ower mobility parallel manipulators having fewer than six degrees of freedom (DOF) continue to draw interest from both industry and academia because they regularly offer improved balance between speed, accuracy, rigidity and reconfigurability when compared with conventional machine tools and industrial robots having serial architectures. Velocity, accuracy, stiffness and rigid body dynamic behaviours are important performance factors that"s should be considered in the design of lower mobility parallel manipulators. Particularly in circumstances where high speed is the priority, rigid body dynamics become a major concern for the dynamic manipulability evaluation, motor sizing and controller design, all of which involve acceleration analysis as the prerequisite. Although general, systematic approaches are available for velocity analysis of lower mobility parallel manipulators using either kinematic influence coefficient methods or screw theory based methods (Huang et al., 2000;Joshi & Tsai, 2002;Jhu et al., 2007) it is by no means an easy task to use these approaches for acceleration analysis owing to the nonlinearity arising from the second order partial derivatives. A number of approaches have been proposed for acceleration analysis of either serial or parallel manipulators. The most straightforward method is to take time derivatives of a set of velocity constraint equations. This, however, involves a tedious and laborious process as shown by many case-by-case studies (Tsai, 2000;Khalil & Guegan, 2004;Li et al., 2005;Callegari et al., 2006). Therefore, a recursive matrix method was proposed in order to reduce computational burdens (Staicu & Zhang, 2008;Staicu, 2009). Having a goal of achieving a general and compact form of the Hessian matrix, the kinematic influence coefficient method was proposed for dealing with the acceleration analysis of serial manipulators (Thomas & Twsar, 1982). This idea was then extended to cover full and lower mobility parallel manipulators (Huang, 1985a;Huang, 1985b;Zhu, 2005;Huang, 2006;Zhu et al., 2007). Along this track, kinematic analysis of a number of parallel manipulators with different architectures has been carried out (Fang & Huang, 1997;Lu, 2006;. Despite the plausibility and merits of the kinematic influence coefficient method, only an implicit form of the Hessian matrix can be achieved because of the unavoidable partial derivative implementations. Recently, an approach for acceleration analysis was proposed that introduced an auxiliary Hessian matrix derived from the differentiation of the auxiliary Jacobian of a class of parallel manipulators containing a passive properly constrained limb (Lu & Hu, 2007a;Lu & Hu, 2007b;. Its suitability for other types of parallel manipulators, however, remains an issue to be investigated. Screw theory based approaches (Hunt, 1978;Mohamed & Duffy, 1985;Kumar, 1992;Ling & Huang, 1995;Bonev et al., 2003;Fang & Tsai, 2003;Zoppi, 2006) could potentially be the most powerful method for acceleration analysis. In order to overcome the difficulty of expressing the twist derivatives in a screw form, a novel term named the "accelerator" (Sugimoto, 1990) or "acceleration motor" (Brand, 1947) was proposed and employed for the acceleration analysis of serial and parallel kinematic chains (Rico & Duffy, 1996;Rico & Duffy, 2000). However, the terms associated with the second derivatives in the acceleration equations can only be written in a lengthy form of Lie brackets rather than in a compact form of the Hessian matrix. # L Introduction achieve an explicit and compact form of the Hessian matrix. Having outlined in Section I the significance of acceleration analysis and its existing problems, the paper is organized as follows. Sections II and III systematically develop the formulations of forward/inverse velocity and acceleration models of serial and parallel kinematic chains, leading to new expressions of the Hessian matrices in a general and compact form. A practical illustration is presented in Section IV before conclusions are drawn in Section V. Based upon the authors" previous work (Huang et al., 2011), this section briefly addresses the velocity analysis of an f-DOF parallel manipulator using the "generalized Jacobian" and adds extensions necessary to its use in acceleration analysis. Without loss generality, assume that the manipulator is composed of l ( 1 f l f ? ? ? ) limbs connecting the platform with the base, each essentially containing i n ( 1, 2, , il ? ? ) 1-DOF joints with at most one of them actuated. Thus, two families of parallel manipulators can be taken into account. The first family covers fully parallel manipulators having f constrained active limbs ( 6 i n ? for all limbs). The second family contains those having f unconstrained active limbs (i.e. 6 i n ? for each of these f limbs) plus one properly constrained passive limb designated by 1 lf ?? . Any other parallel manipulators not belonging to these two families can be dealt with in a manner similar to that used below. It has been shown (Huang et al., 2011) that entire set of the variational twists of the platform spans a 6-dimensional vector space T , known as the twist space. As the dual space of T , the entire set of wrenches exerted on the platform spans a 6dimensional vector space W, named the wrench space. For an f-DOF manipulator, T can be decomposed into an f dimensional subspace, a ? TT , and a 6-f dimensional subspace, c ? TT , known respectively as the twist subspaces of permissions and restrictions. Correspondingly, W can also be decomposed into two subspaces, a ? WW and c ? WW , known as the wrench subspaces of actuations and constraints. It has been proved that the following commutative relationships hold: Direct sum: ac ?? T T T , ac ?? W W W (1a) Orthogonality: ac ? ? WT, ca ? ? WT(1b) Duality: * aa ? WT , * cc ? WT (1c) a) Velocity analysis of a limb Let , , , ?a ta j i a i ? $ T ( 1,2, , a i j n ? ? ), , , , ?a wa k i a i ? $ W ( 1,2, , a i k n ? ? ), , , , ?c tc j i c i ? $ T ( 1,2, ,6 c i j n ? ? ? ) and , , , ?c wc k i c i ? $ W ( 1,2, ,6 c i k n ? ? ? ) be the basis elements of four vector subspaces associated with the ith limb. Note that the commutative relationships given in Eq. (1a, b, c) also hold for each limb since all limbs share the same platform. The variational twist of the platform can then be represented by a linear combination of the basis elements of , ai T and , ci T ,,6 , , , , , , , , 1 1 ?îi a a c c a c t ta tc ta i tc i n n a j i ta j i c j i tc j i j j ii ?? ?? ? ? ? ? ? ? ? ? ? ? ? ?? J ? $ $ $ $ $ $ $ , 1, 2, , il ? ?(2) where, ? ? T TT t ? ? =r ? $ ,, J J J i a i c i ?? ? ?? ,, J J J i a i c i ?? ? ?? , ,1, , , ??i ai ta i ta n i ?? ? ?? J ? $$ , ,1, ,6 , ?? J i c i tc i tc n i ? ?? ? ?? $$ ? ? ? T TT ,, i a i c i ? ? ? ? ? ? ? ? ? T , ,1, ,2, , , i ai a i a i a n i ? ?? ?? ?? ? ? ? ? ? T , ,1, ,2, ,6 ,i c i c i c i c n i ? ?? ?? ?? ? ? ? ? ?r and ?? denote, respectively, the linear variation of the reference point and the angular variation of the platform. ? ? ? ? T T T T T T , , , i a i c i i a i ? ? ? ? ? ? ? ? ? ? ? ?? 0 ,, a i a i ? ? ? ? ? , ci ? ? 0 ? ? ? T , ,1, ,1, , , ( A ) ear 2012 Y © 2012 Global Journals Inc. (US) ? ? ? ? T T T T T T t t ? ? ? ? ? r ? v ? $ $ II. # Velocity Analysis Drawing primarily on the generalized Jacobian but also on the strengths of the kinematic influence coefficient method and the concept of accelerator, we propose a new approach for acceleration analysis of lower mobility parallel manipulators. Its goal is to where, ? i are individual joint velocities. Thus (Huang et al., 2011) t i i ? J ? ? $ , 1, 2, , il ? ?(3) where, Using the commutative relationships given in Eq. ( 1) and taking inner products on both sides of Eq. ( 2) with # ? ? ,, ?a wa k i $ ( 1, 2, , a i k n ? ? ) and ,, ?c wc k i $ ( 1,2, ,6 c i k n ? ? ? ) leads, after the same replacements, to 1 i i t ? ? ? J ? $ , 1, 2, , il ? ? (4) , 1 , L ai i L ci ? ?? ? ?? ?? ?? J J J TT ,1, ,1, ,1, , TT , , , , , , ???î ii wa i wa i ta i L ai wa n i wa n i ta n i ?? ?? ? ?? ?? ?? ?? J ? $ $ $ $ $ $ TT ,1,1 ,1 , TT ,6 , ,6 , ,6 , ???î ii wc wc tc L ci wc n i wc n i tc n i ? ? ? ?? ?? ? ?? ?? ?? ?? J ? $ $ $ $ $ $ Here the superscript L simply identifies that the matrix applies to a single limb. Thus, we have ,, L a i a i t ? ? J ? $ , 1, 2, , il ? ?(5) b) Velocity analysis of a parallel manipulator Building upon the work in Section 2.1, the velocity modeling of a parallel manipulator can be carried out with little extra effort. Let ,, ?k wa g k $ be the unit wrench of actuations associated with the one actuated joint, numbered k g , in the kth ( 1, 2, , k f ? ? ) limb and ,, ?c wc k i $ be the th c k ( 1,2, ,6 ck kn ?? ? ) unit wrench of constraints in the ith ( 1, 2, , i l ? ? ) limb. Again, using the commutative relationships given in Eq. ( 1), taking inner products on both sides of Eq. ( 2) with ,, ?k wa g k $ and ,, ?c wc k i $ , respectively, and making replacements as at Eq. ( 3) results in t ? Jq ? $(6) where, a c ?? ?? ?? J J J = 1 1 1 2 2 2 T T , ,1 , ,1 , ,1 T T , ,2 , ,2 , ,2 T T , , , , , , ????f f f wa g wa g ta g wa g wa g ta g a wa g f wa g f ta g f ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? J ? $ $ $ $ $ $ $ $ $ ,1 ,2 , c c c cl ?? ?? ?? ? ?? ?? ?? ?? J J J J ? TT ,1, ,1, ,1, TT ,2, ,2, ,2, , T T ,6 , ,6 , ,6 ,????î i i wc i wc i tc i wc i wc i tc i ci wc n i wc n i tc n i ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? J ? $ $ $ $ $ $ $ $ $ ? ? T T T a qq ?? =0 ? ? 12 T , ,1 , ,2 , , f a a g a g a g f q q q ? ? ? ? ? ? q In order to distinguish the joint rates of actuated joints from those of the passive joints, we use ,, k a g k q ? to represent the rate of actuated joint numbered k g in the kth limb. For convenience, ,, k a g k q ? will be simplified as k q ? in what follows. J is an ? ? ? ? 1 66 l i i fn ? ? ? ? ? matrix known as the generalized Jacobian of parallel manipulators with 6 f ? DOF. Since ? ? ? ? 1 66 l i i fn ? ? ? ? ? , the left pseudo-inverse of J exists. Using superscript P to identify explicitly platform terms, this leads to P t a a ? ?? J q J q ?? $ , ? ? 1 TT PP ac ? ? ?? ?? ?? J J J J J J(7) Substituting Eq. ( 7) into Eq. ( 4), i ? ? can then be expressed in terms of q ? 1 LP i i i ?? ?? ? J J q J q ? ?? , 1, 2, , il ? ? (8) where, 1 LP ii ?? ? J J J is a ? ? ? ? 1 66 l i i fn ? ? ? ? ? matrix. Furthermore, the linear map between a q ? and where, , ai ? ? is , , , L P LP a i a i a a a i a ?? ? J J q J q ? ?? , 1, 2, , il ? ? (9) ,, LP L P a i a i a ? J J J is a i nf ? matrix. Following the scheme in Section II, acceleration analysis will first be carried out on an -DOF i n limb, with the results then being extended to cover an f-DOF parallel manipulator. # a) Acceleration analysis of a limb Taking the variation of Eq. ( 2) and expressing the derivatives of screws in the form of Lie brackets as given in (Gallardo et al., 2003), yields 2 iii ? ?? AJ ? $ , 1, 2, , il ? ? (10) where, ? ? ? ? T TT 22 ? ? ? ? ?? ? ? ? ?? ?? Ar ? r ? ? ? ? ? T TT 2 2 2 ,, i a i c i ? ? ? ?? ? ?? ?? ? ? ? ? ? T 2 2 2 2 , ,1, ,2, , ,i a i a i a i a n i ? ? ? ? ? ? ? ? ? ? ? ? T 2 2 2 2 , ,1, ,2, ,6 ,i c i c i c i c n i ? ? ? ? ? ? ? ? ? ? ? ,1,,1, ,2, ,2, ,6 , ,6 , ,2, ,2, ,3, ,3, ,6 , ,6 , ,5 , ,5 , ,6 , ,6 , ???î ii i i i i i i a i ta i a i ta i c n i tc n i a i ta i a i ta i c n i tc n i c n i tc n i c n i tc n i ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? $ $ $ $ $ $ $ $ $ 2 ? r , 2 ? ? , 2 ,, a a j i ? ? , and $ denotes the Lie product (Gallardo, 2006). From the properties of the Lie product, i $ can also be written as , , , i a i ac i c i ? ? ? $ $ $ $(11) where, , ,1, ,2, ,1, ,2, ,1, , , ,1, , , ,2, ,3, ,2, , , ,2, , , ,2, , , , 1, , , , 1, , , ????î ii i i i i i i a i a i a i ta i ta i a i a n i ta i ta n i a i a i ta i ta n i a i a n i ta i ta n i a n i a n i ta n i ta n i ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ? ? ? ? ? $ $ $ $ $ $ $ $ $ $$ , ,1, ,1, ,1, ,1, ,1, ,6 , ,1, ,6 , ,2, ,1, ,2, ,1, ,2, ,6 , ,2, ,6 , , , ,1, , , ,1,????î i i i i i # ac i a i c i ta i tc i a i c n i ta i tc n i a i c i ta i tc i a i c n i ta i tc n i a n i c i ta n i tc i ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? $ $ $ $ $ $ $ $ $ $ $ , ,,6 , , , ,6 , ?î i i i a n i c n i ta n i tc n i ?? ?? ? ? ? ? ?? ? ? ? $ $ , ,1, ,2, ,1, ,2, ,1, ,6 , ,1, ,6 , ,2, ,3, ,2, ,3, ,2, ,6 , ,2, ,6 , ,5 , ,6 , ,5 , ,6 ???????îi i i i i i c i c i c i tc i tc i c i c n i tc i tc n i c i c i tc i tc i c i c n i tc i tc n i c n i c n i tc n i tc ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ? ? ? $ $ $ $ $ $ $ $ $ $ $ , i ni ? ? ? ? ? Then, Eq. ( 10) can be rewritten as 2T i i i i i ? ? ? ?? AJ ? ? H ? , 1, 2, , il ? ?(12) ,, , a i ac i i ci ?? ? ?? ?? HH H H 0 ,1, ,2, ,1, ,3, ,1, , , ,2, ,3, ,2, , , , , 1, , , ????????i i ii ta i ta i ta i ta i ta i ta n i ta i ta i ta i ta n i ai ta n i ta n i ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ?? ?? ? ? ? ? ? ? ? ? ?? ?? ? ?? ?? ?? ?? ?? ?? ?? H ? ? ? ? ? ? ? ? ? 0 $ $ $ $ $ $ 0 0 $ $ $ $ 0 0 0 $ $ 0 0 0 0 ,1, ,1, ,1, ,6 , , , , ,1, , , ,6 , ?????ii i i ta i tc i ta i tc n i ac i ta n i tc i ta n i tc n i ? ? ?? ? ? ? ? ? ? ? ? ?? ?? ? ?? ? ? ? ? ?? ? ? ? ? ?? H ? ? ? ? ? $ $ $ $ $ $ $ $ ,1, ,2, ,1, ,3, ,1, ,6 , ,2, ,3, ,2, ,6 , , ,5 , ,6 , ????????ii ii tc i tc i tc i tc i tc i tc n i tc i tc i tc i tc n i ci tc n i tc n i ? ? ?? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ?? ? ? ? ? ? ? H ? ? ? ? ? ? ? ? ? 0 $ $ $ $ $ $ 0 0 $ $ $ $ 0 0 0 $ $ 0 0 0 0 666 i ?? ? H ? is known as the Hessian matrix of the ith limb. It is a three-dimensional matrix having six layers, each containing a 66 ? upper triangular matrix as shown in Fig. 1, where ? ? i K ?? ( 1, 2, , 6 i K ? ? ) denotes the th i K element of the Lie bracket ? ? ?? . The constituent parts of i H , 6 , ii nn ai ?? ? H ? , ? ? 66 , ii nn ac i ? ? ? ? H ? and ? ? ? ? 6 6 6 , ii nn ci ? ? ? ? ? H ? , are also three-dimensional matrices having six layers. In acceleration analysis where only ideal motions of the platform are considered, replacements can be made in Eq. ( 12) such that: The bracket Thus, ? ? ? ? T T T T T T , , , i a i c i i a i ? ? ? ? ? ? ? ? ? ? ? ?? 0 ? ? ? ? ? ? T T T T 2 2 2 T , , , i a i c i i a i ? ? ? ?? ? ? ? ?? ?? ? ? ? ? ? ?? ?? 0 ? ? ? ? ? ? T T T TT T 22 ? ? ? ? ?? ?? ? ? ? ? ? ? ? ?? ??T i i i i i ?? AJ ? ? H ? ?? ? ? , 1, 2, , il ? ?(13) . limb with v? and ? ? being the linear acceleration of the reference point and the angular acceleration of the platform (Gallardo, 2003). The th i K element in A has the expression T ,, IIIi i i K i K i i i K i ?? AJ ? ? H ? ?? ? ? , 1, 2, , il ? ? (15) ( A ) ear 2012 Y © 2012 Global Journals Inc. (US) Fig. 1 : Hessian matrix, H i , of the i th limb. ,1, , ?0 0 0 0 0 0 0 i i i i i i i i i i ta i ta i ta i ta i ta i tc n i K K K ta i ta i ta i tc n i K K tc n i tc n i K ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? $ $ $ $ $ $ $ $ $ $ $ $ non-zero element zero element 1 2 3 4 5 6 1 2 3 4 5 6 1 i K 6 The K i th layer ( 1, 2, ,6 i K ? ? )2, ,1, ,3, ,1, ,6 , ,2, ,3, ,2, ,6 , ,5 , ,6 , ?????0 ???0 0 where, , i iK J is the th i K row of i J , while , i iK H is the th i K layer of i H . In addition, the inverse acceleration equation of the ith limb can easily be obtained, by recalling Eq. ( 4), as ? ? 1 T T 1 i i t i i i t ? ? ? ?? ? J A J H J ?? $$ , 1, 2, , il ? ? (16) Acceleration analysis of a parallel manipulator Following the same approach as at Eq. ( 6), the acceleration equation of a parallel manipulator is obtained by taking inner products on both sides of Eq. ( 13) with ,, ?k wa g k $ and ,, ?c wc k i $ , respectively, and noting the relationship given in Eq. ( 8), to give T ?? JA q q Hq ?? ? ?(17) where, ? ? T T a qq ?? ?? =0 ? ? 12 T , ,1 , ,2 , , f a a g a g a g f q q q ? ?? ?? ?? ?? ? q Here, ,, k a g k q ?? is the acceleration of the actuated joint numbered k g in the k th limb ( 1, 2, , k f ? ? ). For convenience, ,, k a g k q ?? will be simplified as k q ?? in what follows. NNN ?? ? H ? ( ? ? 1 6 l i i N f n ? ? ? ? ? ) is known as the Hessian matrix of an f-DOF parallel manipulator. It is a three dimensional matrix composed of f N N a ?? ? H ? and ? ? 1 6 l i i n N N c ? ? ? ? ? ? H ? . The expression for the th a K acceleration equation of a parallel manipulator: ? ? T ? ?? A J q q Hq ?? ? ?(19) Furthermore, substituting Eqs. ( 8) and ( 19) into Eq. ( 16) expresses the joint acceleration in the ith limb in terms of the velocity and acceleration of the actuated joints: ? ? ? ? T T 1 T LP LP LP i i i i i i ? ? ? ? ? J q q Hq J q J H J q ?? ?? ? ? ? ? , (20)1, 2, , il ? ? ( th c K ) layer of a H ( c H ) is given in Fig. 2. Eq. ( 17) readily yields the inverse acceleration equation of a parallel manipulator, TT q JA J HJ tt ?? $$ ??(18) Moreover, multiplying both sides of Eq. ( 17) by the left pseudo-inverse of J gives the forward The above analyses formulate the forward/inverse velocity and acceleration equations of lower mobility parallel manipulators in a consistent manner under the umbrella of the generalized Jacobian. The velocity and acceleration of each # DOF 1? actuated joint of the manipulator can be evaluated using Eqs. ( 8) and (20). Note, also, that the velocity and acceleration analyses given for a limb are also valid for serial manipulators with 6 f ? DOF. Detailed execution of velocity and acceleration analyses for a 3-PRS parallel manipulator serves to i. # IV. An xample E illustrate the generality and effectiveness of the proposed approach. Fig. 3 shows a schematic diagram of a 3-PRS parallel manipulator which is used as a 3-axis module named the Sprint Z3 (Wahl, 2002) as part of a 5-axis high-speed machining cell for extra large components. The manipulator consists of a base, a platform, and three identical limbs, each connecting the base with the platform in sequence by an actuated prismatic joint, a revolute joint, and a spherical joint. f ? ? 1 6 l i i f n ? ? ? ? c H a H ? ? T T T , ,,1, ,2, , , ,1, ,3, , , ,1, ,6 ,T T T , ,,2, ,3, , , ,2, ,6 , T , , , , , ????????0 ?????0 0??0 0 0 c c c i c c i c c # wc k i ta i ta i wc k i ta i ta i wc k i ta i tc n i LP wc k i ta i ta i wc k i ta i tc n i i wc k i tc k i wc k ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? J ? ? ? ? ? ? ? ? $ $ $ $ $ $ $ $ $ $ $ $ $ $ $ $$ $ T , ,5 , ,6 , ?0 0 0 0 c i i LP i i tc n i tc n i ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? J ? $ $ The K a th layer of H a ( Fig. 4 (a, b, and c) shows the CAD model of the selected example 3-PRS parallel kinematic machine, which helps to visualize and internalize the physical mechanism. 1, 2, , a K f ? ? ; a kK ? ) The K c th layer of H c 1 1, 2, ,6 c k n ? ? ? ( 1 1 1,2, ,6 , ,6l c i i K n n ? ? ? ? ? ? ? ;; 1, 2, , i l ? ? ) ? ? 1 6 l i i f n ? ? ? ? ? ? 1 6 l i i n ? ? ? ? ? T T T , ,,1, ,2, , , ,1, ,3, , , ,1, ,6 ,T T T , ,,2, ,3, , , ,2, ,6 , T , , , , , ????????0 ?????0 0??0 0 0 J k k k i k k i k k wa g k ta k ta k wa g k ta k ta k wa g k ta k tc n k LP wa g k ta k ta k wa g k ta k tc n k i wa g k ta g k wa g ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? $ $ $ $ $ $ $ $ $ $ $ $ $ $ $ $ $ $ ? ? ? ? ? ? ? ? T , ,5 , ,6 , ?0 0 0 0J k i i LP k k tc n k tc n k ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? $ $ ? 4-c 4-b # a) Inverse kinematics A reference frame R is attached to the base and a body fixed frame 0 R to the platform, with O and O? located at the centers of the equilateral triangles # B B B ? and 1 2 3 A A A ? , as shown. The z and 0 z axes are normal to the planes of those triangles, the x axis is parallel to 21 BB and the 0 x axis is parallel to 21 AA. Also, an instantaneous reference frame R? is set with its origin at point O? and its three orthogonal axes remaining always parallel to those of R . Consequently, the orientation matrix of 0 R with respect to R can be obtained using three Euler angles, ? , ? and ? in terms of precession, nutation, and body rotation according to the z-x-z convention ? ? ? ? ? ? Rot , Rot , Rot , C C S C S C S S C C S S S C C C S S S C C C C S S S S C C z x z ? ?? ? ? ? ? ?? ?? ? ? ? ? ? ?? ?? ?? R ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?(21) where, "S" and "C" represent "sin" and "cos", respectively. Then, the position vector, ? ? T x y z ? r , of O? can be expressed as The constraint imposed by the revolute joint restricts the translational motion of i A along the axis of the revolute joint in limb i. This leads to three additional constraint equations, 1, 3 3, r b s s a i i i i i ql ? ? ? ? , 1, 2,3 i ?(22) where, 1,s i i i i q B P ? ? ? T 1, 0 0 1 s i ? 3 3, s i i i l P A ? ? ? ? ? T T cos sin 0 b i ix iy iz i i b b b b ?? ?? ? ? T 0 a Ra i i ix iy iz a a a ?? ? ? T 0 cos sin 0 i i i a ? a ?? ? ? 11 2 1 63 i i ?? ? ? ? ? , 1, 2,3 i? ? T 2, 0 r a s ii ?? , 1, 2,3 i ?(23) where, ? ? T 2 ? ? 1 S2 1 C 2 xa ?? ? ? ?(25) ? ? 1 C2 1 C 2 ya ?? ? ? ?(26) Thus the three desired motions, ? , ? , and z , can be considered as three independent Cartesian variables, leaving the translations along the x and y axes, and rotation about the z? axis (angle? ) as the constrained variables. Given a set of ? , ? , and z , the inverse position problem is determined by , ? ? T 3 3, 1, # r a b s s i i i i i ql ? ? ? ? , 1, 2,3 i ? (27) ? ? T 3,3 , 3 , 3 ,s i x i y i z i s s s ? , ? ? 3, 3 1 x i ix ix s x a b l ? ? ? ? ? 3, 3 1 y i iy iy s y a b l ? ? ? ,22 3 , 3 , 3 ,1 z i x i y i s s s ? ? ? b) Velocity analysis Given the bases for the four vector subspaces of its ith limb (Huang et al., 2011), the generalized Jacobians of the ith limb and the generalized Jacobian of the manipulator can be formulated as follows. For the i th limb ( 1, 2,3 i ? ): , i a i ci ?? ? ?? J J J (28) ? ? , ,1, ,2, ,3, ,4, ,5, 1, 3 3, 2, 3, 4, 5, 2, 3, 4, 5,i i i i i i i i i i i i i l ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? $ $ $ $ $ 0 ? ? 3 3, 1, ,,1, 1, ? a s n J n i i i c i tc i i l ?? ?? ?? ?? ?? ?? # $ For the parallel manipulator: J J J a c ?? ?? ?? =(29)? ? ? ? ? ? ? ? ? ? ? ? ?? ? ?? ?? ?? ?? ?? ?? ? ?? where, Substituting Eq. ( 29) into Eqs. (6) and Eq. ( 7) generates the inverse and forward velocity equations of the manipulator qJ t ? $ ? (30)1 Jq t ? ? $ ?(31) where, ? ? T T T a qq ?? =0 and ? ? T 1 2 3 q= a q q q ? ? ? ? . # c) Acceleration analysis The Hessian matrix, H , of the manipulator can be found by substituting the expressions just found for ? ? T , , T 3 1, 3, J M J H ss c LP LP i c i i cK ii l ? , 1, 2,3 c Ki ??(33) where, 11 J J J LP ii ?? ? 1, , 2, M M M ai ai ai ?? ? ?? ?? 0 0 ? ? ? ? T T T 3, 1,3, 1, 4, 3, 1, 5, 1,i i i i i i i i ai i i i i ll ?? ?? ?? ? ?? ?? ? ? ? ? T 3 4, 1, 2, T3 5, 3, 3, 1, 0 0 snM s s s n ii ai i i i i l l ?? ?? ? ?? ? ?? ?? ?? ?? ?? 1, , M M ci ci ?? ? ?? ?? 0 00 T T T 3, 1, 4, 1, 5, 1, 1, TT 3 3 5, 3, 3 3, 1, 1 0 s n s n s n M s s s n i i i i i i ci i i i i l l l ?? ? ? ? ? ? ?? ? ? ? ?? ?? Here, , H a aK ( , H c cK ) represents the th a K ( th c K ) layer of a H ( c H ). Then, substituting Eqs. ( 32) and (33) into Eqs. ( 18) and ( 19), the inverse and forward acceleration equations of the manipulator are TT q JA J HJ tt ?? $$ ?? (34) ? ? 1T A J q q Hq ? ?? ?? ? ?(35) where, ? ? T T a qq ?? ?? =0 and ? ? T 1 2 3 q a q q q ? ?? ?? ?? ?? . # d) Coordinate transformation for numerical simulation Numerical simulations for the inverse velocity and acceleration, require the explicit relationships of the velocity twist and accelerator to the first and second derivatives of three independent coordinates, ? , ? , and z because they are used for path planning. Taking the time derivative of Eqs. ( 25) and ( 26) gives v r J g vc ?? ?? (36) ? ? ? ? C2 1 C 0.5 S2 S 0 S2 1 C 0.5 C2 S 0 0 0 1 v a ? ? a ? ? a ? ? a ? ? ?? ? ? ? ?? ? ? ? ?? ?? ?? J , g c ? ? z ?? ?? ? ?? ?? ?? ? ? ? ? Then, taking the time derivative of Eq. ( 36) results in T v J g g H g v c c v c ?? ? ?? ? ?(37) where, 333 H v ?? ? ? is a three dimensional matrix with , H vi ( 1, 2,3 i ? ) being its ith layer; g c ? ? z ?? ?? ? ?? ?? ?? ?? ?? ?? ?? , ? ? ,1 2 S2 1 C C2 S 0 C2 S 0.5 S2 C 0 0 0 0 H v a ? ? a ? ? a ? ? a ? ? ?? ?? ?? ? ? ? ?? ?? ?? ? ? ,2 2 C2 1 C S2 S 0 S2 S 0.5 C2 C 0 0 0 0 v a ? ? a ? ? a ? ? a ? ? ?? ? ?? ?? ?? ?? ?? H ,3 3 3 v ? ? H 0 The angular velocity vector of the platform, ? ? T x y z ? ? ? ? ? , can be derived by recalling, e.g. ? ? T 1 2 3 a q q q ? q ? ? ? ? and ? ? T 1 2 3 q a q q q ? ?? ?? ?? ?? can be evaluated using Eqs. (36)-( 40), (6), and (18). Consider now a specific system having the geometry: 250 mm a ? , 312 5 mm . b ? , and 3 540 mm l ? Also, assume as an example that the path and motion rules of the platform are: When the platform of the manipulator moves according to the preceding rules, the velocity/acceleration of the actuated joints, the linear velocity/acceleration of the reference point O? , and the angular velocity/acceleration of the platform versus time can be evaluated using the proposed approach. These results, shown in Fig. 4, have been verified by a CAD model of the manipulator. There was no discernable difference between the results obtained using this approach and the CAD software. This paper presents a general and systematic approach for the forward and inverse velocity and acceleration analysis of lower mobility parallel manipulators using screw theory. With this approach, the process of acceleration modeling of serial and parallel kinematic chains can be integrated into the unified framework of the generalized Jacobian. It results in a new Hessian matrix being developed in a general and compact form. This allows rigid body dynamic modeling of lower mobility manipulators to be integrated into a single mathematical framework. 3![Fig. 3 : Schematic diagram of Sprint Z3 head.](image-2.png "Fig. 3 :") 2![Fig. 2 : Hessian matrix, H, of a parallel manipulator.](image-3.png "Fig. 2 :") 4![Fig. 4 : (a) Shows the motor configuration, (b) The platform tilted with ?=20 0 about X c , and (c) The platform tilted with 20 0 about Y c .](image-4.png "Fig. 4 :") ![Journals Inc. (US)](image-5.png "") or, a i a i , ?? AJ? ??T , a i a i a i , , ? H ? ? ? ,1, 2, , il ? ?(14)where,? ?? ? T T ? ? ?? ?? Av ? v ? ? ??Tbecomes theaccelerator of ani n ?DOF * JAngeles Fundamentals of robotics mechanical systems: Theory, methods, and algorithms. 3 rd ed New York Springer-Verlag 2003 * Singularity of 3-DOF planar parallel mechanisms via screw theory IABonev DZlatanov CMGosselin ASME Journal of Mechanical Design 125 3 2003 * Vector and tensor analysis LBrand 1947 John Wiley and Sons New York * Dynamics modelling and control of the 3-RCC translational platform MCallegari MCPalpacelli MPrincipi Mechatronics 16 2006 * A dynamic analysis of a spatial manipulator to determine the payload weight IiiCrane CDDuffy J Journal of Robotic Systems 90 7 2003 * Kinematics of a threedegree-of-freedom in-parallel actuated manipulator mechanism. Mechanism and Machine Theory YFang ZHuang 1997 32 * Inverse velocity and singularity analysis of low-DOF serial manipulators YFang LWTsai Journal of Robotic Systems 20 4 2003 * Kinematics and singularity analyses of a 4-dof parallel manipulator using screw theory. Mechanism and Machine Theory JGallardo JMRico GAlici 2006 41 * Dynamics of parallel manipulators by means of screw theory. Mechanism and Machine Theory JGallardo JMRico AFrisoli DCheccacci MBergamasco 2003 38 * THuang HTLiu DGChetwynd 2011 * Generalized Jacobian analysis of lower mobility manipulators. Mechanism and Machine Theory 46 * Modeling formulation of 6-dof multi-loop parallel mechanisms ZHuang Proceeding of the 4th IFToMM International Symposium on Lingkage and Computer Aided Design Methods eeding of the 4th IFToMM International Symposium on Lingkage and Computer Aided Design Methods 1985a II * Modeling formulation of 6-dof multi-loop parallel mechanisms ZHuang Proceeding of the 4th IFToMM International Symposium on Lingkage and Computer Aided Design Methods eeding of the 4th IFToMM International Symposium on Lingkage and Computer Aided Design Methods 1985b II * The advanced spatial mechanism ZHuang YSZhao TSZhao 2006 The High Education Press Beijing * Kinematic geometry of mechanisms KHHunt 1978 Oxford University Press Oxford * Jacobian analysis of limited-DOF parallel manipulators SJoshi LWTsai ASME Journal of Mechanical Design 124 2 2002 * Inverse and direct dynamic modeling of Gough-Stewart robots WKhalil SGuegan IEEE Transactions on Robotics 20 4 2004 * Instantaneous kinematics of parallel-chain robotic mechanisms VKumar ASME Journal of Mechanical Design 114 9 1992 * Dynamic formulation and performance comparison of the 3-DOF modules of two reconfigurable PKMs-the Tricept and the TriVariant MLi THuang JPMei XMZhao DGChetwynd SJHu ASME Journal of Mechanical Design 127 6 2005 * Kinestatic analysis of general parallel manipulators SHLing MZHuang ASME Journal of Mechanical Design 117 12 1995 * Using CAD variation geometry and analytic approach for solving kinematics of a novel 3-SPU/3-SPU parallel manipulator YLu ASME Journal of Mechanical Design 128 5 2006 * Analyzing kinematics and solving active/constrained forces of a 3SPU+UPR parallel manipulator. Mechanism and Machine Theory YLu BHu 2007a 42 * Unified solving Jacobian/Hessian matrices of some parallel manipulators with n SPS active legs and a passive constrained leg YLu BHu ASME Journal of Mechanical Design 129 11 2007b * Unification and simplification of velocity/acceleration of limited-dof parallel manipulators with linear active legs. Mechanism and Machine Theory YLu BHu 2008 43 * Kinematic analysis of two novel 3UPPU I and 3UPU II PKMs YLu YShi BHu Robotics and Autonomous Systems 56 2008 * Conclusion References Références Referencias V * A direct determination of the instantaneous kinematics of fully parallel robot manipulators MGMohamed JDuffy Journal of Mechanisms, Transmissions, and Automation in Design 107 2 1985 * A mathematical introduction to robotic manipulation RMurray ZXLi SSastry 1994 FL, CRC, Boca Raton * An application of screw algebra to the acceleration analysis of serial chains. Mechanism and Machine Theory JMRico JDuffy 1996 31 * Forward and inverse acceleration analysis of in-parallel manipulator JMRico JDuffy ASME Journal of Mechanical Design 122 9 2000 * Inverse dynamics of the 3-PRR planar parallel robot SStaicu Robotics and Autonomous Systems 57 2009 * A novel dynamic modelling approach for parallel mechanisms analysis SStaicu DZhang Robotics and Computer-Integrated Manufacturing 24 2008 * Existence criteria for over constrained mechanisms: An extension of motor algebra KSugimoto ASME Journal of Mechanical Design 112 3 1990 * Dynamic modeling of serial manipulator arms MThomas DTwsar ASME Journal of Mechanical Design 104 9 1982 * Solving the inverse dynamics of a Stewart-Gough manipulator by the principle of virtual work LWTsai ASME Journal of Mechanical Design 122 3 2000 * Articulated tool head. US Patent 6431802 JWahl 2002 * SJZhu ZHuang HFDing 2007 * Forward/reverse velocity and acceleration analysis for a class of lower-mobility parallel mechanism ASME Journal of Mechanical Design 129 4 * SJZhu ZHuang XJGuo 2005 * Forward/reverse velocity and acceleration analyses for a class of lower-mobility parallel mechanisms Proceedings ASME Design Engineering Technical Conferences and Computers and Information in Engineering Conference ASME Design Engineering Technical Conferences and Computers and Information in Engineering Conference * On the velocity analysis of interconnected chains mechanisms. Mechanism and Machine Theory MZoppi DZlatanov RMolfino 2006 41