You are on page 1of 6

Proceedings of DETC ‘97

1997 ASME Design Technical Conferences
September 14-17, 1997, Sacramento, California

DETC97/DAC-3851
INVERSE KINEMATICS FOR PLANAR PARALLEL MANIPULATORS
Robert L. Williams II
Brett H. Shelley
Department of Mechanical
Engineering
Ohio University
Athens, Ohio

ABSTRACT
This paper presents algebraic inverse position and velocity
kinematics solutions for a broad class of three degree-of-freedom
planar in-parallel-actuated manipulators. Given an end-effector pose
and rate, all active and passive joint values and rates are calculated
independently for each serial chain connecting the ground link to the
end-effector link. The solutions are independent of joint actuation.
Seven serial chains consisting of revolute and prismatic joints are
identified and their inverse solutions presented.
To reduce
computations, inverse Jacobian matrices for overall manipulators are
derived to give only actuated joint rates. This matrix yields conditions
for invalid actuation schemes. Simulation examples are given.

planar parallel robot. Gosselin et.al. (1996) present the position,
workspace, and velocity kinematics of one planar parallel robot.
Recently, more general approaches have been presented.
Daniali et.al. (1995) present an in-depth study of actuation schemes,
velocity relationships, and singular conditions for general planar
parallel robots. Gosselin (1996) presents general parallel computation
algorithms for kinematics and dynamics of planar and spatial parallel
robots. Merlet (1996) solved the forward position kinematics problem
for a class of planar parallel robots (see Fig. 1).
A2
B2
C2
Y0

INTRODUCTION
Parallel manipulators are robots that consist of separate
serial chains that connect the fixed link to the end-effector link. The
following are potential advantages over serial robots: better stiffness
and accuracy, lighter weight, greater load bearing, higher velocities
and accelerations, and less powerful actuators. A major drawback of
the parallel robot is reduced workspace.
Parallel robotic devices were proposed over 17 years ago
(MacCallion and Pham, 1979). Some configurations have been built
and controlled (e.g. Sumpter and Soni, 1985). Numerous works
analyze kinematics, dynamics, workspace and control of parallel
manipulators (see Williams, 1988 and references therein). Hunt
(1983) conducted preliminary studies of various parallel robot
configurations. Cox and Tesar (1981) compared the relative merits of
serial and parallel robots.
These past works have focused on only a few different
architectures. For example, Aradyfio and Qiao (1985) examine the
inverse kinematics solutions for three different 3-dof planar parallel
robots. Williams and Reinholtz (1988a and 1988b) study dynamics
and workspace for a limited number of parallel manipulators.
Shirkhodaie and Soni (1987) , Gosselin and Angeles (1988), and
Pennock and Kassner (1990) each present a kinematic study of one

x,y
X0

φ
C1

A1

B3

C3
A3

B1

Figure 1. General Class of Manipulators
The current paper presents a purely algebraic approach to
solve inverse position and velocity kinematics for Merlet’s class of 3dof planar parallel manipulators. Geometric approaches have been
applied with success in the past for specific architectures (e.g.
Williams and Reinholtz, 1988b and Gosselin and Angeles, 1988). The
algebraic methods in this paper are suitable when one desires
kinematic solutions for control of all members in a broad class of
planar parallel manipulators. For instance, the current paper would
enable implementation of modular reconfigurable 3-dof planar parallel
manipulators.
This paper is organized as follows. First, the class of
manipulators is discussed.
Next the general inverse position

1

Copyright © 1997 by ASME

Table 1 reduces to seven chains. the angle at Bi is fixed. 2 is the angle relationship ( ψ i are defined in Fig. The passive joint values may also be determined for use in velocity. 1 and 2 must be solved three times.2. 1). one of the two binary links is variable. For an overall manipulator. Actuated Serial Chain Combinations x j θ +θ j θ +θ +θ E =   = Ai + L1i e jθ 1i + L2i e ( 1i 2i ) + L3i e ( i1 2i 3i ) y   The PPP chain is not used because only two planar P joints in a chain are independent. End-Effector Triangle Geometry P P 1 (2) 2 ψ1 R R R (1) The inverse kinematics problem requires three independent inverse solutions. The coupled transcendental Eqs. Kinematic Diagram for all Seven Serial Chains Table 1. Merlet. RRR RRP RPR RPP PRR PRP θ1i + θ2i + θ3i + π = φ + ψ i C PPR Table 2. the actuation scheme does not affect the solutions. PRP.y) L 3i C i θ3i φ L A RRP PRR PPR L 1i i 2i θ2i Bi θ1i Figure 3. For the inverse position problem of the current paper. the combinations RPP. The inverse position problem may be solved for each serial chain independently. for all seven chains in Fig. There are 343 (73) robots (the number is considerably lower if the ordering of distinct chains is not considered important). there is a revolute joint at Ai. Simulation examples are given for position and velocity. Each Ci is: P Cix = x + L3i cos(φ + ψ i ) P Ciy = y + L3i sin(φ + ψ i ) Figure 2. The solution is not dependent on joint actuation. One joint per chain is actuated and the remaining two are passive. Serial Chains When solving the forward position kinematics problem for this robot class the actuated joint in each serial chain must be specified. RRR RPR PRR RRR RPR PRR RRR RPR PRP RRP RPP PRP RRP RPP PPR The inverse position kinematics problem is stated: Given {x the end-effector pose y φ} . Therefore. Serial Chains for Inverse Kinematics C P R R P R R R R P R R P P R Y ψ2 ψ3 X C 3 Figure 4. i=1. and the overall inverse Jacobian is derived. one for each serial chain. The general inverse velocity solutions are presented. if joint k on chain i is prismatic. Underlining indicates actuation. there are three independent serial chains. 2. The active and/or passive joints may be revolute (R) and/or prismatic (P) (see Fig. dynamics. Seven Serial Chains 2 (3) Copyright © 1997 by ASME . singularity conditions are presented for each chain. each having seven possible configurations. E (x. The given manipulator pose fixes the position and orientation of the end-effector triangle in the plane. A general method is required to solve inverse kinematics for all. The kinematic diagram for the ith chain is Fig. acceleration. If serial chain i is RPR. 3. calculate the three actuated joint T (R or P) values. 2. 4). Any one of the three joints in a chain may be actuated unless the remaining passive joints are PP. in terms of independent solutions for all possible serial chains. Lki is variable. and there is a revolute joint at Ci. Three serial chains with 3-dof each connect the fixed link to the end-effector link.kinematics solutions are presented. (1996) identifies eighteen possible serial chains. Determination of invalid actuation schemes is presented. If joint k on chain i is revolute. given in Table 1.3. Equation 1 is the position vector equation for and Eq. CLASS OF MANIPULATORS INVERSE POSITION KINEMATICS SOLUTIONS Overall Manipulator The class of manipulators in this paper are planar and actuated in parallel. given in Table 2 and Fig. and PPR are omitted from Table I. The remaining terms are fixed. θki is variable.

θ3i . where E. where c12 = cos(θ 1i + θ 2i ) and s12 = sin(θ 1i + θ 2i ) . 1987). ( θ1i . θ2i . The remaining unknowns L1i and L2i are found from Eqs. Cix − Aix − L1i c11. Inverse position solutions are given below for the seven serial chains. The T (Cix − Aix )s1 + L2i s2 = (Ciy − Aiy )c1 ( )  Ciy − Aiy c1 − (Cix − Aix )s1   θ2i = sin −1    L 2 i   PRP Chain Given {x (14) (15) y φ} . The two solutions correspond to a shorter and longer L2i joint length. 4 are simplified to eliminate θ 1i . the three serial chains are solved independently. L1i and θ3i are solved using Eqs. θ3i . and G are Eq. Two solutions exist. A unique θ3i is solved with Eq. and θ3i from Eq. θ2i and L2i may now be evaluated from θ 2i = ζ i − θ 1i and Eq. RRR Chain {x Given E = Ciy − Aiy F = Aix − Cix y φ} . A in the unknowns L1i . 3 is used. PPR Chain Given {x y φ} . For each θ1i . Equation 2 is rewritten θ 2i = ζ i − θ 1i where ζ i is the constant L2i = (Cix − Aix )s12 − (Ciy − Aiy )c12 )     ) T 2 PRR Chain  − F ± E 2 + F 2 − G2 θ 1i1. the result is Eq. 12 and 13. sum of angles formulas are Overall Manipulator For the inverse position kinematics problem for all 343 robots. 1 to x and y components: evaluated from Eq. L2i is isolated in the x equation (Eq. determine L1i . L2i . E = L1i sζi F = − L1i cζi ) s2 Given T φ + ψ i − θ 3i − π . 10 and 2. F = 2 Ciy − Aiy L1i {x unique θ 1i is found from Eq. L2i . Equation 6 is solved for θ1i using the tangent half-angle substitution (Mabie and Reinholtz.into the y equation and greatly simplified to the Eq. Two values for θ1i ( −(Cix − Aix )s1 + Ciy − Aiy c1 (12) a near (Eq. 6. 8. 6) of the RPR. Cix − Aix − L1i c1 cζi s2 solution starts with Eq. 2. Eqs. a unique θ3i is found from Eq. L2i . θ2i ). 4 are then two linear equations θ2i is found by a y to x ratio of Eqs. where E. θ3i . 2 RRP Chain Given {x (7) are obtained using Eq. y φ} . determine L1i . L1i . 2 = a tan 2 Ciy − Aiy − L1i s11. The number of inverse position used. a unique RPP Chain Given L1i = E = 2(Cix − Aix ) L1i ( ) (5) ( G = L2i − L1i − (Cix − Aix ) − Ciy − Aiy 2 2 2 L2i = ( (6) θ2i1. 4. Eqs. A T unique θ2i is found from Eq. 2. 5 in (10) y φ} . 11 (and Eq. 5 form. In this case the variable to be solved is θ2i . 2. θ1i .θ 2i . and G are Eq. Eq. 2 . θ1i and T θ2i are constant so the unknowns L1i and L2i are found from Eqs. 12 and 13. L1i may now be Making use of Eq. θ2i . F. 10). 2 − θ1i1. squaring and adding yields Eq. determine θ 1i . not the constant θ1i . The T Ai terms of Eqs. 21) and far ( π − θ 2i ) solution. F. 7. θ3i . determine L1i . For each Ec1 + Fs1 + G = 0 (11) G = − L2i s2 T Isolating θ 1i + θ 2i terms in Eqs. There are two solutions Aix + L1i c1 + L2i c12 = Cix (4) Aiy + L1i s1 + L2i s12 = Ciy but one has negative L1i . 4. L2i . determine θ1i . (8) ( (9) ) G = Ciy − Aiy cζi + Aix − Cix RPR Chain Given {x {x (13) y φ} . 11. 5 form. determine θ1i . L2i with a unique solution. Equations 4 and L1i = Cix − Aix − L2i (c1c 2 − s1s2 ) c1 2 are three equations in three unknowns to solve for each of the three serial chains in a given manipulator. 3 and expanding Eq. 2 = 2 tan −1   G−E  y φ} . 4 are brought to the RHS. 2. 10. 8) and substituted into the y equation to give the Eq. 6. Given the commanded pose {x y φ} . 9. Then the unknown T joint parameters for each serial chain are solved by choosing the proper algorithm from above. given in Eq. θ2i . 2. and L1i is isolated in the x equation (Eq. L1i . Two θ1i are found from Eq. There are two solutions (elbow-up and elbow-down). determine θ1i . L1i is substituted 3 Copyright © 1997 by ASME . 6.

s123 = sin(θ 1 + θ 2 + θ 3 ) . calculate the three actuated joint (R or P) rates. Since Eqs.solutions is the product of the number of solutions (1 or 2) in each serial chain. c1 PRP  s1  0 INVERSE VELOCITY KINEMATICS SOLUTIONS The inverse velocity kinematics problem is stated: Given the end-effector rate command X = {x y ω z } T and all position variables. Jacobian matrices for each of the seven serial chains are: j12  L2 L3s3  − L3 ( L1s23 + L2 s3 ) L1( L2 s2 + L3s23 )  L3c23   − L3 ( L1s23 + L2 s3 )  − L1 − L2c2 − L3c23  s2  L1c2 + L2 + L3c3  − L1 − L2 c2 − L3c23  (19) th  j11 RRR  j21  1 − s12 1  s12 L1c2   L1c1  L2 s12 − L1s1 − L2 s12 L1s1 RPR The velocity relationship for each serial chain can be expressed: X (27) providing Ji has full rank. 1 and 2 (where the index i is dropped for notational convenience):  x  j (θ 1 +θ 2 )  j θ +θ jθ  jθ θ 1 + θ 2 + L 2 e ( 1 2 ) +    = L1 je 1θ 1 + L1e 1 + L2 je  y j θ +θ +θ j θ +θ +θ L3 je ( 1 2 3 ) θ 1 + θ 2 + θ 3 + L 3e ( 1 2 3 ) ( ( ) j13 = − L3s123 . 20-26) 1  j23 = L3c123 . 18 are written with respect to {0} coordinates. PRP. The inverse velocity solution for the ith serial chain is: ) x = − L1s1θ 1 + L1c1 − L2 s12 θ 1 + θ 2 + L2c12 − L3s123 θ 1 + θ 2 + θ 3 j13  j23  (Eqs. First it is assumed that L1 ≠ 0 and L2 ≠ 0 . RPP. R and/or P). The general translational and rotational velocities for all seven serial chains are: 1 c12  c1 c12 s12  PPR  s1 s12  0 0 0  j11 = − L1s1 − L2 s12 − L3s123 . 28-34) The singularity conditions (cases where Ji has less than full rank) for each chain are determined as follows. 16 into real and imaginary parts yields the x y and velocities. Jacobian matrices for each of the seven chains are determined from Eqs. These inverse Jacobians are presented symbolically below for the seven serial chains. Additionally. The RRR. Ji is j22 1 j13   j11 j23  RRP  j21  1 1  − L2 − L3c3   L1c2 + L2 + L3c3  L1( L2 s2 + L3s23 ) c12 − c12 L1s1 − c1 − L1s1 − L2 s12 c1 0  0 1 s c − 12 12 s2  − s1 c1  L2 c12 1  PRR − s1 L2 c2   s1 the i serial chain 3x3 Jacobian matrix. where: (16) ω z = φ = θ 1 + θ 2 + θ 3 j12 j22 j12 j22 1 c12   j11 c1 s12  RPR  j21 s1  1 0 0   j11 c1 c12  c1 RPP  j21 s1 s12  PRR  s1  0  1 0 0  j12 j22 1 j13  j23  1  j13  j23  1  PRP PPR L2 s12 c1 − c1  s12 − c12 1 0 0 s2  − s1 c1   L c − 3 23  L2 c2 + L3c23  L2 L3s3 L2 + L3c3   s2  − L2 c2 − L3c23   s12 − c12 1 − s1 c1 s2   0 0 L3c3   − L3c23  s2  (Eqs. ( ) ( ) ( y = L1c1θ 1 + L1s1 + L2 c12 θ 1 + θ 2 + L2 s12 + L3c123 θ 1 + θ 2 + θ 3 )  L2c12 1  L c − 1 1 − L2 c12 L1 L2 s2   L1c1  RRP where = {x y ω z } T s1  −1  Lc L c − 1 1 − 2 12 L1 + L2c2   − s1 ) (18) RPP is the Cartesian end-effector velocity. and ρ i is the vector of joint rates for the ith serial chain (one active and two passive. RRR Separating Eq. and PPR serial chains are all singular when s2 = 0 4 Copyright © 1997 by ASME . The inverse velocity problem may be solved for each serial chain independently. c123 = cos(θ 1 + θ 2 + θ 3 ) . 18 by zeroing terms that do not apply to a specific chain. The solution is not dependent on actuation scheme. ρ i = Ji−1 X (17) ( j12 = − L2 s12 − L3s123 . these correspond to degenerate robot chains. Velocity relationships for each independent serial chain i are obtained from the first time derivatives of Eqs. ω z = θ 1 + θ 2 + θ 3 X = Ji ρ i j22 = L2c12 + L3c123 . j21 = L1c1 + L2c12 + L3c123 . Note that L3i = 0 because L3i is part of the rigid triangle link. the passive joint rates may be determined for use in acceleration and dynamics. that is the frame of expression for X and Ji.

and Merlet (1989).PRR Robot (37) 10 15 20 Elbow-Down. For example. for more efficient real-time computation.PRR Robot ) Elbow-Up. This robot has 4 solutions (1x2x2). respectively. Sefriou and Gosselin (1995). Overall Manipulator The preceding section is sufficient to solve the inverse velocity problem. The manipulator overall inverse Jacobian matrix is given in Eq. (1996) where AΡ + BX = 0 . Therefore. (Note this method is equivalent to the rate kinematics in Gosselin et. for a 3-RPP robot (where j1=j2= j3=1): when L1 + L2 c2 = 0 . Equation 35 gives the general overall inverse Jacobian M:   x   ρ 1 j   Row( j1 ) st 1 chain   1    ρ 2 j2  =  Row( j2 )2 nd chain   y   ρ   Row( j ) rd  ω  3 3 chain   z   3 j3   ( Ρ = MX ) (35) where ρ iji is the actuated joint rate (joint ji) for the ith serial chain and Row( ji )i th chain is the jith row of the ith serial chain inverse Jacobian matrix Ji−1 . consider an RRP-PRP-RRR manipulator. π ). Near Solution of RPP. SIMULATION EXAMPLES An RPP-RRR-PRR robot is arbitrarily chosen to demonstrate the inverse position solutions. If M is invertible. are solved. Certain manipulators constructed of three identical chains will produce M matrices with columns of zero. M has rank one and is singular.3).” This rule. Invalid Serial-Chain Actuation Schemes L21 s21 + L31 s231 / c21   x    L2 2 + L32 c32 / s2 2   y  (36)  ( L2 s2 + L3s23 ) / L2 3 s23  ω z  ( (38) 5 10 15 20 0 Elbow-Down.RRR.2. Far Solution of RPP. the 3-RPP robot has an uncontrollable fourth degree-offreedom for all motion. 28-34) represents the mapping of the end-effector’s velocity to the jth joint. Three appropriate inverse mappings (Eqs. Near Solution of RPP. In this robot.RRR. Four Solutions for RPP-RRR-PRR Manipulator 5 Copyright © 1997 by ASME . Note that in some cases the relative angle θ 2i is fixed so it can be designed to avoid the singularity for all motion. j2=1. its overall inverse Jacobian M may be constructed from the rows corresponding to the actuated joints in each independent chain. T active and passive.PRR Robot 20 20 18 18 16 16 14 14 12 12 10 10 8 8 6 6 4 4 2 2 0 0 0 Invalid Actuation Schemes The matrix M maps Cartesian end-effector velocity into actuated joint velocities. The Cartesian rates X = {x y ω z } are commanded to each chain. all joint rates. certain parallel robots have invalid actuation schemes. the RRR chain has elbow-up and elbow-down solutions. If a manipulator is not always singular. the overall manipulator has an additional uncontrollable freedom. The actuated joint in each chain is underlined. the third. The jth row of the seven serial chain inverse Jacobian matrices (Eqs. the relationship is thus M = − A −1 B . However.RRR.( θ 2i = 0. it is still subject to the serial chain singularity conditions identified previously. al. In this case. one for each independent serial chain. This mapping is called the overall manipulator inverse Jacobian matrix. Due to this problem.) For example. and the PRR chain near and far. For a more complete treatment of planar parallel manipulator singularities. the forward velocity kinematics problem for the overall robot can be solved. and third joints.PRR Robot 20 20 18 18 16 16 14 14 12 12 10 10 8 8 6 6 4 4 2 5 2 0 0 0 5 10 15 20 0 5 10 15 20 Figure 5. the above procedure yields six passive results which are not required for resolved-rate control. j3=3). Table 3 shows the invalid actuation schemes. 2. derived in the position domain. or 3 for i=1. The RRP and PRR serial chains are singular when c2 = 0 ( θ 2i = ±π / 2 ). 28-34) must be chosen. where in other cases θ 2i is variable.RRR.al. are actuated for serial chains 1. 2. 5. 36:  L2    1  L12  θ   33   c1 / c2 1  1 =  s12 2 / s2 2  c13 / L2 3 s2 3 s11 / c21 − c12 2 / s2 2 s13 / L2 3 s2 3 ( 0 0 1   M = 0 0 1 0 0 1 Clearly. first. the actuated joints must be considered. and 3 (j1=3. The RPP chain has a unique solution. The 3x3 matrix is built from actuated row j for each serial chain i (j can be 1. a method is developed in this section to map X into active joint rates only. as shown in Fig. Far Solution of RPP. Any parallel manipulator constructed with permutations of the serial chains in Table 3 has an invalid actuation scheme (not just for identical chains). −1  PRP Table 3. for any robot configuration. RPP ) X = M Ρ PPR Merlet (1996) states that “each joint of a chain may be actuated [provided] the chain obtained when locking the actuated joint is not of the PP type. Therefore. (1995). The RPR serial chain is singular The overall inverse Jacobian matrix M may not be invertible based on a manipulator’s actuation scheme. The physical interpretation of this mathematical singularity is that when the three actuated joints are locked. This singularity condition is not configuration dependent. Elbow-Up. see Daniali et. is verified in the velocity domain as shown in Table 3.

which agrees with previously published results. DE25. No.J. 341-351.. Journal of Dynamic Systems. and Reinholtz C. and control of any 3-dof planar parallel robot. 533-551. and Pham D. 1985.. Vol.. one actuator per chain. Zsomber-Murray P. Kissimmee FL. The primary contributions of this paper are: 1) general algebraic approach.1-VI. 5th World Congress on TMM. 45-56. "The Dynamic Modeling and Command Signal Formulation for Parallel Multi-Parameter Robotic Devices". Vol. pp. 4. simulation.. 20th Biennial ASME Mechanisms Conference. Mechanism and Machine Theory. Shirkhodaie A.-P. 3744-3749. "Planar Robotic Mechanisms: Analysis and Configuration Comparison". 1988. 22-28. Vol 8. Measurement. and R and P joints. Williams II R. 1996... 1995. John Wiley & Sons. 30. Cox D. 1981. Vol. where one computer program solves all possible inverse position and velocity problems.. 1990.. DE Vol. pp. ASME Journal of Mechanisms. 118. and Soni A.H. Since the actuation scheme does not affect the inverse solution.D. 1988a.H. pp.L. 1983. Gosselin C. "Simulation Algorithm of Oklahoma Crawdad Robot". 9th Applied Mechanisms Conference. and Reinholtz C.. 4. and Soni A.H. pp. "Kinematic Analysis of a Planar EightBar Linkage: Application to a Platform-type Robot". 30. "On the Quadratic nature of the Singularity Curves of Planar Three-degree-of-freedom Parallel Manipulators". Aradyfio D.. Hunt K. 20 18 16 REFERENCES Y-direction 14 12 10 8 6 4 2 0 0 5 10 X-direction 15 20 Figure 6.-P. 71-78. Williams II R.R. pp.M. Gosselin C.L. The first author would like to acknowledge the reviewers for their helpful comments.. No. pp. The class has 3-dof. three serial chains of 3-dof each connecting to the end-effector triangle. 110. 1. "Closed-Form Workspace Determination and Optimization for Parallel Robotic Mechanisms". 1. and Gosselin C. International Journal of Robotics Research. New York. "The Optimum Kinematic Design of a Planar Three-Degree-of-Freedom Parallel Manipulator". Gosselin C. No. Merlet J.-P. 851-856. The results of this paper can be used for the design.Z. DE Vol. pp. Williams II R. Montreal. 15-3. 2) all possible independent 3dof planar serial chains with R and P joints are classified and their inverse position and velocity problems solved in a unified and succinct manner. 1996. Inc. 665-678. there are seven possible serial chains. 5. "Kinematic Simulation of Novel Robotic Mechanisms Having Closed Chains"..R. Journal of Mechanisms. and Automation in Design. "Singular Configurations of Parallel Manipulators and Grassman Geometry".M. Mechanism and Machine Theory.M. pp. 37-43. and Qiao D. VPI&SU.H. 1987..J. 1989. Vol 4. ASME Paper 85-DET-81.F. Kissimmee FL.T.. No. Ph. Vol 4. No. "Direct Kinematics of Planar Parallel Manipulators". and Merlet J. VA. "Forward Dynamic Analysis and Power Requirement Comparison of Parallel Robotic Mechanisms".M. 105. MN.M. and Reinholtz C. for a total of 343 distinct robots (less if the order of chains is unimportant). pp.D. "A New Architecture of Planar Three-Degree-of-Freedom Parallel Manipulator".. Merlet J.F.. Vol.. Vol. Sumpter B. Each serial chain is solved independently given the endeffector pose and rate. 15-3.ACKNOWLEDGMENTS An PRP-PRP-PRP robot is arbitrarily chosen to demonstrate the resolved rate simulation in Fig. "Structural Kinematics of In-Parallel-Actuated Robot Arms". Daniali H. pp. Kansas City. and Automation in Design. 35-41. Lemieux S. and Control. 20th Biennial ASME Mechanisms Conference. Sefriou J. "Singularity Analysis of Planar Parallel Manipulators". 1988. 5. ASME Mechanisms Conference. 1985. Pennock G. and Angeles J. and Angeles J. Summer Computer Simulation Conference. No. Mechanisms and Dynamics of Machinery. 3) overall inverse mapping for actuated joint rates only. VI. 1987. 6 Copyright © 1997 by ASME . Montreal.F. "Parallel Computation Algorithms for the Kinematics and Dynamics of Planar and Spatial Parallel Manipulators".L. pp. 1996. 1979.H. Dissertation.. IEEE International Conference on Robotics and Automation. DOE Report. "Forward and Inverse Synthesis for a Robot with Three Degrees of Freedom". 1995.. 1988b. and Tesar D. "The Analysis of a Six-Degree-ofFreedom Workstation for Mechanized Assembly". MacCallion H.. Resolved-Rate Simulation of PRP-PRP-PRP CONCLUSION This paper has presented the general inverse position and velocity kinematics solutions for a broad class of planar parallel robots. 6. pp. Transmissions. Transmissions. Blacksburg. and Kassner D. 3738-3743. Minneapolis. and 4) identification of invalid actuation schemes.J. IEEE International Conference on Robotics and Automation.. Mabie H..3.