Scribd Logo
Upload

Welcome to Scribd!

  • Skid Steer

    Uploaded by

    Hala El Ouarrak
    9 views7 pages
    Skid Steer

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    Skid Steer

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    9 views7 pages

    Skid Steer

    Uploaded by

    Hala El Ouarrak
    Skid Steer

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    of 7
    Proceeding ofthe 1999 BEE. Ieemutonl Conference on Robatis& Automation ‘ett, Michigan May 1999, ‘Trajectory Tracking Control of a Four-Wheel Differentially Driven Mobile Robot Luca Caracciolo Alessandro De Luca Stefano Iannitti Dipartimento di Informatics ¢ Sistemistien ‘Universita deglt Studi di Roma “La Sapienza" Via Eudossiana, 00184 Rome, Italy {detuca}elabrob,ing.untrosat it Abstract We consider the trajectory trucking control problem for a 4-wheel differentially driven mobile robot mov- ing on an outdoor terrain. A dynamic model is pre sented accounting for the effects of wheel skidding. A ‘model-based nonlinear controller 18 designed, follow ing the dynamic feedback linearization paradigm. An ‘operational nonholonomic constraint is added at this stage, s0 as to obtain a predictable behavior for the ine Stantaneous center of rotation thus preventing exces sive skidding. The contvoller is then robustified, using conventional linear techniques, against uncertainty in the soil parameters at the ground-uheel contact. Sim- ‘ulation results show the good performance in tracking spline-type trajectories on a virtual terrain with vary- ing characteristics. 1 Introduction Robotic autonomous navigation tasks in outdoor ronments can be effectively performed by skid- steering vehicles (1). Typically, these are d-wheel dif- ferentially driven (4wdd) vehicles in which rotational notion is achieved by a differential thrust on wheel pairs at opposite sides. One commercial example is the ATRV-2 mobile robot by RWI (see Fig. 1). The absence of a steering system makes dwdd vehicles me- chanically robust and able to move on rough terrains with ease and good maneuverability. In applications here the robot needs to move also on paved grounds, skid-steering vehicles are preferred to tracked vehicles because they do not corrupt the contact surface, Al- though the dynamie behavior of ddd skid-steering ve hicles is similar to that of vehicles with tracks, the lt- erature on autonomous navigation has focused mostly fn the latter elass[2, 3} 0-7803-5180-0-5/99 $10.00 © 1999 IEEE Figure 1: ATRV-2 mobile robot ‘When considering the problem of accurate trajec- tory tracking, dwdd vehicles are quite difficult to con- trol. In fact, in order to follow a curved path, the wheels need to skid laterally and cannot be tangent to the desired path. Moreover, the Instantaneous center ‘of rotation (ICR) of dwdd vehicles may move out ofthe robot wheelbase, causing loss of motion stability This is diferent from eay-like vehicles, whose ICR is always theoretially fixed along the rear wheels axis [4]. From the modeling point of view, the equilibrium equation of the forees orthogonal to the wheels should be taken into account and this prescribes the use of a dynamic ‘model for control design purposes, instend ofa simpler kinematic one. In this paper, we present a robust trajectory track= ing control system for 4wdd vehicles. After deriving a dynamic model of a dwdd vehicle, we design a model- bbased tracking controller, by borrowing an approach used for motion planning and control of nonholonomic 2632 wheeled mobile robots (4, 5]. The basic idea is to spec ify the longitudinal coordinate of the ICR, in order to force it to remain inside the robot wheelbase. Adding this operational kinematic constraint, we proceed with full linearization of the system via dynamic state feed- back. The obtained closed-loop system is linear and ‘input-output decoupled, thus stabilization to a desired trajectory is easily achieved by means of linear con- trol techniques. However, the overall nonlinear com- ‘rol law depends on soil parameters {3}. We study the effects of uncertain soil parameters on the dynamics of the closed-loop system with respect to a specific class of trajectories and propose then a robust con- trol scheme that reject these disturbances. Sinuulation results on tracking of robot trajectories over a virtual terrain characterized by varying parameters are finally presented 2 Dynamic Modeling We develop a vehicle dynamie model useful for control design, neglecting some side effects introduced, e.g. by suspensions and tire deformation (ee, ¢, (6) In patticular, we make the following assumptions: 1. Rigid vehicle moving on a horizontal plane. 2. Vehicle speed below 10 km/h (about 6 mph). 3. Longitudinal wheel slippage neglected. 4. Tire lateral force function of its vertical load. 2.1. Skid-steering motion analysis Define a fixed reference frame F(X,Y) and a moving frame f(x,y) attached to the vehicle body, with origin at the vehicle center of mass G (see Fig. 2). The center of mass is located at distances a and b (usually, a < ) from the front and rear wheels axes, respectively, and is symmetric with respect to the vehicle sides (at distance ¢) Let #,j, bbe, respectively, the longitudinal, lateral, ‘and angular velocity of the vehicle in frame f. In the fixed frame F, the absolute velocities are X) _ [eeose— pane 3 [F]-Eeecfese] = [5] Dileretiton with espe to tne ses [#]-#0[5+3]-#0[5) ‘where a, and ay are the absolute accelerations ex pressed in the moving frame f- At each instant the vehicle motion is a pure rotation around a point C, the instantaneous center of rotation, in which the lin- ear velocity components in f vanish. Its coordinates [e] - [24] Lue] =| #/6 ‘The angular velocity and the lateral velocity y both vanish during straight line motion, and the ICR goes to infinity along the y-axis. On a curved path, the ICR shifts (forwards) by an amount [2-|. When j= 0, there is no lateral skidding. If re goes out of the robot ‘wheelbase, the vehicle skids dramatically with loss of motion stability. Finally, note that the longitudinal velocity i; and the lateral (skidding) velocity sj of each wheel (7 = Aye--y4) are given by Aah = 2-08 (let) fnein = 2410 (eh) Gy isin = +ad (front) in=in = 9-06 (vont) 2.2. Equations of motion ‘The freo-body diagram of forces and velocities is shown in Fig. 2, with the vehicle having instantaneous posi- tive velocity components # and # and negative velocity. 4. Wheels develop tractive forces Fy and are subject to longitudinal resistance forces Rey, for i = 1,..- 4 We assume that wheel actuation is equal on each side 50 as to reduce longitudinal ship. ‘Thus, it will always. be Fag = Fet and Fag = Fig. Lateral forces Fy; act on the wheels as @ consequence of lateral skidding, Also, f resistive moment M, around the center of mass is induced in general by the Fy; and Re forces. For a vehicle of mass m and inertia I about its een- tor of mass, the equations of motion can be written in frame f as: maz = 2Fi+2Fen~ Re -% @ 20(F, ~ Fea) ~ Mr 6 ‘To express the longitudinal resistive force Re, the lateral resistive force Fy, and the resistive moment M,, we should consider how the vehicle gravitational load mg is shared among the wheels and introduce ‘Coulomb friction model for the wheel-ground contact. We have Pa=Fa et mg 2 mg. 2 Fas Fa ; 2633 Figure 2: Free-body diagram At low speed, the lateral load transfer due to centrifu gal forces on curved paths can be neglected. In case ‘of hard ground, we can assume that the contact patch between wheel and ground is rectangular and that the tire vertical load produces an uniform pressure distri- bution. In this condition, Rey = fr sgu(t;), where fr is the coefficient of rolling resistance, assumned in- dependent from velocity {7}. The total longitudinal resistive force is then Dre R, fe 78 (ogn(és) +sen(éa)). 3) Introducing a lateral friction coefficlent 1, the lateral force acting on each wheel will be Fy; = 4F:.8gn(jh) The total lateral force is thus adhe TE (bsen(ir) +asen(io)), (4) while the resistive moment is Mp = o(Fy + Fa) WR + Py) +t[( Rez + Res) ~ (Rei + Rea)] = 6 SEE (ounin) — senGds)) ++ fe 532 (oan (da) — sentés)) (5) ‘The dynamic model can be rewritten in frame F, introducing the generalized coordinates q = (X,¥,6) ‘and matrix notation Mij+ c(9,4) = Ela). 6) 2634 ‘m 0 0 Re cos — Fy sind M=|0 m0], cqd)=| Resind + Fyoose oor Mr e080 /r cos /r E(q)=| sing/r sind/r | , 4 = 2rFe: (§=1,2), te tir being r the wheel radius, 7 and 7, the torques pro- duced by the left and right side motors at the load side, respectively. An ideal transmission factor is also assumed. 3. Trajectory Control 3.1 Operative nonholonomic constraint We start by observing that 2- (the 2-axis projection of ‘the instantaneous center of rotation) cannot be larger than a. If this happens, the vehicle would skid along the y-axis thus losing control. In order to have the vehicle move properly, one should have then ‘Therefore, we can introduce the following operative constraint Gt dd =0, 0

    You might also like