3

Locomotion Control for Legged Robot

by Virtual Contact Impedance Method
Fumiaki Takemori
Tottori University
Japan

1. Introduction
In the current infrastructure, the accessible environment is not necessarily maintained, so
that it seems to be hardest for user of wheelchair (invalid chair) such as the aged and
the leg-handicapped to travel in the inaccessible environment. When the user of standard
wheelchair travels up/down the stairs, accompany supporter is usually needed. Therefore a
single action is resultingly restricted within narrow limits. From these reasons, it is desirable to
develop a mobile robot carrying the human that enables to independently travel inaccessible
environment.
In the mobile hardware with the locomotive function for the stairs, the wheeled and crawler
type are major (iBOT , 2010), (SUNWA , 2010). Although the wheeled type is simple and
high efcient for actuating method, because of not so redundancy, degree of freedom on
mobility is not so high. In addition, wheel type tends to be large size due to its multi-wheel
and crawler, so that it is difcult to operate in universal inhabited space. In terms of more
useful and intelligence mobility, it is necessary for the robot itself to take avoidance action
automatically in order to not damage the robot itself and circumferential obstacles. So, to
realize an appropriate method for avoiding the obstacle and stairs is very important issue.
In this paper, a legged mobile robot carrying the human is developed considering the
adaptiveness for inaccessible environment at existing infrastructure. Although the legged
robot is complex as mobile-morphology, its redundancy with multi-links is advantage to
high mobility in universal space. Though earlier researches (Wu, Y. & Higuchi, M. , 2004),
(Sugahara et al. , 2004), have already been reported, these are not so advanced instance that
stair detection and its avoiding method are integrated. So, as a method for detecting and
avoiding the stairs/obstacles, the virtual impedance control method (Tsuji & Kaneko , 1999)
is proposed. This method can avoid an obstacle without contacting by receiving virtual force
from virtual spring and damper installed in virtual surface region. Finally, the mobility of the
legged robot by non-contacting impedance control is presented through some experiments.

2. Legged robot with linear actuators

2.1 Outline of three-legged robot

In this paper, a legged mobile robot which has the ability to carry the human is developed. The
photograph of prototype model is shown in Fig.1. This robot has three legs. Each leg consists
of three linear actuators, so that total degree of freedom is 9. As shown in Fig.1 and Fig.2,

www.intechopen.com

42

Intelligent Mechatronics

to maintain the rigidity, three actuators are assigned to triangular conguration and farther
the part of so-called shin has twin screw axle mechanism that are synchronously actuated by
single DC-motor. Main specication of proposed legged robot is presented in Table 1.
Width
Height
Weight
DC motor
Power supply
Sensors
Control computer
I/O

600[mm]
710910[mm]
35[kg]
9ch (150[W]14[W])
7.2[v] battery 5
3-axis accelerometer : 3ch
potentiometer : 9ch
sole load sensor : 12ch
PSD : 2ch
Intel(R) Pentium(R) M 1.6GHz
AD : 32ch / DA : 16ch

Table 1. Specication

Linear
actuator

Fig. 1. Human carrying robot

2.2 Inverse kinematics

A local servo control in terms of linear actuator length is generally convenient to implement
whole robot control algorithm. Then it is need to formulate the relationship between each
link length and a specic position at free leg and/or at the seat. So, in this section, inverse
kinematics for legged robot with linear actuated links is described.
As illustrated in Fig.3, dene the heel point of grounded sole is the origin in sagittal plane and

www.intechopen.com

43

Locomotion Control for Legged Robot by Virtual Contact Impedance Method

DC motor

screw axle

screw axle

Fig. 2. Twin screw mechanism

also dene pt : = [ xt , zt , t ] T is the vector of free-legs tip of toe and p w : = [ xw , zw ] T is the
vector of seat position. Farther, in Fig.3, F, W are sole and seat length, L is base length of each
link and w is short offset, which are all xed.
g g g
At rst, the approach to calculating link length r g : = [r1 , r2 , r3 ] T of the grounded leg is given.
Based on geometrical structure,

g
r1 =
x2w + z2w
(1)

g
r2 =
( xw F )2 + z2w
(2)
can be immediately calculated. From the vector equation
g

L + w + r 3 = W,
g

the following procedure in terms of r3 is given:

g

r 3 2

(W L w ) T (W L w )

W 2 + L20 2W T ( L + w )

W 2 + L20 2W ( L cos w sin )

where
= tan 1
As a result we obtain
g

r3 =

www.intechopen.com

zw
.
F xw

W 2 + L20 2W ( L cos w sin ).

(3)

(4)

44

Intelligent Mechatronics
f

Next, link length r f : = [r1 , r2 , r3 ] T of free leg also can be given by similar approach. The
followings are resultingly obtained:

f
( p t F p w ) T ( pt F p w )
(5)
r1 =

( xt F cos t xw )2 + (zt F sin t zw )2
=

f
r2 =
( x w x t )2 + ( z w z t )2
(6)

f
W 2 + L20 2W ( L cos w sin )
(7)
r3 =
where

zt zw
.
(8)
xt xw
According to above proceeding, if the kinematics in terms of positional vector is represented
by

f w (r g )
pw
= f (r g , r f ) =
,

pt
f t (r g , r f )
= tan1

then we dene the inverse kinematics explicitly solving all links length as

1
f
rg
g ( pw )

= f 1 ( p , p ) =
.

w t

1
rf
f f ( pw , pt )

3. Control strategy for advanced mobility

3.1 Non-contact impedance method

In the case where mobile robot is used in public facility and the home environment, in
order to not damage the robot itself and environmental objects, it is necessary to take into
account for interactive force between end-effector and circumstance. Impedance control is
often used for such control method that makes robots end-effector have desired force and
realizes mechanical compliance.
In a general way, the realization of impedance control necessitates a certain contact with
external objects. For safetys sake, however, it is desirable for mobile robot to avoid obstacles
without contacting external objects. From these reasons, this paper proposes a method of
avoiding obstacles by non-contact impedance control.
The concept of non-contact impedance control is illustrated in Fig.4. Assume that virtual
contact force occurs via virtual spring and damper, if an obstacle trespasses on virtual surface
postulated at the tip of a toe. Assume : = [ x , z ] T to be trespassing vector on virtual surface,
virtual contact force is given by
T
J (r f )( D v + K v ) ( < 0)
(9)
uv =
0
( 0)

www.intechopen.com

45

Locomotion Control for Legged Robot by Virtual Contact Impedance Method

pw ( xw , z w)

r3g

L0

ph

r3f

L
L

w
w

r1
g
1

r2g

r2f

pt ( xt , zt )

pt
F

Fig. 3. Linear actuated legs conguration

here, K v , D v are virtual spring and viscosity coefcient matrix respectively.
Jacobian-matrix dened by
t

f 1 f 1t f 1t
f
f
f
r
r2
r3

1
J (r f ) = t
t.
t
f2 f2 f2
f
f
f
r1
r2
r3

J (r f ) is

(10)

In actual robot control, virtual surface can be formed in realtime by measuring the distance
between obstacle and end-effector using position sensitive detector (PSD).
3.2 Control method

The ideal task of mobile robot is to realize intelligent behavior robustly even if planed
reference path is obstructed by irregular pitch steps or wall projection. As mentioned above
section, our control strategy combines tracking control for planed reference path with force
control by virtual contact impedance.
The proposed control method, as shown in Fig.5, consists of three control law Eq.(11), which
are reference trajectory control ut , virtual contact control uv and sole counter force control u s .
u = ut + us + uv

(11)

u t is determined by servo control loop for reference trajectories in terms of links length which
are precedently given to realize a walking pattern. u s is used to compensate the gap between
the sole and the oor if the robot does not poise owing to the transition of center of mass,
external force, etc.

www.intechopen.com

46

Intelligent Mechatronics

virtual surface

robot's leg
virtual contact force

x
spring-damper

K v Dv

obstacle

z
Fig. 4. Concept of non-contact impedance control

Reference
Reference
generator
generator

pd

f 1(p )

rd

ut

PID
PID

Inverse kinematics
Sole forces

Sole
Solecounter
counter
force
forcecontrol
control

Virtual
Virtual
contact
contact

us

+
+

J T (r )

Legged
Legged
Robot
Robot

uv

f (r )
Kinematics

Fig. 5. Control system

phase1
Fig. 6. Walking sequence

www.intechopen.com

phase2

phase3

phase4

Locomotion Control for Legged Robot by Virtual Contact Impedance Method

47

3.3 Walking pattern

In order to make the mobile robot proceed to a certain direction, some locomotion pattern
is needed beforehand. Then it is useful to generate corresponding trajectories along the
locomotion pattern in realtime and to realize feedback control for such trajectories.
A human carrying robot proposed in this paper must be able to move three legs individually
in consideration of high mobility and stability. This paper considers the gait in which a
single leg travels for free-leg and the other two legs keep on standing as fundamental walking
pattern. Since such gait can posture robots center of mass inside polygonal area formed by
ground soles, static walk becomes to practicable. Each phase in proposed walking pattern, as
illustrated in Fig. 6, is briey explained as follows:
phase-1 : As swing up center leg, detect the edge of step based on the degree of change of
PSD value. At the same time compute the height of step by downward PSD.
phase-2 : Give a reference elliptical orbit in order that for free sole to land on the step. Then
give a horizontal orbit for seat position in keeping current height.
phase-3 : Swing left leg up to step height (which has already been given at phase-1. And
give a reference elliptical orbit in order that for its sole to land on the step.
phase-4 : For right leg, give same operation as phase-3.

rv
PSD(1ch)

PSD(2ch)

Fig. 7. Location of PSD

4. Experiment
4.1 Walking up stair

An experiment to avoid a stair installed in front of the robot is tried. In this experiment, while
the distance between robot and stair can be measured by PSD in real time and the radius of
virtual surface is given as rv = 5 [cm], notice that the height of stair is unknown to human.
Fig. 7 shows the location of 2ch PSD assigned for virtical and horizontal measures.
Fig. 8 presents sequential motion captures in walking up a stair, where a man with about
50[kg] weight is embarking on the seat. Indeed, because the robot travels in constantly
standing by two legs, it is conrmed that the behavior of the robot loading a human is stable.
Fig. 9(b) shows trajectories and virtual force at tip of toe, where dashed line means reference
trajectory planned at phase-1, phase-2 in walking pattern. For comparison, the trajectory
without virtual contact control is also shown in Fig. 9(a). As shown in Fig. 9, it is found

www.intechopen.com

48

Intelligent Mechatronics

that as tip of toe approaches edge of stair closer, virtual force increases by virtual contact
efciency. Consequently it is conrmed that proposed virtual contact method can avoid the
obstacle even if reference trajectory which approaches obstacle closely is applied.

10

Fig. 8. Sequential motion captures

0.2

pt
p td

0.2

pt
ptd

0.16

fv

0.16

Z [m]

0.12

Z [m]

0.12

0.08

0.08

stair

0.04

0
0

0.1

0.2

0.3

0.04

0.4

0.5

0
0

0.1

0.2

X [m]

0.3

0.4

0.5

X [m]

(a) without Virtual contact control

(b) with Virtual contact control

Fig. 9. Trajectory and virtual force at tip of toe

4.2 Soft landing

The efciency that alleviate the robots impact in landing, i.e. soft-landing is expected by
applying a variable gain to control law
u = u t + ( I )u v + us (0 < i < 1) .

(12)

Then, as second experiment, we verify the efciency of soft-landing by measuring the degree
of pitch in landing phase. Fig. 10(a) shows vertical acceleration and velocity at the seat when
= I, while Fig. 10(b) shows same datum when virtical contact control is given by
= diag{0.15, 0.25}

www.intechopen.com

49

Locomotion Control for Legged Robot by Virtual Contact Impedance Method

in trial and error. Obviously the degree of pitch controlled by virtual contact is more
suppressed than the case not controlled. As a result, it is expected that mechanical landing
impact and the fearfulness for the passenger are alleviated.
Acceleration [m/s2]

0.5

Az
0.5
0

10

11

Velocity [m/s]

0.1
0.05
V

0
0.05
0.1
0

5
6
Time [s]

10

11

10

11

(a) without Virtual contact control

Acceleration [m/s2]

0.5
Az

0.5
0

0.1
Velocity [m/s]

Vz
0.05
0
0.05
0.1
0

5
6
Time [s]

10

11

(b) with Virtual contact control

Fig. 10. Degree of pitch at the seat

5. Conclusion
In this paper, we have developed three-legged mobile robot which has the ability to carry
the human by using has 9 linear actuated links. Each leg consists of three linear actuators
assigned to triangular conguration in order to maintain the rigidity. For such linear structure,
the inverse kinematics has been given. The inverse kinematics for local servo control is also

www.intechopen.com

50

Intelligent Mechatronics

given. As a method for avoiding the obstacles, we have proposed the virtual impedance
control method which can avoid the stairs without contacting by receiving virtual force
from virtual spring and damper installed in virtual surface region. Finally, the mobility for
avoiding unknown height step and soft-landing motion have been conrmed through some
experiments.

6. References
http://www.ibotnow.com
http://www.sunwa-jp.co.jp/en/
Sugahara, Y. et al. (2004). Realization of Dynamic Human-Carrying Walking by a Biped
Locomotor, Proceedings of the IEEE ICRA 2004, pp.3055-3060.
Tsuji, T. & Kaneko, M. (1999). Noncontact Impedance Control for Redundant Manipulators,
IEEE Trans. on Systems, Man, and Cybernetics-Part A: Systems and Humans, Vol.29, No.2,
pp.184-193.
L., Villani et al. (2000). An Experimental Study of Adaptive Force/Position Control
Algorithms for an Industrial Robot, IEEE Trans. on Control Systems Technology, Vol.8,
No.5, pp777-786.
Wu, Y. & Higuchi, M. (2004). Development of a Power Assisting System of a Walking Chair,
Proceedings of 2004 IEEE/RSJ Int. Conf. on IROS, pp.3207-3212.

www.intechopen.com

Intelligent Mechatronics

Edited by Prof. Ganesh Naik

ISBN 978-953-307-300-2
Hard cover, 248 pages
Publisher InTech

Published online 28, February, 2011

Published in print edition February, 2011

This book is intended for both mechanical and electronics engineers (researchers and graduate students) who
wish to get some training in smart electronics devices embedded in mechanical systems. The book is partly a
textbook and partly a monograph. It is a textbook as it provides a focused interdisciplinary experience for
undergraduates that encompass important elements from traditional courses as well as contemporary
developments in Mechtronics. It is simultaneously a monograph because it presents several new results and
ideas and further developments and explanation of existing algorithms which are brought together and
published in the book for the first time.

How to reference

In order to correctly reference this scholarly work, feel free to copy and paste the following:
Fumiaki Takemori (2011). Locomotion Control for Legged Robot by Virtual Contact Impedance Method,
Intelligent Mechatronics, Prof. Ganesh Naik (Ed.), ISBN: 978-953-307-300-2, InTech, Available from:
http://www.intechopen.com/books/intelligent-mechatronics/locomotion-control-for-legged-robot-by-virtualcontact-impedance-method

InTech Europe

University Campus STeP Ri

Slavka Krautzeka 83/A
51000 Rijeka, Croatia
Phone: +385 (51) 770 447
Fax: +385 (51) 686 166
www.intechopen.com

InTech China

Unit 405, Office Block, Hotel Equatorial Shanghai

No.65, Yan An Road (West), Shanghai, 200040, China
Phone: +86-21-62489820
Fax: +86-21-62489821