Sei sulla pagina 1di 12

J Intell Robot Syst (2006) 46: 151162

DOI 10.1007/s10846-006-9056-2

Flatness Conservation in the n-trailer System Equipped


with a Sliding Kingpin Mechanism
V. Deligiannis G. Davrazos S. Manesis
T. Arampatzis

Received: 9 June 2004 / Accepted: 3 May 2006 /


Published online: 3 August 2006
Springer Science + Business Media B.V. 2006

Abstract Nonlinear systems, which are differentially flat, have several properties
that can be useful on designing effective controllers. In this paper we show that
the n-trailer system equipped with a sliding kingpin mechanism is a differentially
flat system, like its non-sliding kingpin counter part. The sliding kingpin technique
is used to eliminate the undesired deviation of the path of each intermediate vehicle
from that of the leading one (off-tracking phenomenon). The linearizing outputs of
the flat system are the Cartesian coordinates of the middle of the last semi-trailers
axle. The state space and the kinematic equations of the new modified system are
derived and the conditions for flatness are examined. The flatness conservation is
also checked relatively to several kinds of dynamic sliding feedback control.
Key words flatness multi-articulated vehicles n-trailer system sliding kingpin

1. Introduction
A major problem in automatic control is driving a system from one state to another.
This is the motion planning problem in mobile robotics. For systems described as
differentially flat, motion planning has simple and specific solutions [1]. This arises
from the fact that every system variable is described as a function of the flat output
and a finite number of its time derivatives [2]. Additionally, knowledge of the systems
flat output allows the design of open loop control and assists the design of control
loops [3, 4]. Because the behaviour of differentially flat systems is determined by the
linearizing outputs, we can design trajectories in output space and then calculate the
appropriate inputs. The concept of differential flatness was originally introduced by
Fliess et al. [5], in the context of differential algebra and later using LieBcklund
transformations [6]. It is known that many mechanical systems are differentially flat,

V. Deligiannis G. Davrazos S. Manesis (B) T. Arampatzis


Electrical and Computer Engineering Department, University of Patras, Patras 26500, Greece
e-mail: Stam.manesis@ee.upatras.gr

152

J Intell Robot Syst (2006) 46: 151162

such as, the n-trailer system [3], the Bi-steerable car [4], the hopping robot, the planar
rigid body chain and the towed cable system [6], pendulums in series, the heavy chain
[7] or even the fed-batch bioreactor [8].
The n-trailer systems or multi-articulated vehicles can be found in two different
research fields: Autonomous robotics and transportation systems. In autonomous
robotics the main aim is to build mobile multi-body robots that accomplish useful
tasks without human intervention while operating in unknown environments. On the
other hand, in intelligent transportation systems the goal is to construct long and
multi-articulated transportation vehicles, known as a hard platooning system, which
is intelligent enough to be driven with as less human intervention as possible. One
basic problem, in both above areas, is the undesired excess in motion due to the offtracking phenomenon. Off-tracking is defined as the deviation of the semi-trailers
axles or the kingpin hitch from the path of the steering axle of the leading vehicle.
The off-tracking deviation can be eliminated using a sliding kingpin mechanism, a
technique introduced in [9] and various types of sliding control have been examined
in related works [10, 11].
As mentioned above, the n-trailer system is a differentially flat system, whose flat
(or linearizing) outputs are the Cartesian coordinates of the middle point of last
trailers rear axle. In [2], Rouchon et al. show that mobile robots with trailers are
flat systems as soon as the trailers are hitched to the middle point of the axle of the
previous ones. The placement of the kingpin outside the axle makes the system not
flat. The kingpin sliding mechanism allows the kingpin to slide along the axle of the
previous trailer. A natural question, from a steering and control theory perspective,
is whether the system remains flat or not. In this paper we prove that the n-trailer
system equipped with a sliding kingpin mechanism is a differentially flat system.
The flatness property of a system is more significant in the robotics field than the
transportation one, because it is related to various problems such as path and motion
planing, path following, steering etc. A representative example among a large number
of similar works can be found in [12]. Fliess et al. [2, 5] have shown that chainedform systems are a special case of differentially flat systems, where the bottoms of
the chains play the role of flat outputs. The advantages of the chained form are that
many methods are available for the open-loop steering of such systems as well as for
point-stabilization. The conversion of the n-trailer system and of the extended one
(multi-steering n-trailer) into a multi-input chained form has been done by Tilbury
et al. [13].
The rest of the paper is organised as follows. Section 2 presents the mathematical
model of the n-trailer system equipped with a sliding kingpin mechanism and the
necessary assumptions used in the derivation. In Section 3 the flatness property of
the defined n-trailer system with kingpin sliding is examined and proved. The last
section presents some aspects about controllers selection for the sliding mechanism
and its influence in flatness conservation.

2. The n-trailer System Equipped with a Sliding Kingpin Mechanism


The n-trailer system is defined as a long and complex vehicle system consisting of a
suitable power tractor pulling a number of passive robot bodies or semi-trailers as
shown in Figure 1. Such systems are modeled as having one constraint on each axle:
That the velocity vector of the axle midpoint is constrained to move perpendicular

J Intell Robot Syst (2006) 46: 151162

153

Figure 1 Illustration of the


multi-articulated vehicle
coordinates.

to the axle direction or in other words that the wheels are allowed to roll but not
slip. These non-slipping constraints are nonholonomic or nonintegrable and do not
reduce the configuration space of the system.
Hence, the n-trailer system is a nonholonomic system subject to n+1 nonholonomic
constraints
y i cos (i ) x i sin (i ) = 0 i = 0, 1, ..., n
which lead to the relations
tan (i ) =

y i
x i

i = 0, 1, ..., n

(1)

As the term semi-trailer expresses, each trailer is hooked up to the mid-point of the
rear wheels of the preceding trailer or body. This means that the system is submitted
to 2n holonomic equations
xi = xi1 `i cos (i )
yi = yi1 `i sin (i )

i = 1, 2, ..., n

(2)

Also, we have the inputs equations


x 0 = U 1 cos (0 )
y 0 = U 1 sin (0 )
= U 2

(3)

where x0 ,y0 are the Cartesian coordinates of the leading vehicle (tractor), xi ,yi the
coordinates of the ith trailer and U1 ,U2 are the two control inputs, the linear velocity
and the steering angle rate, respectively. The other state variables represent the
orientation angles i for each trailer i, as shown in Figure 1, and is the tractors
steering angle. The only difference between a multi-body robotic system and a similar
multi-trailer vehicle (truck or road train) is the magnitude of the different physical
quantities (length, velocity, steering angle limits, weight, etc.).
Off-tracking is defined as the deviation of the semi-trailers axles or the kingpin
hitch from the path of the steering axle of the leading vehicle. Figures 2 and
3 illustrate off-tracking phenomenon showing two examples of multi-articulated

154
Figure 2 Trajectories for a
multi-articulated vehicle with
three trailers.

Figure 3 Trajectories for a


multi-articulated vehicle with
five trailers.

J Intell Robot Syst (2006) 46: 151162

J Intell Robot Syst (2006) 46: 151162

155

vehicles with three and five trailers, respectively. In the case of truck-trains, it is
more imperative than in any other case that the last semi-trailer follows exactly the
path of the lead tractor during a turn for lane change or a turn due to the curvature
of the highway. Otherwise, it will be possible for the last semi-trailer to violate the
outer boundary of the highway or to crash with an adjacent car during a lane change
although both keep invariant their relative velocity. It is known that the driver of any
long truck-train, because of the off-tracking of the rear trailers, turns the tractor far
towards the desired path in order to preserve rear trailers in acceptable boundaries.
When we deal with mobile robots the major problems are to find an obstacle-free
path and path following control. However, in the case of multi-articulated robotic
vehicles we must take into consideration the off-tracking phenomenon when finding
an obstacle-free path. The reason is that the last trailer may collide with obstacles
if the vehicle attempts to follow the designed path for the leading vehicle with offtracking neglected. One efficient way to solve this problem is to find an obstaclefree path for the leading vehicle, add a controller for path following and use another
kingpin controller for off-tracking elimination.
As mentioned above, the off-tracking can be eliminated by sliding the kingpin of
each trailer with respect to the previous one. Consider two intermediate semi-trailers
of a truck train as shown in Figure 4. According to this technique the kingpin hitch of
the i + 1th semi-trailer slides in a direction perpendicular to the longitudinal axle (i.e.
along the rear axle) of the ith trailer. The position of the ith semi-trailer Pi is taken
to be the middle point of the ith semi-trailers rear axle. The original configuration
space must be reobtained since the set of holonomic constraints has been changed.
Position Pi is defined by the pair (xi ,yi ) in the Cartesian coordinates system while
i is the orientation of the ith semi-trailer with respect to the horizontal axis. It has
been pointed out [14] that when the lead car of a single trailer system is traveling
along a circle of radius Rl , then the trailer is traveling along a circle of radius Rt with
the same center, where Rt < Rl , as shown in Figure 5.

Figure 4 The kingpin slides


along the axle when the
semi-trailer turns.

156

J Intell Robot Syst (2006) 46: 151162

Figure 5 Illustration of
off-tracking phenomenon.

In order to compensate for this path deviation of the trailer, we suppose that the
kingpin hitching point slides from the point Pi to the point Psi by a distance Si . The
following assumptions are necessary for deriving the mathematical model:
a) All trailers have the same length L.
b) Each trailer is modeled as having only one axle.
c) Each trailer is assumed to be hooked to the midpoint of the rear axle of the
preceding trailer.
d) By sliding the location of the kingpin, the weight of the trailer shifts toward an
outer direction, which doesnt affect the kinematic behavior of the train. This
assumption is more significant for heavy transportation multi trailer vehicles.
e) The unbalanced pulling point (when the kingpin sliding is nonzero) does not
cause skidding of the whole axle.
f) The sliding of the kingpin can be performed with the trailer fully loaded via a
hydraulic mechanism.
While the nonholonomic constraints of the new system remain the same (Equation
(1)), the 2n holonomic equations introduced by the corresponding links take a
modified form because of the sliding distance Si = Pi Psi (see Figure 2),
xi = xi1 `i cos (i ) + Si1 sin (i1 )
yi = yi1 `i sin (i ) Si1 cos (i1 )

i = 1, 2, ..., n

(4)

The inputs equations for the new n-trailer system with a sliding kingpin mechanism remain the same as without it (Equation (3)). After successive algebraic

J Intell Robot Syst (2006) 46: 151162

157

manipulations of Equations (1), (3) and (4) we can derive the mathematical model of
the n-trailer system with sliding kingpin,
x 0 = U 1 cos 0
y 0 = U 1 sin 0
= U 2
U1
0 =
tan
L


U1
1 = 2 sin (0 1 ) L + S0 tan
L



U1 
2 = 3 L + S0 tan sin (1 2 ) L cos (0 1 ) + S1 sin (0 1 ) .
L
..
.
n1
Y

U1 
n = n+1 L + S0 tan sin (n1 n ) [L cos (i1 i ) + Si sin (i1 i )]
L
i=1
Details can be found in [10]. The (x0 ,y0 ) position of the first vehicle along with all
the orientation angles {0 ,1 ,...,n } and steering angle define the entire state of the
system, while (U1 ,U2 ) is the control input.

3. Flatness of the n-trailer System with Sliding Kingpin


The notion of differential flatness was first introduced in [5] for nonlinear finite
dimensional systems. A system is said to be differentially flat if there exists a finite set
of outputs y = {y1 ,..., ym } that are differentially independent, and satisfy the following
conditions:

The outputs yi can be expressed as functions of the system variables and of a


finite number of their derivatives.
Any system variable can be expressed as a function of the outputs yi and of a
finite number of their derivatives.

The set y = {y1 ,..., ym } is called the flat (or linearizing) output and its number of
components equals the number of independent input channels. Given that there is
no general method to compute the linearizing outputs when the system is flat, some
options for the flat outputs were tried before deriving the following proposition.
PROPOSITION. The n-trailer system with a sliding kingpin mechanism is a differentially flat system with the Cartesian coordinates of the middle point of last trailers
rear axle y = [xn , yn ]T as flat output.
Proof. From Equation (4) after some algebraic manipulations we have

n
n1
X
X
`i cos (i ) +
Si sin (i )
" # x0

xn
i=1
i=0

n
n1
yn
X
X

y0
`i sin (i )
Si cos (i )
i=1

i=0

(5)

158

J Intell Robot Syst (2006) 46: 151162

The first condition for proving differential flatness property is satisfied by Equation
(5). Also from Equation (4) after simple calculations we have
xi = xi+1 + `i+1 cos (i+1 ) Si sin (i )
yi = yi+1 + `i+1 sin (i+1 ) + Si cos (i )

i = 0, 1, ..., n 1

(6)

From the nonholonomic constraints (1) we have for the n-trailer system with sliding
kingpin,
 
y n
1
n = tan
(7)
x n
and equivalently for the n 1 trailer
tan (n1 ) =
tan (n1 ) =

y n1 (6)
(y + `n sin (n ) + Sn1 cos (n1 ))0
tan (n1 ) = n

x n1
(xn + `n cos (n ) Sn1 sin (n1 ))0
y n + `n cos (n ) n + S n1 cos (n1 ) Sn1 sin (n1 ) n1
x n `n sin (n ) n S n1 sin (n1 ) Sn1 cos (n1 ) n1

(8)

By replacing n in Equation (8) from (7) we have


n1 = f xn , yn , x n , y n , x n , y n , Sn1 , S n1

(9)

Hence, the satisfaction of the second condition depends on the existence of the sliding
distance Si and its derivative in Equation (9). If the sliding distance in each trailer is
dependent on the trailers orientation angle,
Si = Ki i

(10)

then the orientation angle n 1 can be expressed as a differential function of (xn ,


yn ) and their derivatives up to order two. Similarly, we can prove that each i can be
expressed as a differential function of (xn , yn ) and their derivatives up to order n + 1
for 0 .
The linear controller of Equation (10) provides a good balance between simplicity
and effectiveness. Trajectories for two multi-articulated vehicles with three and five
trailers are shown in Figures 6 and 7, respectively.
From Equation (6) after successive substitutions we have
xn1 = xn + `n cos (n ) Sn1 sin (n1 )
xn1
z
}|
{
xn2 = (xn + `n cos (n ) Sn1 sin (n1 )) + `n1 cos (n1 ) Sn2 sin (n2 )
..
.
n
P
xi = xn +
(`k cos (k ) Sk1 sin (k1 ))
(11)
k=i+1

and similarly
yi = yn +

n
X

(`k sin (k ) + Sk1 cos (k1 ))

(12)

k=i+1

By replacing i in Equations (11) and (12) we prove that xi ,yi can be expressed as
differential functions of (xn , yn ) and their derivatives up to order n + 1 for x0 ,y0 .

J Intell Robot Syst (2006) 46: 151162


Figure 6 Trajectories for a
multi-articulated vehicle with
three trailers and linear
controller.

Figure 7 Trajectories for a


multi-articulated vehicle with
five trailers and linear
controller.

159

160

J Intell Robot Syst (2006) 46: 151162

For systems inputs and steering angle we will use Equation (3) from which we get
U1 =

x 0
cos (0 )

and hence input U1 can be expressed as a differential function of (xn , yn ) and their
derivatives up to order n + 2. For the steering angle we have
= tan1

`0 0
U1

thus steering angle can be expressed as a differential function of (xn ,yn ) and their
derivatives up to order n + 2. For the second input we have
U 2 =
and consequently input U2 can be expressed also as a differential function of (xn , yn )
and their derivatives up to order n + 3. Finally for the sliding distances Si , because we
supposed that Si = Ki i , the Si can be expressed as differential functions of (xn , yn )
and their derivatives up to order n + 2. It must be noted that the n-trailer system with
sliding kingpin has been proved to be a differentially flat system under the assumption
that a linear controller was used for the sliding distance. However, as soon as other
sliding controllers and especially nonlinear are introduced the system may be not flat
anymore.

4. On Flatness and Sliding Kingpin Controller Selection


The sliding kingpin technique was proposed in order to eliminate the off-tracking
phenomenon. The main issue is to design the controller with the best results. Various
controllers were tested through simulations for the system with three or five trailers.
The nonlinear controller of Equation (13) gave us the best results. Figures 8 and
9 show simulations results, for two multi-articulated vehicles with three and five
trailers, where off-tracking is practically zero in all phases of motion. Details can be
found in [11].
Si = `i

1 cos (i i+1 )
sin (i i+1 )

(13)

If we use this type of controller we must again prove the satisfaction of the second
condition of differential flatness property. By substituting Si from Equations (13) to
(9) we have

n1 = f xn , yn , x n , y n , x n , y n , n , n
and by using (7) we can prove that the orientation angle n 1 can be expressed as
a differential function of (xn , yn ) and their derivatives up to order two. Similarly,
we can prove that i can be expressed as a differential function of (xn , yn ) and their
derivatives up to order n + 1 for 0 . For the other systems variables we have the same
results as for the case of the linear controller Si = Ki i .

J Intell Robot Syst (2006) 46: 151162


Figure 8 Trajectories for a
multi-articulated vehicle with
three trailers and nonlinear
controller.

Figure 9 Trajectories for a


multi-articulated vehicle with
five trailers and nonlinear
controller.

161

162

J Intell Robot Syst (2006) 46: 151162

5. Conclusions
This area of work is part of an ongoing research effort in improved control of
multi-articulated vehicles with application to problems in both mobile robotics and
transportation complex vehicles. In this paper we have shown that the n-trailer
system keeps its structural flatness property when it is equipped with a kingpin sliding
mechanism, which moves dynamically the hitching point of each trailer. The goal is to
exploit the advantages of flatness in order to generate effective control strategies of
the mechanical system. In this case, problems related with the n-trailer system such as
motion planning, path following, steering etc. can be reduced to simple algebra and
computationally efficient algorithms.
Acknowledgment This research work is partial supported by Karatheodori Program of the Research Commission of the University of Patras.

References
1. Rouchon, P.: Motion planning, equivalence, infinite dimensional systems. Mathematical Theory
of Networks and Systems, MTNS Perpignan France, (2000)
2. Rouchon, P., Fliess, M., Levine, J., Martin, P.: Flatness and motion planning: The car with n
trailers. In: European Control Conference, pp. 15181522, 1993
3. Van Nieuwstadt, M., Rathinam, M., Murray, R.M.: Differential flatness and absolute equivalence
of nonlinear control systems. SIAM J. Control Optim. 36(4), 12251239 (1998)
4. Sekhavat, S., Rouchon, P., Hermosillo, J.: Computing the flat outputs of Engel differential
systems. The case of the Bi-steerable car. In: American Control Conference, Arlington, USA,
2001
5. Fliess, M., Lvine, J., Martin, P., Rouchon, P.: Flatness and defect of nonlinear systems: Introductory theory and examples. Int. J. Control 61(6), 13271361 (1995)
6. Murray, R.M., Rathinam, M., Sluis, W.: Differential flatness of mechanical control systems: A
catalog of prototype systems. In: 1995 ASME International Mech. Eng. Congress, 1995
7. Petit, N., Rouchon, P.: Flatness of heavy chain systems. SIAM J. Control Optim. 40(2), 475495
(2001)
8. Mahadevan, R., Agrawal, S.K., Doyle, F.J.: Differential flatness based nonlinear predictive control of fed-batch bioreactors. Control Eng. Pract. 9, 889899 (2001)
9. Manesis, S.A.: Off-tracking elimination in road trains of heavy duty trucks with multiple semitrailers. In: Proceedings of 8th IFAC International Symposium on Large Scale Systems: Theory
and Applications, Patras, Hellas, 1998
10. Davrazos, G.N., Manesis, S.A., Koussoulas, N.T.: Controller design and validation for offtracking elimination in multi-articulated vehicles. Int. J. Syst. Sci. 27(4), 129137 (2001)
11. Manesis, S., Kousoulas, N., Davrazos, G.: On the suppression of off-tracking in multi-articulated
vehicles through a movable junction technique. J. Intell. Robot. Systems 37, 399414 (2003)
12. Altafini, C., Gutman, P.O.: Path following with reduced off-tracking for the n-trailer system. In:
Proceedings of the 37th IEEE Conference on Decision and Control, Florida, USA, 1998
13. Tilbury, D., Sordalen, O.J., Bushnell, L., Sastry, S.S.: A multisteering trailer system: Conversion
into chained form using dynamic feedback. IEEE Trans. Robot. Autom. 11(6), 807817 (1995)
14. Bushnell, L.G., Mirtisch, B., Sahai, A., Secor, M.: Off-tracking bounds for a car pulling trailers
with kingpin hitching. In: Proceedings of the 33rd Conf. On Decision and Control, Lake Buena
Vista, Florida, 1994

Potrebbero piacerti anche