Vous êtes sur la page 1sur 6

12th IFToMM World Congress, Besanon (France), June18-21, 2007

CK-xxx

Special new configuration of wheeled robot to work over uneven terrain: sand, mud and snow
T. Akinfiev* M. Armada A. Ramirez Industrial Automation Institute Madrid, Spain
However, usually they have a very complex design, even more complex than legged robots; they also have complicate control system, they require a big number of motors, etc. In this paper it is considered a new and simple configuration of mobile robot which takes the advantages of positive properties of both: wheeled and legged robots. The construction of the robot is based on a new concept of design of wheeled mobile robot, which can work in two different regimes. Like wheeled robot it can work on even (solid and smooth) surfaces and it can also move as hybrid robot on an uneven rugged terrain (like sand, mud and snow) [11].

Abstract In the paper, a wheeled robot with changeable structure is presented. The robot is able to work under two modes: under continuous movement mode over sufficiently solid and smooth surfaces and under discontinuous movement mode over soft surfaces. Under discontinuous movement mode the wheels of the robot alternately act as supporting legs (blocked wheels) and as freely rolling wheels. Specific dynamic properties of the robot have been discovered using analytical methods. The problem of time optimal control is solved. Experimental results obtained using a prototype confirmed theoretical considerations. Field of application of the robot is connected with disabled people transportation. Keywords: mobile robot, design, changeable structure, dynamical analysis.

I. Introduction Principally, two main kinds of mobile robots are known: legged robots and wheeled robots. Legged robots can travel over uneven terrain; however, they have a very complex construction. Among some of their features are: usually they are working with on-board computer, they have numerous DOF, they have low efficiency, very high energy expenses and low autonomy; in general, they are very expensive and act with low velocity [1]. On the other hand, wheeled robots have comparatively simple construction, they are not expensive, they have high efficiency, they can work with high velocity and usually they have a simple control system. However, they can move only over solid and smooth surfaces and not able to cross big obstacles [2-7]. There exist special devices with wheels of big diameter and big width [12, 13]. These devices allow for transportation of disabled people along soft surface; however, because of the huge dimensions they cannot go through standard door so that they cannot be used inside buildings. The authors of several studies [8-10] attempted to mix positive properties of legged and wheeled robots. Normally, it is done on the bases of legged robot configuration adding wheels at the end of the legs. This kind of robots is called hybrid robots. Hybrid robots can have high velocity over smooth surfaces and use legs to cross obstacles.
*E-mail: teodor@iai.csic.es E-mail: armada@iai.csic.es E-mail: aramirez@iai.csic.es

Motor 1 Reducer Belt Pulley Passive pulley Mobile element

Stopper Motor 2

Front wheel Stopper

Fig. 1. Configuration of the robot

Figure 1 shows the robots configuration. The robot is formed by a body of the robot, by several axis connected to the wheels and to the body of the robot, and by wheels which are able to rotate around the axis. Due to the new design concept, the axis of the rear wheels is directly connected to the body of robot and the axis of the front wheels is located on a special mobile element, which is connected to the body of robot and can slide forward and back. Motor 1 is fixed on the body and is kinematically connected to the mobile element (through a transmission belt-pulley or screw-nut). The rear wheels are fed from motors 2, which rotate these wheels and determine the direction of the movement of the robot. Each wheel has a guided stopper. With two-side stoppers, rotation of wheels can be prevented.

12th IFToMM World Congress, Besanon (France), June18-21, 2007

CK-xxx

III. Robots working regime If the surface is sufficiently plain, smooth and solid, the robot is working under the first regime of work. In this regime, the motor 1 is disconnected, the stoppers are disconnected and the robot is working helped by the motors 2 like any other traditional wheeled robot with all advantages for such kind of machines. If the robot starts to work under the same regime on a sandy surface, the torque on the powered wheel (or wheels) starts to move the under-wheel sand, the wheel would be caved in the sand and the robot would not be able to continue moving forward. For this case the robot works under its second regime (the motor 2 is disconnected to prevent self-excavation of wheels). The second regime of work consists of: (1) The stopper of front wheel is disconnected, the stoppers of rear wheels are connected; motor 1 starts to move mobile element forward together with front axis and front wheel; in this step the front wheel makes passive forward rotation; the rear wheels do not move because of the applied stoppers. (2) When mobile element arrives to the switching point, motor 1 starts to apply force to the mobile element in opposite direction. The motor stops when mobile element arrives to the extreme position. (3) At this moment, the stoppers of rear wheels are switched off and are now disconnected; the stopper of front wheel is connected. Motor 1 tries to move the mobile element back but the mobile element does not execute any movement because of the force of friction between the front wheel and the ground, and because front wheel is fixed. Consequently, the body of robot starts to be moved forward with positive acceleration; in this step the rear wheels make passive forward rotation; (4) When the body of robot arrives to the switching point, motor 1 starts to apply force to the mobile element in opposite direction. The body of robot is moved forward with negative acceleration; motor 1 stops when mobile element arrives to the extreme position. After that the whole process starts again. Figure 2 shows the above described steps of the second regime. Thanks to the stoppers applied on rear wheels, during steps 1 - 2 these wheels behave together like legs in walking robot, which can hold a given position on the ground. During these steps the front wheel acts as conventional passive wheel. In the same way, in steps 3-4, thanks to the stopper on the front wheel, it together with the mobile element, behave like a leg in walking robot, which can hold the given position on the ground. During the same steps, the rear wheels act as conventional passive wheels. Along one full cycle of movement, front and rear wheels are playing consecutively two roles: wheel and leg, conferring to the robot a hybrid robots typical performing.

Initial position

a
Step 1

V V

a
Step 2

Step 3

a
Step 4

Fig. 2. Full cycle of movement of robot

IV. Robots dynamical properties Dynamical properties of the robot working under the first regime of movement are out of consideration in this paper because they have been discussed by other authors and their results are very well known [2-7]. Only dynamical properties of the robot working under the second regime of movement will be considered. One full cycle of movement under the second regime of work consists of 4 steps. A. Step 1 To analyze the dynamical properties it is necessary to make an independent description of the forces that affect the body of the robot (including rear wheels, fig. 3) and the mobile element (including front wheel, fig. 4). N21 FP1 L1

d FFR21

FT1 Mg

FFR11

X Fig. 3. Forces applied to the body of the robot

Figure 3 describes all forces applied to the body of the

12th IFToMM World Congress, Besanon (France), June18-21, 2007

CK-xxx

robot, so that the following system of equations can be written: N 21 + FP1 Mg = 0 (1)

Equations (5) and (6) correspond to the projection of all forces on X and Y coordinates. FP1

FFR11 + FFR 21 FT 1 = 0 MgL1 FP1 2 L1 FFR 21d = 0 FFR 21 K 2 N 21

(2) (3) (4) FT1 FFR11

where: M - mass of body of robot. g - acceleration of gravity. K2 - coefficient of static friction between wheels of robot and the ground. O - pivot point (for calculation of moments of forces). L1 - distance between the pivot point O and the centre of gravity of the body of robot. d - distance between the pivot point O and the ground. N21 - normal force applied to the rear wheels of robot from the ground. FP1 - force applied between the mobile element and the body of robot. FFR11 force of sliding friction applied between the body of robot and the mobile element. FFR21 force of static friction applied between blocked wheels of robot and the ground. FT1 - force applied to the body of robot or to the mobile element from the transmission system. N11 - normal force applied to the front wheel of the mobile element from the ground. The equations (1) and (2) correspond to the projection of forces on X and Y coordinates. Because of the stoppers applied on rear wheels and the force of friction between the rear wheels of robot and the ground, the robots body does not move and the acceleration of the body of robot is equal to zero. The equation (3) corresponds to the rotational equilibrium of the body of robot. The stoppers of rear wheels are connected; in consequence there is no rotation of the rear wheels. Besides, it is assumed that rear wheels do not slide as well. In this case, the force of friction applied between the rear wheels and the ground, when there is no motion, is static friction force. The magnitude of the static friction force is given by the inequality (4). Figure 4 describes all forces applied to the mobile element (together with front wheel). The following system of equations can be established: N11 FP1 mg = 0 (5)

a1
Y X N11
Fig. 4. Forces applied to the mobile element

mg

During step 1 the mobile element is in motion; it means that the force of friction applied between the mobile element and the body of robot is a sliding friction. The magnitude of the sliding friction force is given by the equation (7). In this point an additional remark has been done. The front wheel is not fixed by the stoppers, and it can rotate freely. The inertia of front wheel is very small and in consequence the magnitude of the force of friction applied between the front wheel and the ground is assumed to be very small, so in the present analysis this force of friction is not taken into account. An important condition of the considered system consists of the normal force applied to the wheels of robot from the ground has to be positive (N11>0) in order to keep the front wheel in contact to the ground all the time. This limitation is especially important when the robot is working over inclined surfaces. From the system of equations (1) (7) it is easy to find the maximum possible value of acceleration ( a1max ) for the mobile element. It is interesting to note, that the magnitude of acceleration is a robots self-property, which is connected to the forces of friction and the mass of mobile element. This means, if motor 1 is not powerful enough, then the mobile element would not be able to reach the maximum possible acceleration; however, even if motor 1 is very powerful, then it would be impossible to reach acceleration higher than a1 max . Fig. 5 shows the maximum possible velocity of mobile element and body of robot versus time. According to this figure, the mobile elements movement takes place with maximum possible acceleration a1max and a 2 max (step 1 and step 2, accordingly) and the bodys movement takes place with maximum possible acceleration a3 max and

FT 1 FFR11 = ma1 FFR11 = K 1 FP1

(6) (7)

where: m - mass of the mobile element including front wheel. a1 - acceleration of mobile element. K1 - coefficient of sliding friction between the mobile element and the body of robot. 3

a 4 max (step 3 and step 4, accordingly).

12th IFToMM World Congress, Besanon (France), June18-21, 2007

CK-xxx

Velocity (m/s)

1,5 1 0,5 0 0 Step 1 & 2 0,5 1 Step 3 & 4

higher would be the maximum possible value of acceleration a1max. The relationship between values K1 and K2 has to be taken into consideration while designing the robot. In consequence, once the problem solved analytically, it is necessary to take in to account some margin of reserve for coefficients, which are known only approximately. B. Step 2 Similarly, for the step 2, the following system of equations can be written: N 22 + FP 2 Mg = 0 (10)

t (s)

Fig. 5. Maximum possible velocity of the robots body and mobile element.

The maximum possible magnitude of acceleration (a1max) during step 1 can be achieved when the static force of friction applied between the rear wheels of robot and the ground (FFR21) is equal to its limit value before sliding, i.e., when FFR21 = K2N21 In this case:

FT 2 + FFR12 FFR 22 = 0 MgL1 FP 2 2 L1 + dFFR 22 = 0 FFR 22 K 2 N 22 v N12 FP 2 mg = 0 FFR12 FT 2 = ma2 FFR12 = K 1 FP 2


From this system the value of acceleration a2 is:

(11) (12) (13) (14) (15) (16)

K2 m

MgL1 2 L dK 2 1

= a1 max

(8)

K 1 ( MgL1 + dFT 2 ) FT 2 = a2 m(dK 1 2 L1 ) m K2 MgL1 ( ) = a 2 max . m 2 L1 + dK 2

(17)

The force applied to the body of robot or to the mobile element from the transmission system corresponding to the maximum possible value of acceleration (a1max) is:

And then,

(18)

FT 1 max = ma1 max + K 1 Mg

K 1 MgL1 . (9) (2 L1 dK 2 )

It is important to mention that during step 1 the value of FT1 has to be less or equal than FT1max; it has to be more than some minimal value (FFR11), i.e., FFR11<FT1FT1max. If FT1<FFR11, the mobile element would not execute any movement. On the other hand, if FT1>FT1max, the rear wheels of robots body would start to slide. Another limitation (related to FT1) is associated to the possibility of the body of robot to turn around its rear wheels. To prevent this effect it is necessary to comply the condition FT1<F*T1max, where F*T1max corresponds to the maximum possible value of the force applied to the body of robot or to the mobile element from the transmission system, when N11 is positive. Finally, during step 1, there must be a special relationship between K1 and K2. A critical value of this relationship is related to geometrical characteristics of the robot, position of the mobile element, etc. For example, in case when the robot is symmetrical (like it is shown in figure 1), and the distance of displacement of the mobile element l is small (l L1), the critical value of the relationship between K1 and K2 is 1 (K1/K2=1). In this case, if K1>K2, the robot would not be able to work properly because its rear wheels would slide; if K1<K2 , the robot can work, and as smaller is the value of K1 , 4

The force, corresponding to the maximum possible value of acceleration (a2max), applied to the body of robot or to the mobile element from the transmission system, can be described by:

FT 2 max = ma 2 max K 1 mg +
C. Step 3

K 1 MgL1 dK 2 + 2L1

(19)

The system of equations corresponding to step 3 is:

N 23 + FP 3 Mg = 0 FX = FT 3 FFR13 = Ma3 FFR13 = K 1 FP 3 N 13 FP 3 mg = 0 FFR13 + FFR 23 FT 3 = 0 FFR 23 K 2 N 13 MgL1 N 13 2 L1 + mg 2 L1 dFFR 23 = 0


From the above system of equations follows:

(20) (21) (22) (23) (24) (25) (26)

a3 =
Then,

FT 3 K 1 (dFT 3 L1 Mg ) M M (dK 1 2L1 )

(27)

12th IFToMM World Congress, Besanon (France), June18-21, 2007

CK-xxx

K 2 gL1 M + 2m M 2 L1 + dK 2

= a3 max

(28)

where FT3max can be described by:

M + 2m mg (29) FT 3 max = Ma3 max + K1 gL1 2 L1 + dK 2


D. Step 4 For step 4 it is possible to write the following system of equations: N 24 + FP 4 Mg = 0 (30)

FT 4 FFR14 = Ma4 FFR14 = K 1 FP 4 N14 FP 4 mg = 0 FT 4 + FFR14 FFR 24 = 0 FFR 24 K 2 N 14 MgL1 N 14 2 L1 + mg 2 L1 + dFFR 24 = 0


From the above system of equations follows:

(31) (32) (33) (34) (35) (36)

distance of each step Xi is calculated. This allows finding switching points for motor control. The figure 5 illustrates the relationship between velocity and time during steps 1-4. In the initial moment mobile element starts the movement with zero velocity. The mobile element moves with maximum possible acceleration, and when it is at a switching point, step 1 is over and step 2 begins. During step 2 the mobile element moves with maximum possible negative acceleration. Step 2 is completed when velocity of mobile element is equal zero. Steps 3 and 4 are performed in the same manner, the body of the robot is in motion instead of mobile element. Solving the problem of optimization for step 1, it is possible to find:

VS1 = 2a1 max X 1


VS 1 = 2 a2 max X 2

(40) (41) (42) (43)

X1 + X2 = l 2X1 . t1 = a1 max t2 = 2X 2 a 2 max

FT 4 K1 dFT 4 + MgL1 = a4 + M M dK 1 2 L1 K 2 gL1 M + 2m M 2 L1 dK 2 = a 4 max

(37)

(44)

Then we obtain:

(38)

where the value of FT4max corresponding to a4max can be described by:

where l is a full distance of the displacement of the mobile element. From the above system of equations it is easy to find X1, X2, t1, t2. For one full cycle of movement, the average robots movement velocity is:

M + 2m FT 4 max = Ma 4 max K1 gL1 2 L dK mg (39) 2 1


As it has been shown above, for each step it is possible to obtain the maximum possible value of acceleration aimax, i=1,,4. This takes place when the static force of friction, applied between the wheels of the robot connected to the stoppers (rear wheels during steps 1-2, and front wheels during steps 3-4) and the ground, is equal to its limit value before sliding. The values of aimax are not related to the properties of the motor 1. Those values are obtained considering only self properties of robot and the working surface. Even if the motor 1 is very powerful, it is impossible to support values of acceleration higher than aimax. V. Time optimal control In order to improve the system performance, time optimal control has to be included. Time optimal control for one full cycle of movement must be found, according to the values of maximum possible acceleration calculated in each step. The optimization has to be made independently for steps 1-2 and 3-4. To solve the problem of time optimal control the 5

l ~ V14 = T0

(45)

where the total time required to execute one full cycle of movement is described by: T0 = t1 + t 2 + t 3 + t 4 (46) VI. Experiments To confirm the effectiveness of the use of new designed robot, which can work over uneven terrain, a special prototype was constructed, built and tested. A bottom view of prototype is presented in fig. 6. The tests made with the prototype confirmed the dynamical properties described above. During experiments a new property of the robot was found. It was observed that if any of the moving wheel (for example, front wheel) contacts to an obstacle, then fixed wheels (in this case, rear wheels) start to slide back and move back a top layer of sand. Sand is elevated in front of the sliding rear wheels, and a support is created preventing further sliding of the rear wheels. At this moment the front wheel begins to cross the obstacle. This effect is shown in fig. 7.

12th IFToMM World Congress, Besanon (France), June18-21, 2007

CK-xxx

The same effect helps robot to continue movement even in a situation, when K1 < K2. Self creation of a special sand support is equivalent to increasing static force of friction. Of course, in this case average velocity of movement decreases because during some period of the whole working time robot is moving back. VII. Conclusion New design of a special configuration of wheeled robot for displacements over uneven terrain (sand, mud and snow) has been developed. Several working regimes of the robot have been described. Dynamical properties of robot have been calculated. The problem of time optimal control has been solved. Experimental results with a prototype confirmed theoretical considerations. Such robot can be used for transportation of disabled people along soft or hard surfaces. References
[1] [2] Walking Machine Catalogue: http://www.walking-machines.org/. Fierro, R. and Lewis, F. L. Control of a nonholonomic mobile robot: Backstepping kinematics into dynammics. In 34th IEEE Conference on Decision and Control, 1999, pages 3805-3810, New Orleans, USA. Akinfev, T., Armada, M., and Fernandez, R. Vehicle control method. Patent No2187375, Spain. Astolfi, A. Exponential stabilization of a wheleed mobile robot via discontinuos control. Journal of Dynamic Systems, Measurements, and Control, 1999, 121, 121-126. Everett, H. R. Sensors for Mobile Robots: Theory and Application. AK Peters Ltd., 1995. Fukao, T., Nakagawa, H., and Adachi, N. Adaptive tracking control of a nonholonomic mobile robot. In IEEE Transactions on Robotics and Automation, 2000, 16(5), 609-615. Guldner, J. and Utkin, V. I. Stabilization of nonholonomic mobile robots using Lyapunov functions for navigation and sliding mode control. In 33rd IEEE Conference on Decision and Control, 1994, pages 2967-2972, Orlando, Florida, USA. Leppanen I., Salmi S. and Halme A. Workpartner HUT Automations new hybrid walking machine. CLAWAR 98, Brussels. Matsumoto O., Kajita S., Saigo M. and Tani K. Biped type leg wheeled robot. Advanced Robotics, vol. 13, No 3, pp. 235 236. Adachi H., Koyachi N., Arai T., Shimuzu A. and Nogami Y. Mechanism and control of leg wheel hybrid mobile robot. IEEE/RSJ International Conference on Robotics and Automation, pp. 17921797, 1999. Akinfiev T., Fernandez R., Armada M. and A. Ramirez. Device for people or payload transportation and its control method. Patent application P200503060, Spain. http://www-robot.mes.titech.ac.jp/robot/wheeled/helios3/helios 3_e.html http://www.ncaonline.org/products/beach-wheelchairs/index.shtml #skipnav

[3] [4]

[5] [6] Fig. 6. A bottom view of the robots prototype [7]

[8] [9] [10]

[11]

[12] [13]

Fig. 7. The illustration of the sand support creation.

Vous aimerez peut-être aussi