Professional Documents
Culture Documents
I. INTRODUCTION
LEGGED VEHICLE possesses a tremendous potential
for maneuverability overrough terrain, particularly in
comparison to conventional wheeled or tracked vehicles [ 11.
In general, a legged vehicle can offer more degrees-offreedom for movementthan conventional vehicles. Legged
vehicles can provide the capabilities of stepping over obstacles
or ditches, climbing over obstacles, or maneuveringwithin
confined areas of space [ 2 ] , [3].
However, the coordination of the movements of the various
leg joints in such a way as to produce the desired locomotion
of the vehicle is an extremely complex task. Previous studies
have shown that if the leg coordination is left entirely to the
human operator, evena relatively simple walkingmachine
presents such a highly complex task that the operator becomes
exhausted after only a short period of operation [4]. Therefore, it is essential to relieve the operator of as much of this
complex task as possible.
Fig. 1. Hexapodvehicles at the Ohio State University. (a) The OSU
At The Ohio State University, research is being conducted
Hexapod, walkingin
dual tripod mode. (b)Model
of the adaptive
suspension vehicle (ASV).
in several areas of legged locomotioh. A major development
of this research is the OSU Hexapod vehicle (Fig. l(a)). This
six-legged vehicle is an experimental prototypewhich is being evaluation of newsensors and sensing systems. Eachof the six
used to develop various control schemes and leg placement legs of this vehicle is comprised of three independent rotary
algorithms and serves as a test-bed for the development and
joints arranged in an arthropod configuration. The vehicle is
interfaced to a PDP-11/70 computer via an optically isolated
ManuscriptreceivedMarch
7, 1985. This workwassupported
by the
digital-data
link [5].
Defense Advanced Research Projects Agency under contract DAAE07-84-KPresently under construction is a new vehicle referred to as
ROO1.
D. A. Messuri iswithPackard
Electric Division of General Motors the adaptive suspension vehicle (ASV). A preliminary model
Corporation, P.O. Box 431, Warren, OH 44486, USA.
C. A. Klein is with the Department
of Electrical Engineering, The Ohio of this vehicle is shown inFig. l(b). This vehicle will be afullscale self-contained walking machine. Each of the six legs of
State University, 2015 Neil Avenue, Columbus, OH 43210, USA.
0882-4967/85/0900-0132$01.00 0 1985IEEE
133
a general wave gait, the dual tripod scheme does not use the
maximum possible number of supporting legs for agiven
speed, but since the vehicle must already be designedso that it
can be supported by three legs, this does not pose a problem
for close maneuvering mode. The dual tripod algorithm has
been implemented on the OSU Hexapod and simulated on the
ASV.Other recent workdealing with this modehas been
performed by Lee [191.
Precision Footing: In situations involving very irregular
terrain, the operator may want to control individual legs and
body motion with a joystick, keyboard, or other means. The
precision footing mode is, by definition, veryoperator
intensive andmaneuvering the vehicle with this type of control
modecouldbe
anextremelycomplex
task. To make this
control mode useful, it is essential that the precision-footing
computer control algorithm includefeatures to aid the operator
as much as possible without greatly restricting the freedom of
movement inherent to this mode.
This paperis primarily concernedwith the precision-footing
mode of operationand, in particular, the development of
automaticbody
regulation schemes that allowautomatic
movement of the vehicle body in order to aid the operator in
maneuvering
the
vehicle. Section I1 describes how the
operator would use the precision footing mode to control the
vehicle. Since this mode would be used on irregular terrain
where the operator i s concerned with the vehicle tipping over,
a measure of stability is very important and will be discussed
in Section 111. This measure, together with a consideration of
constraints on kinematic limits of individual legs, leads to two
new control schemes described in Section IV.
11. PRECISION-FOOTING
MODE
The precision-footing operational mode can provide maximum maneuverability, particularly for complex tasks such as
climbing over large obstacles or crossing ditches. However,
the control algorithm should provide
the operator asmuch help
as possible in order to alleviate some of the burden of
manipulating the body and limbs.
A specific computer control algorithm has been developed
to implement the precision-footing operational mode [ 181. A
variety of features have been incorporated into this algorithm
to helpsimplify
the operators control task,to
provide
necessary information to the operator, and to assure safety.
For theOSUHexapod the vehicle operator can issue commands to the algorithm by using a three-axis joystick and a
selected set of keys on the computer terminal keyboard.When
construction of the ASV is completed, the operator control
mechanism will consist of a custom-designed arm controller
and a set of function-select switches [9]. Regardless of the
hardware interface, the algorithm functions are the same.
By choosing one of a set of six switches, the operator can
select the desired foot to be moved. The operator canthen use
the joystick to command the desired velocities of the foot in the
longitudinal, lateral, and vertical directions. Foot movement
can be simplified for the operator by theuseofJacobian
control and resolved motion rate control [20]. This allows the
operator to specify rectilinear velocities of the foot rather than
specifying actuator velocities.
134
IEEE JOURNAL OF ROBOTICS AND AUTOMATION, VOL. RA-I, NO. 3, SEPTEMBER 1985
111. ENERGY
STABILITY
MARGIN
Fig. 2.
Graphics displayprovidesessentialguidanceinformation
operator of a walking machine vehicle.
to the
A critical foot is definedas a foot which, if lifted, would cause the body
to
be statically unstable.
135
Fig. 3 . Vehicle standing on an inclined plane with the body horizontal. The
projection of support feet into a horizontal
planedefines the support
pattern while the curve connecting the tipsof the support feet (shown in
dotted lines) defines the support boundary.
136
IEEE JOURNAL OF ROBOTICS AND AUTOMATION, VOL. RA-1, NO. 3, SEPTEMBER 1985
PE = mgh
(1)
COS
!I?.
137
'
I'
#
1-36
IO
FORE-AFT
36 I
(IN)
(0
1-36
IO
FORE-AFT
(IN)
m
rn
1-36
FORE-AFT
(IN)
I
36 I
IO
FORE-AFT
A-
31
(IN)
(C)
Fig. 6 . Optimal paths of the vehicle centerof gravity, for variousvehicle
configurationsforseveraldifferentstartingpoints.The
dotted lines
represent level energy curves. (a) Vehicle standing on level terrain, body
horizontal. (b) Vehicle standing on a20" inclined plane, body horizontal.
(c) Vehicle standingon a 20" inclined plane,body horizontal, left front leg
off the ground.
Fig. 7. Optimal pathsof the vehicle centerof gravity, for various vehicle
on
configurations for several different starting points. (a) Vehicle standing
a 20" inclined plane,body horizontal, with right front legand left rear leg
on rocks. (b) Vehicle standing on a 20" inclined plane, body pitched at
20". (c) Vehicle standing on level terrain, body horizontal, with a tripod
support phase.
138
:
FORE-AFT
( I N)
Fig. 8. Tracing the optimal paths of the vehicle center of gravity from four different starting points, when a blending function is not
included. Note the oscillations, which indicate sharp direction changes in the optimal path, due to the discrete time calculation of
the energy stability margin gradient.
IV. AUTOMATIC
BODYREGULATION
Maneuvering a vehicle overrough
terrain canbean
extremely complextask. The features in the precision footing
control algorithm which have been discussed thus far simplify
the operators control task, provide feedback information,and
assure safety. To enhance the capabilities of the precision
footing operational mode, the concept of automatic body
regulation was developed whereby the operator can allow the
computer control algorithm to automatically adjust the position of the vehicle body in accordance with some predefined
criteria. Two automatic body regulation schemes were developed and have been
incorporated into the computer control
algorithm. The two schemes are
referred to as body accommodation and body stabilization.
A . Body Accommodation
As explainedin Section 11, the precision-footing control
algorithm enablesthe vehicle operator toselect and control the
motion of individual legs of the vehicle. The algorithm also
allows the operator to directly control the body motion in its
six degrees-of-freedom.These features make the vehicle
highly maneuverable for extremely rough terrain situations.
However, eachvehicle leg has a limited reach, due to the legs
kinematic limits. This limited reach may sometimes require
the operator to perform increased maneuvering in order to
place a foot at a desired foothold. In order to alleviate the
operator of some ofthis maneuvering task, a body accommodation feature was incorporated into the control algorithm.
Inthe precision footing control algorithm,whenever the
operator selects a vehicle leg for individual leg control, the
position of that leg is monitored to insure it is never extended
beyond the kinematic limits. With the body accommodation
scheme, if this individually controlled leg reaches the kinematic limits, thenthe vehicle body is automatically commanded to move in such a direction as to accommodate the
operators desired motion of that individual leg. This accom-
139
B. Body Stabilization
While traversing a region of extremely rough terrain, the
vehicle operator may find that, due to the terrain conditions,
the vehicle body and legs have become oriented into a rather
precarious configuration. The feedback information provided
via the cockpits graphics display (as discussed in Section 11)
indicates such things as the critical support feet which cannot
be lifted, the support polygon, and the energy stability level
for eachedge ofthe supportboundary.Furthermore,
the
control algorithm includes safeguards to keep all legs within
kinematic limits and to maintain static stability. If the display
indicates that the present vehicle situation has a low energy
stability margin, the operator may desire to shift the body to a
positionof
greater stability beforeproceeding
with leg
maneuvers. This repositioning task is simplified by incorporating a body stabilization feature into the control algorithm.
The body stabilization feature is activated when theoperator
chooses the appropriate function-select switch. The body
stabilization scheme
determines
the optimal
path
to an
optimally stable position, based
upon
the current body
orientation and leg positions. The vehicle bodyisthen
automatically shifted, with the bodymovement
restricted
within the present plane of the body, to the point where the
body center of gravity coincides with the optimallystable
position. After body stabilization is completed, the operator
can proceed with whatever maneuvers are desired. The body
stabilization feature can also beusefulwhen
the operator
desires to position the vehicle at an optimally stable position
before attempting to maneuver across some obstacles. Since
the body stabilization routine is completely automatic when
activated, it permits the vehicle stability to be optimized
quickly and easily, for any body orientation and leg positions.
It should be noted that there may be occasions where the
body orientation and leg positions are such that the leg
kinematic limits prohibit thebodymovementnecessary
to
140
IEEE JOURNAL OF ROBOTICS AND AUTOMATION, VOL. RA-I, NO. 3, SEPTEMBER 1985
The operation of the body stabilization routine is demonstrated in Fig. 9. Fig. 9(a) shows the OSU Hexapod in a rather
precarious configuration. where the vehicle has climbed over
an obstacle and
has
been
maneuvered intoan
awkward
position. The graphics display information corresponding to
this situation is shown in Fig. 9(b). The small square inside the
support polygon represents the location of the body center of
gravity. The X symbol, which appears when the body
stabilization routine is activated, indicates the optimally stable
position of the center of gravity. Fig. 9(c) shows the results
after invoking the body stabilization routine. The graphics
display information indicates that the body center of gravity is
nowat the optimally stable position as evidenced by the
location of the present center of gravity and the magnitudes of
the energy stability levels.
The general formulation of the equation for calculating the
energy stability margin allows an interesting extension. Since
the OSU Hexapod vehicle is equipped with force sensors on
each foot, this force information can beused to actively
compute the location of the vehicle center of gravity. This
active center of gravity would take into account any effects of
vehicle cargo loading, changes in fuel level, etc. The energy
stability margin could be calculated using the active center of
gravity. The result is improved vehicle stability since any
variation in the vehicle center of gravity could now be
compensated.
V. CONCLUSION
A computer control algorithm has been developed for the
precision-footing operational mode. This algorithm includes
the incorporation of a body accommodation feature and a body
stabilization feature. These automatic body regulation schemes
allow greater vehicle maneuverability, particularly during
rough-terrain locomotion.
One of the major developments presented in this paper is the
concept of energy stability margin. A general equation was
introduced which allows the calculation of the energy stability
margin for any given position of the bodyand legs. By
utilizing the gradient of this function, an optimal path can be
found leading from a given initial location of the body center
of gravity to an optimally stable position. A blending function
was introduced to reduce the oscillation which can occur when
the optimal path approaches a ridge on the energy surface.
The energy stability margin provides an accurate quantitative measure of vehicle stability, particularly for rough-terrain
conditions. Previously implemented stability criteria provided
a qualitative measure of stability, but did not fully account for
rough-terrain conditions. It is this quantitative measure,
provided by the energy stability margin, which allowed the
development of a computer algorithm for determining an
optimally stable position. This capability was incorporated
into the precision footing algorithm to achieve the body
stabilization feature.
One of the particularly interesting future applications for
these concepts and the automatic body regulation schemes
presented here wouldbein
the development of free-gait
algorithms [ 131. For example, the body accommodation
feature would provide a wider selection of allowable foot-
141
VEHICLE
[22] R. B. McGheeandA.
A.Frank, On the stability of quadruped
creeping gaits, Mathematical Biosciences, vol. 3, no. 3, pp.331351, Oct. 1968.
[23] A. A. Frank, On the stability of an algorithmic biped locomotion
machine, J. Terramechanics, vol. 8, no. 1, pp. 41-50, 1971.
the OhioStateUniversity.Heiscurrentlyemployed
Engineer in the Advanced EngineiringDepartment
Division.
Dr. Messuri isa member of Tau Beta Pi.
as a SeniorProject
of PackardElectric