Indonesian Journal of Electrical Engineering and Computer Science V ol. 2, No . 3, J une 2016, pp . 478 485 DOI: 10.11591/ijeecs .v2.i3.pp478-485 478 Bac kstepping Appr oac h f or A utonomous Mobile Robot T rajector y T rac king Ibari Benaoumeur* 1 , Benc hikh Laredj 2 , Hanifi Elhac himi Amar Reda 3 , and Ahmed-f oitih Zoubir 4 1,3,4 Labor ator y of P o w er Systems , Solar Energy and A utomation L.E.P .E.S .A. Univ ersity of Sci ences and T echnology of Or an UST O/MB . Or an, Alger ia 2 IBISC labor ator y (Inf or matics , Integ r ativ e Biology and Comple x Systems), Univ ersit y of Evr y , F r ance * Corresponding author , e-mail: benaoumeur .ibar i@univ-usto .dz Abstract This paper proposes a bac kstep ping controller design f or tr ajector y tr ac kin g of unicycle-type mobile robots . The main object of the control algor ithms de v eloped is to design a rob ust output tr ac king controller . The design of the controller is based on the ly apuno v theorem, kinematic tr ac king controller of an unicycle- lik e mobile robot is used to pro vides t he desired v alues of the linear and angular v elocities f or the giv en tr ajector y . A L y apuno v-based stability analysis is presented to guar antee the robot stability of the tr ac king errors . Sim ulation and e xper imental results sho w the eff ectiv eness of the proposed rob ust controller in ter m of accur acy and stability under diff erent load conditions . K e yw or ds: Non-linear systems , T r ajector y tr ac king, Dynamic model, bac kstepping. Cop yright c 2016 Institute of Ad v anced Engineering and Science 1. Intr oduction Diff erential wheeled robot are becoming more popular f or perf or ming tasks that are too dangerous or tedious f or humans , the y are widely used in: industr y , science , education , enter tain- ment and militar y applications [1], f or this w a y , There are man y research ar ticles emphasizing the impor tance of designing controller at the control of mobile robot. Most of them ha v e f ocused on tr ajector y tr aking [2, 3, 4]. In these algor ithms , the v elocity control inputs is defined to stabiliz e the closed-loop system. In path f ollo wing, the v elocity control is designed to stabiliz e a car-lik e mobile robot [5], this prob lem of stabilization is solv ed about a desired posture in [6]. Most controllers designed in control systems are not based on dynam ic systems and control theor y [7, 8, 9]. In [10] an adaptiv e f ollo wing controller based on the PID f or mobile robot path f ollo wing is presented, one adv antage of their controller is that its the control la w is constr ucted on the basis of L y apuno v stability theor y . Once more , just a model of the robot mobile kinematics is used and no e xper ime ntal results w ere repor ted. Ho w e v er , it is necessar y to using the tools from control theor y and dynamic systems in order to ensure system stability . Man y researchers ha v e sho wn interest in applying controllers designed f or wheeled robot mobile based on dynamic model. Diff erent approaches ha v e been in v estigated using fuzzy control [11, 12], sliding mode control (SMC)[13, 14], adaptiv e control [15, 16, 17], or bac kstepping control [18, 19]. In the presence of par ametr ic uncer ta inties and noises in only its dynamic model, a com- bination of model ref erence adaptiv e control and gain scheduling is de v eloped in [15] to control the robot motion b y the adaptiv e controller . In [16] an adaptiv e controller based on the dynamic model pro vides the torques of the robot actuators f or yielding the required v elocities is designed in the presence of unkno wn dynamics only in its dynamic model. An adaptiv e tr ajector y-tr ac king con- troller based on the robot dynamics is proposed with e xper imental results in [17], and its stability is pro v ed using the L y apuno v stability theorem, The dynamic controller is capab le of updating the estimated par ameters , which are directly related to ph ysical par ameters of the robot. In order to o v ercome tr aj ector y tr ac king prob lems , in [20] an adaptiv e nonlinear control of a wheeled mobile Receiv ed F ebr uar y 2, 2016; Re vised Ma y 9, 2016; Accepted Ma y 20, 2016 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 479 robot approach is proposed in the presence of uncer tainties . A methodo logy based on linear in- ter polation is used in [21] to design control algor ithms f or tr ajector y tr ac king of mobile robots ,The proposed control can be applied to the design of a large class a mobile robot. In this paper , the tr ajector y tr ac king prob lem f or an unicycle-lik e mobile robot has been addressed. First, the dynamic model of the unicycle-lik e mobile is presented and the kinematic controller , which is based on the robot kinematics , is introduced to gener ate the desired linear and angular v elocities f or the giv en tr ajector y . The control la w is de v eloped using bac kstepping technique to gener ate the commands of linear and angular v elocities deliv ered to the robot ser v os , and its stability proper ty is pro v ed using the L y apuno v theor y . The rest of the paper is organiz ed as f ollo ws . Section 2 presents the dynamic unicycle-lik e robot model. The kinematic controller is detailed and the complete equations of the bac kstepping controller design a re de v eloped in section 3. Respectiv ely , in section 4, sim ulations and e xper i- mental results are discussed. Finally , section 5 concludes this paper . 2. Dynamic model In this w or k, the dynamic model of the unicycle-lik e mobile is considered, this model is proposed in [22], The mobile robot is illustr ated in Figure 1. G is the center of mass of the robot, C is the position of the castor wheel, E is the location of a tool onboard the robot, h is the point of interest with coordin ates x and y in the XY plane ,   is the robot or ientation, and a is the distance betw een the point of interest and the centr al point of the vir tual axis linking the tr action wheels (point B ), u and ! are the linear and angular v elocities of the robot. Figure 1. The unicycle-lik e mobile robot. The model of the mobile robot can be obtained as f ollo ws [22]: 0 B B B B @ _ x _ y _   _ u _ ! 1 C C C C A = 0 B B B B @ u cos(   ) aw sin(   ) u sin(   ) + aw cos(   ) ! 3 1 ! 2 4 1 u 5 2 u! 6 2 ! 1 C C C C A + 0 B B B B @ 0 0 0 0 0 0 1 1 0 0 1 2 1 C C C C A u r ef ! r ef (1) The par ameters of the dynamic model are: i ; i = 1 ::::; 6 and defined as f ollo ws: Bac kstepping Approach f or A utonomous Mobile Robot T r ajector y T r ac king (Ibar i Benaoumeur) Evaluation Warning : The document was created with Spire.PDF for Python.
480 ISSN: 2502-4752 8 > > > > > > > > > > > < > > > > > > > > > > > : 1 = R a k a ( mR t r +2 I e )+2 r k D T (2 r k P T ) 2 = R a k a ( I e d 2 +2 R t r ( I z + mb 2 ))+2 r dk D R (2 r dk P R ) 3 = R a k a mbR t 2 k P T 4 = R a k a ( k a k b R a + B e ) ( r k P T ) + 1 5 = R a k a mbR t 2 k P R 6 = R a k a ( k a k b R a + B e ) d ( r k P R ) + 1 (2) where m is the robot mass , I z is the r obot moment of iner tia at G, r is the r ight and left wheel r adius , I e and B e are the moment of iner tia and the viscous fr iction coefficient of th e combined motor rotor , gearbo x, and wheel, and R t is the nominal r adius of the tire , k b is electromotiv e constant of motors , k a is the constant of torque . R a is the electr ical resistance of the motors , b and d are the distances . The robot ha v e PD controllers to control the v elocities of each motor , with propor tional gains k P T and k P R , and der iv ativ e gains k D T and k D R . 3. Rob ust contr oller design In this w or k, tw o diff erent types of controllers are considered : kinematic controller f or e xter nal loop and a bac kstepping controller f or an inter nal loop as see in figure 2. 3.1. Kinematic contr oller F or the giv en tr ajector y , the desired v alues of the linear and angular v elocities are gener- ated b y the kinematic controller , it is based on the kinematic model of the robot. The kinematic equations of mobile robot in Figure 1 are descr ibed b y: _ x _ y = A u r ef ! r ef (3) with A = cos (   ) asin (   ) sin (   ) acos (   ) (4) whose respectiv ely u r ef ; ! r ef are the desired v alues of the linear and angular v elocities and h ( x; y ) is the point of interest. whose in v erse is A 1 = cos (   ) sin (   ) 1 a sin (   ) 1 a cos (   ) (5) Thus , choosing the control la w u k r ef ! k r ef = cos   sin   1 a sin   1 a cos   _ x d + x _ y d + y (6) where x = x d x , y = y d y are the current position errors , h ( x; y ) and h d ( x d , y d ) are the current and the desired coordinates . No w , consider the position error _ x + x _ y + y = 0 0 (7) Thus , choosing the L y apuno v candidate function V = 1 2 T ) _ V = T _ < 0 (8) IJEECS V ol. 2, No . 3, J une 2016 : 478 485 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 481 where = _ x + x _ y + y T ,theref ore , from (7) and (8) it results that: _ V = T x _ x = T x x < 0 T y _ y = T y y < 0 (9) is negativ e definite . Clear ly , if goes to z ero then h con v erges to h d and the tr ac king error is asymptotically stab le , which means that ! 0 as t ! 1 Theref ore , (6) is a satisf actor y tr ac king controller f or system (3). 3.2. Bac kstepping Contr oller design and stability anal ysis The controlle r receiv es from the kinematic controller the ref erences f or linear and angular v elocities , and gener ates another pair of linear and angular v elocities to be deliv ered to the robot ser v os , as sho wn in Figure 2. The bac kstepping control is used to design the controller which is making the error dynamics Figure 2. Control str ucture . stab le [23, 24]. The design of the bac kstepping controller is ba s e d on the ly apuno v theorem, the objectiv e of this technique is to deter mine a control la w that pro vides the system stability . The dynamic par t of equation (1) is: ( _ u = 3 1 ! 2 4 1 u + u r ef 1 _ ! = 5 2 u! 6 2 ! + ! r ef 2 (10) In the first step , the output error betw een the ref erence and actual controls is giv en b y: 1 u = u r ef u = ) _ 1 u = _ u r ef _ u (11) 1 ! = ! r ef ! = ) _ 1 ! = _ ! r ef _ ! (12) Choose a L y apuno v functions candidate as V ( 1 u ) = 1 2 2 1 u ; V ( 1 ! ) = 1 2 2 1 ! (13) The time der iv ativ e of the L y apuno v candidate functions can be wr itten as _ V ( 1 u ) = 1 u _ 1 u ; _ V ( 1 ! ) = 1 ! _ 1 ! (14) Bac kstepping Approach f or A utonomous Mobile Robot T r ajector y T r ac king (Ibar i Benaoumeur) Evaluation Warning : The document was created with Spire.PDF for Python.
482 ISSN: 2502-4752 The stability of the equilibr ium at the or igin of the errors system can be obtained b y: _ u r ef _ u = K 1 u 1 u _ ! r ef _ ! = K 1 ! 1 ! (15) where K 1 u > 0 ; K 1 ! > 0 are design par ameters . and a vir tual controls la w is defined b y: u u 0 = _ u r ef + K 1 u 1 u u ! 0 = _ ! r ef + K 1 ! 1 ! (16) F rom (14) and (15) it f ollo ws that ( _ V ( 1 u ) = K 1 u 2 1 u < 0 _ V ( 1 ! ) = K 1 ! 2 1 ! < 0 (17) At the second step the ne w errors giv en b y: 2 u = _ u u u 0 = ) _ 2 u = u _ u u 0 2 ! = _ ! u ! 0 = ) _ 2 ! = ! _ u ! 0 (18) F rom (16) and (18) it f ollo ws that: _ 2 u = u u r ef K 1 u _ 1 u _ 2 ! = ! ! r ef K 1 ! _ 1 ! (19) The augmented L y apuno v functions are giv en b y: ( V ( 1 u ; 2 u ) = ( 2 1 u + 2 2 u ) 2 V ( 1 ! ; 2 ! ) = ( 2 1 ! + 2 2 ! ) 2 (20) Its time der iv ativ e is then: _ V ( 1 u ; 2 u ) = 1 u _ 1 u + 2 u _ 2 u _ V ( 1 ! ; 2 ! ) = 1 ! _ 1 ! + 2 ! _ 2 ! (21) No w , after substituting (11),(12) and (18) in (21) it results that: _ V ( 1 u ; 2 u ) = 1 u 2 u K 1 u 2 1 u + 2 u ( u u r ef K 1 u ( K 1 u 1 u 2 u )) _ V ( 1 ! ; 2 ! ) = 1 ! 2 ! K 1 ! 2 1 ! + 2 ! ( ! ! r ef K 1 ! ( K 1 ! 1 ! 2 ! )) (22) In order to satisfy the L y apuno v Condition _ V ( 1 u ; 2 u ) < 0 and _ V ( 1 ! ; 2 ! ) < 0 , the controls la w is defined b y : u u = 1 ( u r ef ( K 1 u + K 2 u ) 2 u (1 K 2 1 u ) 1 u ) 2 3 ! _ ! + 4 _ u u ! = 2 ( ! r ef ( K 1 ! + K 2 ! ) 2 ! (1 K 2 1 ! ) 1 ! ) + 5 ( _ u! + u _ ! ) + 6 _ ! (23) In such a w a y that: _ V ( 1 u ; 2 u ) = K 1 u 2 1 u K 2 u 2 2 u < 0 _ V ( 1 ! ; 2 ! ) = K 1 ! 2 1 ! K 2 ! 2 2 ! < 0 (24) V along the tr ajector ies is negativ e definite . This pro v es the asymptotic stability of the r ac king tr ajector y of the wheeled mobile robot, which allo ws v er ifying the stability of the equilibr ium at the or igin of the error system. IJEECS V ol. 2, No . 3, J une 2016 : 478 485 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 483 Figure 3. Arduino robot mobile . −1 −0.5 0 0.5 1 −1.5 −1 −0.5 0 0.5 1 1.5 x [m] y [m]     Reference Actual (a) Robot T r ajector y 0 20 40 60 80 100 −0.1 0 0.1 0.2 0.3 u [m/s]     Reference Actual 0 20 40 60 80 100 0 0.5 1 1.5 time [s] ω  [rad/s]     Reference Actual (b) The Linear V elocity and Angular V elocity . 0 20 40 60 80 100 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 time [s] error [m] (c) Distance errors f or e xper iments . 0 20 40 60 80 100 −1 −0.5 0 0.5 1 x [m]     0 20 40 60 80 100 −2 −1 0 1 2 time [s] y [m] Reference Actual (d) Time e v olution of current and the desired coordinates . . Figure 4. Exper imental results 4. Result and Anal ysis In this section, e xper iments and sim ulations w ere carr ied out at Labor ator y of P o w er Sys- tems , Solar Energy and A utomation L.E.P .E.S .A, Univ ersity of sciences and technology of Or an, Bac kstepping Approach f or A utonomous Mobile Robot T r ajector y T r ac king (Ibar i Benaoumeur) Evaluation Warning : The document was created with Spire.PDF for Python.
484 ISSN: 2502-4752 Or an, Alger ia. The proposed bac kstepping controller is tested on a Arduino Robot Mobile (Radius of 185 mm, height of 85 mm and w eight of 0.150 kg), see Figure 3, Arduino Robot has tw o pro- cessors based on the A Tmega32u4, which admits linear and angular v elocities as input ref erence signals and it uses tw o DC motor dr iv en wheels . The robot is wirelessly connected to computer with Arduino Y un, these modules can comm unicate point to point, from one point to a PC , or in a mesh netw or k. Card arduino Y un mounted on the robot that is used to tr ansmit data from matlab to the robot using the Wireless netw or k protocol. 4.1. Discussion Exper imental results f or the bac kstepping controller proposed in sectio n 3 are sho wn in figure 4. The task f or a mobile robot is to f ollo w a circular tr ajector y , the mobile robot star ts from the initial p osture P 0 ( x; y ; ) = (1 ; 0 : 5 ; 0 ) , it can be seen that the robot f ollo ws the ref erence tr a- jector y with small e rror as see in figure 4a. At t=20 s , it can be seen in figure 4c that the distance error begins to increase with time , and tends to z ero . The robot v elocities v and ! are plotted in Figure 4b, and asymptotically tr ac ks their desired ref erence , there are used to obtain the r ight and left v elociities ( ! l ; ! r ). F rom Figure 4d, it can be seen that robo t arr iv es at the end of the ref erence tr ajector y and catches up to the desired coordinates . Unlik e the af orementioned w or ks , especially in compar ison with [17, 20], in this method, the tr ac k- ing errors are v er y lo w and the system is guar anteed to be stab le . 5. Conc lusion In this w or k, a re vie w of the kinematics and dynamics of a dierential dr iv e wheeled mobile robot w as giv en and a bac kstep ping controller f or a unicycle-lik e mobile robot w as also studied and tested. The control la w w as created b y giving the robot f orw ard and angular ref erence v eloc- ities , collecting the actual v elocities . The stability analysis based on L y apuno v theor y sho ws the eff ectiv eness of the control la w which are the tr ajector y tr ac king and the stability maintaining of the closed loop dynamics of the mobile robot. The analysis of results sho ws the good perf or mance of the proposed controller f or tr ajector y tr ac king when applied to an e xper imental mobile robot in ter m of accur acy , stability and con v ergence . The tr ac king errors are v er y lo w with respect to the mobile robot dimensions . The proposed controller allo w ed f or use on diff erent robotic platf or ms and it can be tested in an en vironment with obstacles . In future w or k, w e intend to impro v e the mobile robot b y using the augmented reality [25], This w or k can be applied also to remote control using the vir tual reality [26]. Ref erences [1] T . Lozano-P erez, I. J . Co x, and G. T . Wilf ong, A utonomous robot v ehicles . Spr inger Science & Business Media, 2012. [2] P . Antonini, G. Ippoliti, and S . Longhi, “Lear ning control of mobile robots using a m ultiproces- sor system, Control Engineer ing Pr actice , v ol. 14, no . 11, pp . 1279–1295, 2006. [3] M. Corr adini and G. Or lando , “Control of mobile robots with uncer tainties in the dynamical model: a discrete time sliding mode approach with e xper imental results , Control Engineer ing Pr actice , v ol. 10, no . 1, pp . 23–34, 2002. [4] X. F eng, S . V elinsky et al. , “De v elopment of a distr ib uted m ultiple mobile robot control sys- tem f or a utomatic highw a y maintenance and constr uction, in Circuits and Systems , 1997. Proceedings of the 40th Midw est Symposium on , v ol. 1. IEEE, 1997, pp . 489–492. [5] C . C . De Wit, H. Khennouf , C . Samson, and O . J . Sordalen, “Nonlinear control design f or mobile robots , Recent trends in mobile robots , v ol. 11, pp . 121–156, 1993. [6] C . Samson, “Time-v ar ying f eedbac k stabilization of car-lik e wheeled mobile robots , The In- ter national jour nal of robotics research , v ol. 12, no . 1, pp . 55–64, 1993. [7] R. Carelli and E. O . F reire , “Corr idor na vigation and w all-f ollo wing stab le control f or sonar- based mobile robots , Robotics and A utonomous Systems , v ol. 45, no . 3, pp . 235–247, 2003. IJEECS V ol. 2, No . 3, J une 2016 : 478 485 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 485 [8] F . K ¨ unhe , J . Gomes , and W . F etter , “Mobile robot tr ajector y tr ac king using model predictiv e control, in II IEEE latin-amer ican robotics symposium , 2005. [9] W . W . H. C . Y . W ang and P . W oo , “Adaptiv e e xponential stabilization of mobile robots with uncer tainties , 1999. [10] T .-S . Jin and H.-H. T ac k, “P ath f ollo wing control of mobile robot using ly apuno v techniques and pid cntroller , Inter national Jour nal of Fuzzy Logic and Intelligent Systems , v ol. 11, no . 1, pp . 49–53, 2011. [11] T . Das and I. N. Kar , “Design and implementation of an adaptiv e fuzzy logic-based controller f or wheeled mobile robots , Control Systems T echnology , IEEE T r ansactions on , v ol. 14, no . 3, pp . 501–510, 2006. [12] Y . Cai, Q. Zhan, and X. Xi, “P ath tr ac king control of a spher ical mobile robot, Mechanism and Machine Theor y , v ol. 51, pp . 58–73, 2012. [13] G. Guoqin, R. Y i, Z. Haiy an, and F . Zhiming, “Smooth sliding mode control f or tr ajector y tr ac king of g reenhouse spr a ying mo bile robot, TELK OMNIKA Indonesian Jour nal of Electr i- cal Engineer ing , v ol. 11, no . 2, pp . 642–652, 2013. [14] L. Gr acia, F . Garelli, and A. Sala, “Integ r ated sliding-mode algor ithms in robot tr ac king appli- cations , Robotics and Computer-Integ r ated Man uf actur ing , v ol. 29, no . 1, pp . 53–62, 2013. [15] M. Ashoor ir ad, R. Barzamini, A. Afshar , and J . Jouzdani, “Model ref erence adaptiv e path f ollo wing f or wheeled mobile robots , in Inf or mation and A utomation, 2006. ICIA 2006. Inter- national Conf erence on . IEEE, 2006, pp . 289–294. [16] E. Canigur and M. Ozkan, “Model ref erence adaptiv e control of a nonholonomic wheeled mobile robot f or tr ajector y tr ac king, in Inno v ations in Intelligent Systems and Applications (INIST A), 2012 Inter national Symposium on . IEEE, 2012, pp . 1–5. [17] F . N. Mar tins , W . C . Celeste , R. Carelli, M. Sarcinelli-Filho , and T . F . Bastos-Filho , “An adap- tiv e dynamic controller f or autonomous mobile robot tr ajector y tr ac king, Control Engineer ing Pr actice , v ol. 16, no . 11, pp . 1354–1363, 2008. [18] R. Fierro and F . L. Le wis , “Control of a nonholonomic mobile robot: bac kstepping kinematics into dynamics , in Decision and Control, 1995., Proceedings of the 34th IEEE Conf erence on , v ol. 4. IEEE, 1995, pp . 3805–3810. [19] G. Y anf eng, Z. Hua, and Y . Y anhui, “Bac k-stepping and neur al netw or k control of a mobile robot f or cur v ed w eld seam tr ac king, Procedia Engineer ing , v ol. 15, pp . 38–44, 2011. [20] J . T aher i-Kalani and M. Khosro wjerdi, “Adaptiv e tr aject or y tr ac king control of wheeled mo- bile robots with disturbance obser v er , Inter national Jour nal of Adapt iv e Control and Signal Processing , v ol. 28, no . 1, pp . 14–27, 2014. [21] G. Scaglia, A. Rosales , L. Quintero , V . Mut, and R. Agarw al, “A linear-inter polation-based controller design f or tr ajector y tr ac king of mobile robots , Control Engineer ing Pr actice , v ol. 18, no . 3, pp . 318–329, 2010. [22] C . De La Cr uz and R. Carelli, “Dynamic modeling and centr aliz ed f or mat ion control of mobile robots , in IEEE Industr ial Electronics , IECON 2006-32nd Ann ual Conf erence on . IEEE, 2006, pp . 3880–3885. [23] W . Y . Qiao , “Bac kstepping adaptiv e fuzzy scheme f or scar a g rb400 robot, TELK OMNIKA Indonesian Jour nal of Electr ical Engineer ing , v ol. 11, no . 8, pp . 4229–4237, 2013. [24] L. Chr if , Z. M. Kada, T . Mohamed, and N. Bastaoui, “Flight-path tr ac king control of an aircr aft using bac kstepping controller , TELK OMNIKA Ind onesian Jour nal of Electr ical Engineer ing , v ol. 15, no . 2, 2015. [25] B . Ibar i, K. Bouzgou, Z. Ahmed-F oitih, and L. Benchikh, “An application of augmented reality (ar) in the manipulation of f an uc 200ic robot, in Inno v ativ e Computing T echnology (INTECH), 2015 Fifth Inter national Conf erence on . IEEE, 2015, pp . 56–60. [26] B . Ibar i, Z. Ahmed-F oitih, and H. E. A. Reda, “Remote control of mobile robot using the vir tual reality , Inter national Jour nal of Electr ical and Computer Engineer ing , v ol. 5, no . 5, 2015. Bac kstepping Approach f or A utonomous Mobile Robot T r ajector y T r ac king (Ibar i Benaoumeur) Evaluation Warning : The document was created with Spire.PDF for Python.