis.
A similar shift in the closed-loop system eigenvalues Figure 7 shows the locus of closed loop eigenvalues for
occurs as a function of changing the allowable control 0.01 < E < 100. All other weights are those of case I in
y,,. Shown in Fig. 6 is the locus of roots in the complex Table 2. As in the previously discussed case of
plane as the allowable control y,,is varied from 0.01 rad decreased allowable control, when the tracking task is
to 1 rad. All other weights were the same as case 2 in deemphasized (E < 1) the tracking roots become do-
Table 2. As expected, when more control effort is minant and approach the origin. Hence the tracking
allowed the system becomes faster and more re- performance becomes sluggish.
sponsive. When y. is very small (O.Ol), the slow roots In several of the above cases large values of feedback
correspond to almost purely tracking (6,X) whereas gains and closed-loop roots were produced for certain
the faster roots are purely balance (&d). AS ye sets ofweighting constants. For example, E = 10 in Fig.
approaches zero the two pairs of closed-loop roots 7 yields a root at s = - 46.8 set- ‘. While such results
approach - 2.99 set-’ and the origin, respectively. are mathematically meaningful, they are in no way
Thus when the control is restricted, it is the tracking descriptive of a control law which a human could
subtask which suffers. implement, since the human bandwidth is of the order

Table 2. Optimal feedback gains and closed-loop eigenvalues for various weight&

Performance index weights Feedback gains C in Closed-loop eigenvalues


Case Cp. & 0, .L Y. y=cx s, I= 1.4
bad) bad) (m) (rad) C, c, CB Cx (XC - ’ )

1 0.1 0.32 0.1 0.2 0.1 1.32 2.45 1.89 0.50 -5.59. -3.14, -2.26, -1.06
2 0.1 0.32 0.1 0.1 0.1 8.97 3.00 2.13 1.00 -5.37, -3.16, -2.13+0.63j
3 0.1 0.32 0.1 0.07 0.1 10.29 3.44 3.41 1.41 -5.04, -3.18, -2.56kO.91,
4 0.1 0.32 0.1 0.06 0.1 11.27 3.11 3.93 1.73 -4.56, -3.22, -2.97kO.991
5 0.1 0.32 0.1 0.05 0.1 12.09 4.05 4.36 2.00 -3.41?0.17/, -3.62+1.1Oj
6 0.1 0.32 0.1 0.03 0.1 15.55 5.20 6.20 3.16 -3.04+0.15j, -4.62k2.66,
-___ ____-
Human control of the skateboard 751

i -1

Fig. 7. Locus ofclosed-loop system roots as a function of changes in the relative weight Eof the tracking and
balancing subtasks.

of 6-12 rad set-’ (1 to 2 Hz). Thus care must be taken two complex conjugate pairs, with dominant time
in the interpretation of the optimal human controllers constant of roughly 0.5 sec. Hence the decay of the
generated in this fashion. system states to zero is relatively fast, the position error
To serve as an illustration of the actual time decaying to near zero in about 1Ssec.
response of the closed-loop controllers discussed pre- Figure 9 shows the time response ofanother optimal
viously, time simulations were run which generated closed-loop system (with eigenvalues - 1.56 t_ .i 1.17,
solutions to (18) for a variety of initial conditions and - 6.00, - 1.82s~~~~’ generated from weights 4, =
for three different control laws generated from different 0.1 rad, 4. = 0.1 rad/sec, 0. = 0.1 rad, X, = 0.1 m, y,
weighting matrices in (13). These results are shown in = 0.05 rad) to an initial position error X,, = 0.1 m.
Figs. 8-10. In each case $, X and y are plotted vs time. Here two of the closed-loop eigenvalues are real, the
The two remaining state variables can be inferred in dominant time constant is roughly 0.65 set and the
each plot by graphical differentiation. decay of the system states to zero is somewhat slower
Figure 8 shows the closed loop time response for than in Fig. 8. The initial control action is to turn to the
case 6 in Table 2, for an initial rider tilt error do = right (y>O), away from the reference line. Although
0.02 rad. Note that initially the board tilt angle y is this, at first glance, may seem nonoptimal, the con-
positive, producing a right turn to move the board troller is merely unbalancing the rider to the left (4
under the rider. This produces both heading (8) and < 0) so that it can then begin a left turn and approach
position (X) deviations which must then be corrected the reference line. From the positivity of all feedback
for. Note that the closed-loop eigenvalues consist of gains in Table 2, this same behavior can be seen to be
true for isolated Qinitial conditions. The initial control
action is to turn even further away from the line,
I ozc n I increasing 0 momentarily before the eventual cor-
rective action can be taken.
Such a time simulation ISshown in Fig. IO, for case 1
in Table 2 with an initial error 8, = 0.1 rad. Here all
four eigenvalues are real and the dominant time
constant is roughly 1.0 sec. Thus the system response
is slower still.

ESPERIMESTAL RESULTS

In order to validate the preceding theory. an


experimental study was made of an actual
rider-skateboard system negotiating a turn as de-
scribed in Fig. 5 with 0, = 0.1 rad (5.4 deg). A 16 mm
movie camera (nominal frame speed = 18 frames/set)
Fig 8. Fast closed-loop system response to initial rider tilt was positioned over the final reference line at the
error with feedback gains C = [tS.55, 5.20, 6.19. 3.161. height of the rider c.m. and the angles f/ and ;’ were
752

p ;:
- 0 --0
-8 #

Fig. 9. Slower closed-loop system response to initial transverse position error with feedback gams C = [6.20,
2.13, 1.53, O.SO].

measured from the film. Position deviation X was had a measured forward velocity, u = 1.79 m/set. The
measured from a ground trace left on the concrete floor closed-loop rider-board system appears to have a
by chalk attached under the board center. speed of response somewhere between those shown
The test subject was an experienced professional in Figs. 9 and 10. Furthermore, the qualitative re-
skateboard rider, the reigning world downhill cham- semblance of the experimentally measured X. C#J and 7
pion at the time. The subject was instructed to actuate to those of the theoretical turn shown in Fig. JO is very
with the ankles only, maintaining the rest of the body good. Initially rider tilt $I is negative but it returns to
as rigid as possible, and to track the final reference line zero at t z 1.3 set and then overshoots slightly. The
as closely as possible. No detailed quantitative specifi- board tilt y, after the brief initial positive impulse
cation of the tracking task was made however, as for discussed previously, follows a similar path becoming
example including desired weighting factors, etc. Rider negative and then returning to zero and overshooting
and board parameters were as listed in Table 1 except slightly. The position error X is always positive
for the speed which varied slightly from test to test. reaching a peak at t 5 0.4sec and then decaying to
Although the experimental study included many near zero after roughly two seconds.
factors, only a nominal set of results are presented here. It should be possible, using sophisticated system
An experimental trajectory is shown in Fig. 11 which identification techniques, to determine the exact feed-

Fig. 10. Slow closed-loop system response to initial heading error (equivalently. system dynamic response
for turning through small angle) with feedback gains C = [7.32, 2.45. 1.89. 0 503
Human control of the skateboard 753

Bryson, A. E. and Ho. Y. C. (1969) Applied Optimal Conrrol,


Blaisdell, Waltham, Mass., 167.
Bryson, A. E. and Hall, W. E. (1971) Optimal Control and
Filter Synthesis by Eigenvector Decomposition. SUDAAR
436, Stanford University.
Donaldson, P. E. K. (1964) Error Decorrelation Studies on a
Human Operator Performing a Balancing Task. Med.
electron. biol. Engng. 2, 393-410.
Grunwald, A. J. and Merhav, S. J. (1976) Vehicular Control
by Visual Field Cues - Analytical Model and ExperImental
Validation, IEEE Trans. Sysr.. Man.. Cybemer SMC-4,
835-845.
Hayes, D. (1974) Risk Factors in Sport, Hum. Facrors 15,
454-458.
Hemami, H. and Wyman, B. F. (1979) Modeling and Control
of Constrained Dynamic Systems with Applicallon IO
Biped Locomotion in the Frontal Plane, IEEE Trans.
Automat. Contr. AC-N. 526-535.
Fig. 11. Experimentally measured vehicle turning response. Hubbard, M. (1979) Lateral Dynamics and Stabihty of the
Skateboard, J. appl. Mech. 46, 931-936.
Jacobs, R. A. and Keller, E. L. (1977) Skateboard Accidents,
Pediatrics 59, 939-942.
back gains which the rider actually implemented, using James, H. M., Nichols. N. B. and Phillips, R. S. (1947 i Theor!,
ojSeruomechanisms, McGraw-Hill, New York
experimental data from a system trajectory such as
Kane, T. R. (1968) Dynamics, Holt, Rinehart. and Wmsron,
shown in Fig. 11. Such research is now in progress New York.
using the technique of quasifinearization described by Kleinman, D. L., Baron, S. and Levison. W. H. (1970) An
Bellman and Kafaba (1965). But even though the exact Optimal Control Model of Human Response. Parr I:
feedback gains are unknown, the strong qualitative Theory and Validation, Auromarica 6, 357-369
McRuer, D. T.. Graham, D., Krendel, E. S. and Relsener. W.
agreement between theory and experiment shown in (1965) Human Pilot Dynamics m Compensator! Systems:
Figs. 10 and 11 supports the notion of a human Theory, Models and Experiments with Conrrolled-element
optimal controf in the skateboard tracking task. and Forcing Function Variations, AFFDL-TR-6% 15.
Morawskl, J. A. (1973) Conlrol Systems Approach to a Ski-
turn Analysis J. Biomechanics 6, 267-279.
CONCLUSION Rutherford, G. W., Friedman, J. I.. Beale. S. P. and Brown. V.
R. (1978) Hazard Analysis - Injuries Associated with
A model has been presented for the dynamics of a Skateboards, U S. Consumer Product Safety Commission.
skateboard and rider in a simple tracking task. A linear Washington.
state-feedback control law was derived by assuming Shuman, S. H. (1967) Skateboard InJurIes In a Campus
Community, C/in. Pediar. 6, 252.
the human attempts to minimize an integral of quad-
Tustin,A. (1947) An Investigation ofthe Operaror’s Response
ratic forms in the state and the control; that is, to in Manual Control and its Implications for C’onrroller
obtain maximum mean square performance with Design. J. Inst. ejecr. Engnrs 94, 190-202
minimum mean square control. The effects of tradeoffs van Lunteren, A. and Stassen, H. G. (1969) In\estigatlon of
in the specification of the performance index on the !he Characteristics of a Human Operator Stablhzing a
Bicycle Model, Occup. Sajer!, HIrh 14, 349-369
optimal controller were investigated and time simu- Young, L. R. and Meiry, J. L. (1965) Manual Conlrol of an
lations of these different controllers were presented. Unstable System with Visual and Motion Cues. IEEE Inr.
Experimental results were shown which strongly sup Corn... Rec.,.Part 6, 123-127.
port the hypothesis that the human functions as an Wiener. N. (1950) Exrrapolation. Inrerpolarum and Smoothiq
o/Srarionarx Time Series. Wiley. New York
optimal controller in the skateboard tracking and
Wonham, W. M. (1968) On the Separation Theorem of
balancing task. Stochastic Control. SI_4M J. Conrrol 6, 312~3%

Acknowledgemenrs-The author gratefully acknowledges the


assistance of J. Hutson, P. Marshall and S. Glass in the
gathering and processing of the experimental data.
NOMENCLATURE

a wheelbase
A state weighting matrix in performance Index
REFERENCES b control weighting parameter In performance
index
Atlenza, F. and Sia, C. (1976) The Hazards of Skateboard- B.E endpoints of desired trajectory
Ridmg, Pediarrics 59, 939-942. constant relating board till angle to steering
Baron, S., Kleinman, D. L. and Levison, W. H. (1970) An angle
Optimal Control Model of Human Response, Part II: C feedback gains
Prediction of Human Performance in a Complex Task, d distance of rider c.m. behind front axle
Auromarica 6, 371-383. F system dynamics matrix
Bellman, R. and Kalaba, R. (1965) Quasilinearizarion and G control distribution matrix
Non-Linear Boundary-Value Problems, Elsevier, New, h board cm height above axles
York mass moment of inertia
754 MONT HUBBARD

performance index of control task weighting parameter for tracking subtask


k torsional spring stiffness rider tilt angle
height of rider cm. above board pivot axis angle from horizontal
m mass yaw angle of board centerline
P point on line connecting axle centers with zero
lateral velocity
Subscripts
Laplace variable
Riccati matrix a allowable
time b board
speed desired
position coordinates in system attached to board /” front
system state vector 0 initial
inertial coordinates of board center r rear, rider
dimensionless measure of longitudinal rider t turn
position
B dimensionless truck asymmetry
Superscripts
Y board tilt angle
6 steering angle T matrix transpose

You might also like