100% found this document useful (1 vote)
449 views49 pagesChapter 7 RK Mittal Robotic
trajectory planning only
Uploaded by
tycat evlewttCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
100% found this document useful (1 vote)
449 views49 pagesChapter 7 RK Mittal Robotic
trajectory planning only
Uploaded by
tycat evlewttCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 49
Trajectory Planning
fhe end-effector of the manipulator is required to move in a particular
fashion to accomplish a specified task, The previous chapters discussed the
niatical models for geamettic, kinematic, static, ancl dynamic hehaviour of
nical manipulators, which form the hackground for this chapter. The
execution of the specific task requires the manipulator co fallow a preplanaed
path, whieh is the larger proble
control for the manipulator. The goal of trajectory planning is to describe une
requisite motion of the manipulator asa time sequence of jomUlink/end-el lector
locations and derivatives of locations, which are generated by “interpolating” or
“approximating” the desired path by a polynomial function. ‘These ume base
sequence locations, also called time law or time history, obtained from the
trajectory planning serve as reference input or “control set points” to the
manipulator’s control system. The contol system, in tum, assures that the
manipul annedl trajectories. The mation control is discussed in
the next chapter.
This chapler focuses On various Wajector
with describing the terminalag:
bf motion or trajectory planning and motion
ar executes th
planning issues. The chapter begins
«in trajectory planning. Some techniqu
involw
s
nefation
for trajectory 2 hwo lypes of motion, poi ( motion, and
continuous-path motien, are presented. First, the problem of trajectory planning
joint space is consiclered and then Cartesian space planning is discussed, The
chapter concludes with examples illustrating various trajectory-planning
wehniques.
‘The user usually defines the desired trajectory by a number of parameters. A
polynomial function is wentified to “approximate” the rajectory. The trajectory
planning consists of generating a time sequence of values of different parameters
attained by the polynomial interpolating the trajectory.
Every manipulator has at least capability to move from an initial location
(position and orientation) toa final desired location. The kinematics and dynamics
of the manipulator govern the transition of posture with actuators exerting the
o-prai desired wryues, As no Limits must be violated
should be exec
jell, iL is meeessary ter dew ise pl
oth trajectories
nbjectiveset any teajectory-pla isto achiene a
A sieoth function is ane thal is enatinuows
7.1 DEFINITIONS AND PLANNING TASKS
hich the
sa black box tw
cory pl
ay be thowghe ¢
ence of intermediate con’
e trajectory pli
expressed in cith
Fig, 7.1 shows the parameters involved in the trajectory-planni
solve the trajectory-planning problem tie terminolagy used in trajectory planning
bs discussed first
ner generates atime igurations
rant in
joint or Cartesian cverdinate frames, The block tia
problem, To
te Tire sequence of
i ie rinks wilh loca
Planner
ath is the locus af poi
esian space) tw be traversed hy the itor te execute the
specified task, A path is a purel of the
metia
Git Trajectory A trajectory is a path with specified qualities of motion,
that is, 9 pation which a time law is specified in termsnf velocities andsor
accelerations ai each point In other words, a trajectory is the time Mics and Ce
und accelera
sequence tor time hiss velocity. 1 for each
Joint or end-effector of the manipulator. A teajcetory is both a spatial and
temporal representation of motion. [teat he specified either ia joint space
my) of posi
or in Cartesi
space
(ii) Knot Points or Vi
e the set of intermedi:
1s oF via points ar interpolation
Points Knot ju
ions between the start and
bwhtieh ther
yslination.
a
(iv) Spli Ibis the smooth fumetion that pe
aint Space Trajectory Planning — In joint spy
each path point is specified in terms of a desired position and orientation
ofthe end-effecte relative We the base frame Unese points is
sired Joint positions by application of inverse
nd for cack of the jolts, which
through the set of via pos
ory planning
traje
then fo
Lan
ut Space Trajectory fin Cartes
ily specified tn the ¢
space uraje
n space. The
s (velocity, a ele.) are specified i
ervanead in jenn
jh as ee utes
ified trajectory
eneration [i is the act of ce
time sequence of values in real time, using
algorithm based on the spatial and temporal c«
(iii) Path Update Rate The rate at which the trajectory points are eo
at run time is called fe
(vil) Trajectory © y the trajectory asa
the trajectory planning
straints
<2) Steps in Trajectory Planning
Trajectory planning can be divided inte three steps for solving the teajectary-
lanning pe
biem as outlined below
G2 Tusk Description “The first step in te motiel # peoblem is to
ictentil'y the k
yput to the Wajecte
of mai
trajecte
In the ease 0 tle task is
PEP mation. In
< smooth, na pa
eff
cases, the user specifies only the goal point (
nt
lor is giv
known initial poi
If in addition t0
w be w
samt and finish points, a spe
d by the end-effecter
ul trajectory. Cope
fc path between them is required
his Is called consis
are welding and plot
ction "wo polyaomial functions a examined re. one im each of bove approche
() acubic polynomiat ia which euics connect te path pons ofeach joint
ina smooth way. and
(iy a tncar
iystrate the concepts of constraint Klenticaion and their use to compute
polynomial enetieiets,
721 Use of a p-Degtoe Polynomial as
Interpolation Function
The session of a single polyno forte ents Joint path depends om the
number of consis umposed and the typeof motion deste. The mein
Hower pene kis jie
that wile picking up an objet the motion ofthe end-effector must be directed
away from the supporting surface so thatthe grist dacs ot crash into the
supporting surface Tus if positon” slang the outwand normal othe
Surface, fom thei! poston, elected as aditonal specification, On the
basis of snl considerations 4 "setlown position” i selected to specify the
start an Snish pont snot enough fr stisftory
(yal posion,
i) Lito poston
in verially upward dvetion from a horizontal surface). Aer iss moved &
safe distance, volt fs complete. From then onwards, it moves at constant
Those conceps for de end-effector motion can be extended 19 positon
Yelociy, and acceleration of the joins. Further, additional eonsraats ate
foqated for contnulty of each joint variable and is desvatves a etry path
pola, In summary, the consents that can be appli to joint fr join
-
2, Velocity constraints, and
3. Acceleration constrains
(by) Intermediate constants are the constants on he intermedia pons of
thereto. ‘aff position
velocity at lit-olt poi
aL LiIL-Oft pasta
al seb-dowen posilie
vol velasily at sebalown
fof seit
slocitylaecelern i cif fitisvnal comstesint
of continuity
ter Fly
point ef the
Position
point
nia eo
2. Velocity, a
3
th the
serajectary incase af
lendled trajecte
A suitable polynomial function is selected such that the © cl ennstraints
degree polynomial
a
cneral choive
isfied and the trajectory is smenth, [tis known tha
has (p=)
traject
for trajectory plan
ots. Thus, fo
cnefficients and can therefore satisfy (p=1) canst
tra can be the
2. However, if the degece of polynontial is high (say > 5), it
0
with (p+1) constraints, ap-degree poly
Ji tends to cause extraneous motion.
hecomes computati
The comple:
Innenits $e
can be ceduced by splitting the ajeciery into two or more
a in ca
sepment polynomials of lower depree can be us
jectury is split with Iwo intermedi
For each of these three sep
ate the joi ii
wajectory, Fe
exampl
I be three sep
ents, Ewer
points, there wil
de, Iscan be used
ns,
ee polynon
The types pw
like
pial used for
ctory alse-fas optic
t veloc
y trajectory
2. Parabolic position or lingar velocity ar constant acceleration trajectory
Some other comples possibilities of polynomials are:
dy — The first «
para
polated with aquiaratic pe
initial position to the “blend” point, The second segment is imerpelated as
stant velocity followed by 2 quadratic polynomial an the last
vont is inter
ynontial (parabola) specifying trajeciery from
linear or en
(third) segment, ie I, there are Six
so hee wil be (pr V)k-26 |
trajectory consisting of (K-1)
yy
sii fdr interpolation of trajectory,
Gorivatives at the interpolation
‘as cubic-spline functions. The cubic-
the interpolating functions as the
y
jectory iuterpolated with cubic polynomials will be 4k-2. Cubic
jal and other trajectories are discussed in the following sections. For
example, a five-cubic polynomials interpolation will have five segments and six
‘interpolation points and the number of constraint equations will be 22.
7.2.2" Cubic Polynomial Trajectories
‘The problem of obtaining cubic polynomial trajsctories is considered first, for a
point-to-point motion without via points und then it is extended to the case with
via points.
Point-to-Point motion without via points In this scheme, the goal point
and travel time are specified by the user, The set of joint-variable values for the
elven goal point are obtained from the inverse kinematics. It is assumed that the
joint-variable values corresponding to the initial location of the
beh end-effector are
A cubic trajectory is obtained for each joint variable,
. nl » q(), of the n-DO
manipulator, The determination of the interpolation cubic polynomial is
out for each of the n-joint variables, , ° 1g 255
Lota alg be the stating and goat-point valu
watable 4. Note tat the sonotalized joint vane doves the joint vacale
(Bor a) tor jot of the DOF ts
respectively of the joka
dependent values of q,adiferent path points superscripts used, for example
jo) ave w te saath. This reqaement imposes swo constrains each on he
Jnint positon and veloeity fanetons Farther, 4) zor fr and 4, Ths
Along with he continuity requirements gives he four constrains
aor
ad os
a0y=0
H)=0
that, cube polyoma with four costiients. This ties our assrption
that cubic polsromial would give a roth motion profits fr jt variable
Therefore, to describe the Joint motion, assume tht te cub polynomial
‘which ves parabolic velocity profil
und a linear ace
G0 = 2ay + bat 6
sraats of £4.67) 1 Eqs (7 and (75), ghves the
The sluiow of ig 7-7) pivesthe cveticentsof cubic polynomial Eg (74), 28
358) Noro and
Thus the cubic polynomial oierpolate te path connecting the inl join
sti 0 nl joint poston is
wag +
Ie may be noted sha this solution ifr the case when the velit is zea ath
tthe start and Finish pins
The acceleration profile towel near, tas seontinulty at ial! inal
postions andthe motion wil sry atthe start and a the goal point duc ths,
The varatonof the oi variable, velocity and acceleration ae shown in Fig.
Poi
to-point motion with via points Now consider the problem of
determining the cubis bat connect the given "X" path poms smoothly. I the
manipulator come h via pon, den the cubic solution of above
section can be applied between each pa of succesive via points. A differen
“Mjpping eon oes athe ha pts, Fora path ih ph pins, he
pins a sot way . .
eco the
poston and onestaan, Sina 1 th sing: goal pont sae each ofthese path
Assuntos te ais ten by fiat warble 60
ding to path
1) is (0. This
These time atta fork pth po
oat jand te cubic polynomial connecting path points and
bie ts dsfined in the sie inetval <4 fy ih T,
ravel tne between path points jad ‘om with the continuity requirements gives the four constrains as
war
aia oo
any=0
aio
To satisfy these Four constraiats, the polynomial g(M) must be of enler tive
thats, x eune polyvomial with four coetfcients. This jstifies our assumption
that 2 cubic plyseial wold give a sme motion profile forint variable
Therefore 0 escrie he joint motion, ssn tha the eae polyoma
Wat ai bah sat oa
inch gives a parabolic veloeity profile
Gl) 4, Dato Bog? os)
anda linear acceleration profile
a
Applying the constraints of Eq. (7.3) 19 Eqs. (7.4) and 0.5), gives the
following setot four equationsin four unknowns:
days ba co)
=a)
pty tant badd at
420
a) #2ayt, Hay 0
“The solution of Eq. (7-7) giveste egemeents of cube polsornil. Eg. 7.88
=a!
o,=0
paw) 73)
Fig. 74. Cable polyno rjetory for PTP mi
Point-to-point motion with via points Now considst the problem of
‘otorminine the evbics that connect tho given "path points smears HF the
‘man pulator eons tn estat eaeh vis poi, ren the cubic satin oF above
section can ho applied hotcen eae par of sueeassve Via points. 4 ferent
pach iy agesd who the manipulates rogured 9 paste va point witho
sopping or coming ores at dh via poims. Fora pa wit & pal points thete
se (6-2) injermadiate oF via poins, and (5-1) polymomials ate requted 10
sonect te & path pons ina smooth 3
‘Usually, each path point is specified interns of the desired end-effector
position and orientation, Silla tothe single gol poiat cise, each of these pa
Points is tavstormes int & st of joint-variable Values by inverse kinematics,
‘Assume thi, demates the sls taken by join variable y corresponding to path
point j ma the cubic polynomial
etn path points Fal G+) 0), This
pete ty mth T= 1,~ 4, denoting the
leave tine hetwecen path pots j and (jo). These me interval fork pats point
swith (6-2) via pits ane shows Figs 7.5.The atl travel time for he pth
Hea) with Rared'=a
Raed =f 90
Panerai a
Parker epee
Pane
Equation (7-15) represents a set of CK-1) esuations ftom which the four
Uunkoven goelicknts for eas ofthe (\-{Vcubies ean be bland,
reached spo the tenyporal esti species well as ensues the eons
‘ia pins reguite diferent amon of compa
computations, whsteas the last ons is most computationally intensive
possibility 3 cusutes te velocty and acceleration continuity athe via pots
ising a smooth motion throughout, while ia possibilty 2, jerks ars possible a
‘ia polos because acceleration continuity isnot enforced, The Neuss used
‘would also talusnce ae trajectory, Posibiy | iv use when is desis 10
bf desieed velocities asia points, whieh way bea bande
7.2.3 Linear Function with Parabolic Blends
This would cause the velocity tobe discontinuous atthe beginning and end af
cry for PTP motion without via pits selon 7.22), te diseantinity of
physically realizable. One metho to avo hie dificult isto use higher degree
snot with mone constrains, ragiring above complex computations, 38
fiscusedinsction 72.1. Altemsaey, tis posibe to use a lend poly
‘which allows for vertieainn whether th resulting velocities art accelerations.
can be supported bythe physical mechanic manipalatr.
‘Atrapezoidal velocity profile. for example. imposes econstant aeceloratin ia
and a constaat deceleration in the al pha
at both ends. In general, 2 blended trajectory 1s constructed axing polynomial
lncerpolaton function with leading Functions at bowh ends. which ae splined
pon oi with ia point
Blended trajectory for point-to-point motion withow! via points For the
isch start sl taget pins, a siauth saetory, which consists of 3 Kear
orion of the trajectory, constant acceleration is used to achieve a smooth
‘elit wansitinn, The te pazapole bens ars assumes te iene has,
ney have the stm constant acceleration. Tetsfore, parabolic Blends nar the
path points ar of same duration (4) an te whole trajectory i pmamecic aout
fhe Nalfwsy point in time and position (iq, q"). Thus, the tajetory gives
‘of posiion fr this uaestory fsshowa in Fi. 1.6,
gf stave inact, with kaw inal pointy? and the eanstant Wend seceleaton i
Fig 76
‘eloiies ae 7 al the joint positon efile is spt! about the pit
‘locity at mien ft fis lens (ar Beginning of sod blend) must Be gul
fhe velocity ofthe incr samen, tts, at he lend pons Cs)
ean Trajectory Planning 259
Pa )=qi=q
Pty)=qgl=q' =0
Pot =q' Pd )=q' |
Po )= P(t)
. . > for j=2.4(k-1) (7.15)
P(t )= Py) |
Po )= Pc)
qg®
Pot, )= @ =G8 =0
Pe = a"
Equation (7.15) represents a set of 4(A—1) equations from which the four
unknown coefficients for each of the (A-1) cubics can be obtained.
It is important to note that Eg. (7.15) guarantees that the path points are
reached as per the temporal constraints specified as well as ensures the continuity
of joint velocity and acceleration functions at the via points.
The three possibilities discussed above to connect the cubics smoothly at the
via points require different amount of computations. The first requires the least
computations, whereas the last one is most computationally intensive. The
possibility 3 ensures the velocity and acceleration continuity at the via points in two-end segments.
gah
e
Parabolic
q”
Paraboie —Linear
a“
0 >
0 fy tn t-h & t
Fig, 7.6 Time history of position for trajectory consisting of # linear
with parabolic blends
As can be seen from Fig. 7.6, the constraints are: both the initial and final
velocities are zero and the joint position profile is symmetrical about the point
m a
(qty) with
s 8
mG +
g = and fy,
(7.16)
The trajectory has to satisfy the additional constraint for smooth transition
from parabolic blend to linear segment. For joint velocity to be continuous, the
velocity at the end of the first blend (or beginning of second blend) must be equal
to the velocity of the linear segment, that is, at the blend point (q’, ty)
mob
qr, = 2-4 (7.17)
ty — th t BBR) Robotics and Control =
interpolation function with blending functions at both ends, which are splined
together so that the entire path has continuity of position and velocity, including
| the smooth motion at resulting via points.
The trapezoidal velocity profile trajectory planning is discussed, first for point-
| to-point motion without via points and later, it is extended to the case of point-to-
point motion with via points,
Blended trajectory for point-to-point motion without via points For the
given start and target points, a smooth trajectory, which consists of a linear
segment with parabolic blends near the end-points, is used. During the blend
portion of the trajectory, constant acceleration is used to achieve a smooth
velocity transition. The two parabolic blends are assumed to be identical, that is,
they have the same constant acceleration. Therefore, parabolic blends near the
path points are of same duration (t,) and the whole trajectory is symmetric about
the halfway point in time and position (f,, 9”). Thus, the trajectory gives
continuity of position, velocity, and acceleration throughout. The timé variation
of position for this trajectory is'sshown in Fig. 7.6.
For this type of trajectory planning, the’user input is the target point position
qf, travel time, with known initial point * and the eonstant! blend acceleration 4 q°
intwo-end segments.
>|. Parabolic, 3 pede alten
is, fe Tg Hergebonge wan? 6
| Fig. 7.6 Timi of pitino etary oii of linemen
i © with parabolic blends
As can be seen from Fil
the eas rato ‘To predict the time history of g, 4, and g, with given q', 98, and ty,
unknown parameter is ¢,, the blend duration. Solving Eq. (7.19) for t,, giv
ey
Note that there are many solutions to Eq. (7.20), depending on the va
but the answer is always symmetric about the point (q”,f,). The constraint
choice of acceleration during the blend, g°, is ieee
# vk
A
mu
slate a I and g° #0
Solution to Eq. (7.19) will not exist if the above condition is not satisfi
When equality occurs in Eq. (7.21), the two blends meet at t, and there is
linear or constant velocity segment and only acceleration and decelerat
segments are present, giving a triangular velocity profile. a
On the other hand, as the acceleration becomes larger and larger, fy
shorter and shorter, The trajectory tends to linear interpolation, with acceler
tending to infinity orf, tending to zero.
The trapezoidal velocity profile (linear intexpolation with parabolic blends)
‘trajectory using’ Eq.(7.20), generates the following sequence of poly: s
the titne history of joint position, velocity. and acceleration.
la Polynomial q(@) does not touch the via points and the via points become
pseudo via points. This means that different methods to compute |
duration and trajectory for end-points and for via points have to bes
q
Fig. 7.7° Multisegment linear-parabolic-blend ‘splines
The path points, travel time between succes path points and «
acceleration at path points constitute the input for the problem.
parameter required to détermine thestime history of q,
lion at each path point, %,
that there
ni Trajectory Planning
For the via points, the blend duration fy near via point ! is computed from
specified a at the via point g and the constant linear velocities in two
segments g and 4”
(7.25)
ny) <1
sign(g - 4!")|@'|
‘The linear velocity in segment jlis given by
where
(7.26)
g parabolic blend to be
(7.27)
ssible solutions, depending on the value of
blerid duration at the initial
nd of blend is the same as the
is gives
Once again, there are many
accelerations used in each blend. To compute
1, the fac
Of the linear segi
Gt = (7.28)
where @' = sign(q? -4')\q
4, at the initial point and using #,, the linear
and f,
and fy
From Eq. (7.28) the blend tim
segment velocity atid duration, ¢
Finally, si blend duration.at the goal point is cecil Sint ta the.
starting point,:the velocity. continuity constraint, in the middle. searients
Gotmestng:pathepbinss (b+ 1) and piven
oh Cae
i ts — r fs + 13;
Tae — 3% a ‘Thus, all the requisite parameters to detecutte the: time his
position, velocity, and acceleration are obtained using Eqs (7.25) through
‘The acceleration at each via point must be sufficiently large, as as discussed
segment case (see Eq. (7.21)]. To make computations simpler,
blend acceleration may be used for each via point.
7.3 CARTESIAN SPACE TECHNIQUES
In the previous section, some approaches to joint-space act pli
been discussed, Another approach to trajectory planning is to
coordinates and Cartesian space. In Cartesian space, the positic
of a rigid body can be clearly defined.
‘The problem of planning trajectories that enable the manipulato:
to track a given path in Cartesian space is investigated in this section, The
specifies the desired end-effector path, the mavell nea time, and the t
orientations along the path
To tackle the aforesaid problem, a eee descripti
effector path is required. This description specifies |
requisite motion with respect to the base frame. ‘First
parametric description of the path in Cartesian space i briefly 1 revie
some common techniques for Cartesian space trajectory sane
| representation, the path (or curye) ceeds 2
single parameter, often this parameter is time: The: Fig. 7.8 Path ® relative to frame (0)
* ite ott th the Bath coordinate céan be used to express any point Foal
the path with respect to frame {0} as
"P = F(c)
Note that in Eq. (7.36), °P is a3 x | position vector ree
frame {0}<'The vector function F(c), thus, analytically expresses any point on tl
~ path Pwith respect to frame {0}. Hence, the path !Pcan’also be written as! Testy Planning |
point Pon path Pis the length of are of with exttomes P and, Usull, the
Sasting pot oa te pa P, she path eooudiaae origin and at ip path
‘0. This means tha or any point P onthe pa, the path coordinate species
the are length ofthe atts beeen P and P
2 csaiteg a
+ Se vr
Fig. 78 Pai Peli fame
Ie thoa follows hat the path oortinatsc ean he ws to express any point Pon
the path with respect frame JO} a5
Pek, 036)
Note that in Eg, (7.36),"P isa 8 1 postion vector point P with respect t9
feame {0}. The vector forston Fic). this, anasticllyexpressesany point onthe
path respect to frame [0). ence. the pat #?can also be writen a5
ae Fee) os
“Tree unit vectors enn be used to express point P which form a right-handed
lomhononpal coordinate system with three onhonormal planes 2s shown in
Fig. 73
fechin ane
Nom oe
7 pen
owmotne
ene
Fist unitvetor isthe tangent. nit vector dpolatingin the ineeasing dvection
lof The unit vot les nthe osclatng plane ©3t point P. This plane the
mit postin of the plane containiny the unit vector aad point
Pet, when Ponds to P along the pat, as shown in Fig. 7.8, The nora unit
oetord ats theunit voce perpnicula tad andcontined in pane ©. Tie
Aiceetem otf isehosen so thatthe path, a he netptborhood of P, wah spect
{w the plane eataining an noemal to, es cn the same sie af . Dinwrma
func veetor 6 completes height handed entnonormal cooedinae Lame {20}
[As per the definiians given above, these unit kets are anally related
lowe 38
wr =I we »
Use of this paramere seston of path ig (738) made wo compte te
‘eajetay ia Canes spe, This lst for i Gequenly enced
(Canesin space paths in maniptlation asks sight ine an erelar moon,
732 A StraightLine Path,
Consider a stsightFine path with tat a goal points", ane", Any point on
‘he path en he expressed in ers of pats parameter a8
Poe het
" 1139)
Peat )
Nowe hat PO) =P, and PP, "2, ) =, represents tbe overall length of
Iie senens Fortis pth the wn secs ae stem a
a a
Fence the unit vectors 14 and ate not uniquely defined ad are chosen
say ° OO oan
Town
Fig. 78 Poth lace to fae 108
Ic then followsthar the pat coordinate ean he used :0 express any polatP on
the path with respect wo fame {0} as
Note that ang. (7-36), "#183 31 postion vector of point # with respect to
frame [0}."The vector runeton Fle. thas, aalsicalyexpressesany point a the
path? with espeet wo fame {0}. Hence, the pa ean aso be writen 36
"pare 039)
Tes unit ectors can be used 1o expres point P, which form a ight handed
‘orthonormal coordinate system with thee oxthonortal planes 28 showin tn
Fig. 79.
24 nama _
Nonret ane
v(Nema)
(odaina ene
Acne
Fig. 729. The crthoarial srt aston fv pl ®
i as)
Use ofthis parametric description o path, Eg. (7.38), is male tpcampae the
luajectory in Cartestn spe, This sllstealed for Wwe frequently encountered
{Caresian space pas in manipulation iasks! seat Ine and eeu MOOR a
7.3.2 A Straight-Line Path
{Consider a seaight-line path wih start and goal poiaisP and, Any point om
the path ea he express in terms of path parameter
er os)
Nove tha (0) P and Fl, —°P,| = P, represents he oneal eng
tine sgment. orthis pth the wit vets ae determine 38
phy ,
a CBL n cao
de (inn
Peo cay
Henee, the unit veclors and d ate not uniquely defined amd are chosen
anbiteanly 733. A Clecular Path
“The ireulae path 5{¢), shown in Fig. 7.10, 8 specified using the Following
(2) Radas of cecal pat is pan plane vy
(6) Unigvectorerthestsis avs. atthe eons of sive, dicted assonting
torgitchand rue
fe} A point Bon the axis of eile pasing through the eenter oF eee. Is
positon veetor in Tram (0) 38°.
(4) Position veccor'P a eame {0} of potato the ete (his serves as the
path conrdinae origin.
Allthe above parametersare expressed i terms base Frame (0)
Fig. 740. Parc kescrption of clr path
1 3= °F, —"D. as shown i Fig 7.20, the position vector °C ofthe eiele
centse Dis obtained as
‘Cop's ne oan
First obtain the parametric representation of the cite with respect to
framets’ y* 2) defined as foliose” i she unit vector pining rom elle
centr 0 whic alo serves as the origin © towards ,
sirle ans while completes the right-handsdorthwnoml eardnaes fats
Any point Pon the cite Je) can easily he desceibod in tems of
feme{a? yo} as
pos (ofp)
painicip) ay)
oo
isan Conta
path: wih espe oFarme (xyes gven by
‘pcos ci
wer=| psingcip) ity
0
Finally, tae parametrie description of cise ith sapost ao frante (1) is
‘obtained by mapping. to hase Frame as
Mme 804 RR ers)
where isthe tation matt of frame (x yc] wih respect base Fame {9}.
“Thus the position vector of any pointe te Cele given by
"po ="e= RBC) 86)
Fro Fi (7.383 unit wectars are oad as
sn tcp
6 1M al costerps aan
« 8
~eos(clo)
eI) Ssa(cip) mas
ver)
ae
"he unit veetnr dis obtained rom aad
734 Position Plarsming,
(nee the piven path i ese in erm of pat pancters eesti ft
{termine te time history afendeecter position, velocity, afl teeleraion con
sistent with dhe tempor constants preset,
For a path Pto be traversed by agen DOF manipulator by ashe. fist
the exreme values of path coordinate ¢ ase detemmined. For the piven pa
ssuume c= 0 atuine?= 0 te ial positon. and e= 6, at P= isthe goal
Fosition, Nextafunction cf, which intrpolstenchetweenie =O aabe=e, ina
smoot yay, obtained. is smeeut Kurtin cf) sould sir be a Sabie
polynomial or alinesr sopment with parabole blends and isebtaned as seas
‘dic esseof jin spaee sehen fo poine-o pone madi without Va poiMs
‘he une history of Pty sutscutin evans aniined by (0), wesc inant
simeyin the Eq. 7-38.
“Thelfnear Yelocty ofthe en-efectoris determined 3s
ary
de
* 749
£2 sper bg (738) ATU ce ARS AE PENG ENN KE HE YO
o
& .
Fig. 740 Porometrie description of ciculr path
183 "P) —%D, as shown ts Fig. 7. the postion vecae "C ofthe eivee
contte 0 is obtained as
Des! Ae (ay
First, obtain the poramrie representation of the itele with respect 10
Frame(s" 7 e"} defined a folows$" the nie vector pools fron eee
center (which also serves the origin 7) owas
‘the unit sector along
circle ans while completes the ight-harded orthonormal coordinates fame
‘Any point P on dhe ctcke fe) ean easily be described in terms of
framete’ yf "bas
[pcos erp]
Pr! psin lp) 743)
Loo
Me ="C+R TO, 7.46)
From £g.(738) the unit vetors are obtained as
sin (eso)
= LP a] costespy oan
o
costetp)
Ri
= EO since eas)
Plo
a
The unit veclor is obiaine fom d and fi
734. Position Planning
‘Once the piven path ts deserbedt i tems of path parecer, the next am sto
determine the tine bisory of end-effector positon, velocity, and acceleration con
‘stent with the temporal constraints prescribed.
or a path to be traversed hy agiven DOF manipulator in a ime fis
the exiteme values of path coordinate « ate determined. For the given pa
assune =) a time 1= 0 ise iia position, anc = c, al = J, he goal
position. Next futon). which aterpolatese deter e =O ande=e, Ina
smooth Was, Is obtained. This smooth fusetion ct) could either be a euble
Polynouial ort linear sepient with parabolic bls a! obtained as discus
in the ease of jont-space schemes fr point-to-point motion without via pots
[Now that time Jas imposec on te path coordinate. ts possible to bain
thotime history of "Phy substituting the Values atained by ct) at cach instant
time, inthe Eg (7.3),
The linear velocity of the end-effector fs eterined as
AUP)
ae 7.49)
cs
-
wvtere g = 2OP) as per Bg, 7.38) Linear acceleration ofthe tol is obalne by differentiating Eq (2.49) with
respect otime
eer) Py
P (730)
Thus, for 4 Canesian sts
line movin, "Ps obvsine fom gs (F0)
P PR =a si
a
Silay, for iculor motion, from Las (7.87), (78), (79, ana (750th
Xelosty an! acceleration of point P ante cle ates
sin 9)
‘P= Re] cosicip) oss
0
08 (cp) En (cl p)
sia (ei p)+Eeos(elp)
Orientation Planning
The next stop in Cartesian space schemes is t9 obtain the time sequenes of
sriestations tained by the end-effector. Bus, these orientations cannot be
specified ny rotation mainews ecause interpolation of to-otation mazes does
om with respect the Base, the angles 6,
a-l0 6 ol ss
Because aris similario"P discus ia poston planning. any path on the en
effecorsofeattin along with zempora constaiascan he anyposed. The eps
‘uulied.as follows generate asmivoth ajectory joining
G10, 0 el t04, “10, 0 el" ntiner
path joining
Stcp2, Impose a tine lawn pth cooninatec, asdoncn position planing
in Cartesian space
Step, Obisin he tm soquence of vals attained by at based ono.
Step 4, As explained carer for poston
a) (aay wn50)
74 JOINT-SPACE VERSUS CARTESIAN SPACE
TRAJECTORY PLANNING.
The joint space trajectory planning is simple to implement and joint coontinates
fully specify the postion and orientation of end-effector. Dut when it eamesto
spoeiying a task or defiaing an vodinates
offer beer choice. Its etsy to speci the postion and arientaion of a Hale
cle in workspace, Caresian
‘ther denwck is that dia he anlinearnatueof the manips inematies
rol, ef difficult co pret the resulting emd-offecior motion that will be
Drocluced By a pariculartrsjectary excced in the joint space. The above
Inentioned dfawhacks could he overcome by employing Cartesian space
rajectry planning
‘One of the main aaybacks with Carsian space wxhnigues Is its
computational complexity. Be
The sther iiation wth C
sam spice schemes that he planned paths
7 singularities In sich 3
anes.
yeular configuration of manipulator eA, one WF Me oe
‘elociies may incre
initsls. One approach a avoid excessive jot ats ss ory Planning
y e
He limits, while computing the nl
2 30 would violate the
jectory
ion. How doi
ke the overall motion
this will
y alse fi ts of a planned path are
2 within the workspace. Th
jel by the
a
Cartesian space planning requires: more
computational ability, so that in rantime th
desired Ca paths, To sum it up. Cartesian space p better
and simpler reps ofthe task but ata higher co
ints are servecd to follow the
SOLVED EXAMPLES
Example 7.1 Cubie spline trajectory
of a SCARA wired 10 move from
The second fe
= 0" to 150" erate the
smith trajectory ti dl acoeleration
fier this trajectory?
Sotetion The
four coetfici
using Eq. (7.8) as
are computed from the abi
17.80)
ae)
a)
Xr)
velocit
of positic
fied sampling Example 7.1 Cubic spline trajectory
The second joint of a SCARA manipulator is required to move from
@, =30° to 150° in 5 seconds. Find the cubie polynomial to generate the
smooth trajectory for the joint. What is the maximum velocity and acceleration
for this trajectory’?
Solution The given parameters of the trajectory are:
O =30° 6 = 105°
v=0 ve =Q (7.58)
e=5s
The four coefficients of the cubic polynomial are computed from the above
using Eq. (7.8) as:
ay = 8" = 300
a,=0
(7.59)
Hence the polynomials interpolating the position, velocity and acceleration of
the joint are
O(1) = 30.046.07 -1.28
Ot) =12.0r-3.6° (7.60)
Ot) =12.0-7.21
As expected, the velocity profile is parabolic and acceleration profile is linear.
The time sequence of position, velocity, and acceleration can be easily generated
at any specified sampling rate. The plot of these functions will be identical to points. The desired end-effector motion |
Assume a cubic spline trajectory in each segment and the «
ia points is enforced. area
Table 7.1 Desired end-effector motion
Position q (rad)
Velocity q’(rad/s)
Traversal time 1, (5)
Solution ‘The given trajectory hs two via points and, F
segments and three cubic polynomials will be required to define
of motion of end-effector. Let the three cubic polynomials be
RW) =a taut taal? tant
P(t) = yg + anit Fang? tant?
P(t) = ayy + at tay! tas)
where ay are the unknown coefficients ofthe three cubic polyno
“The constraints imposed on these three polynomials are gi
‘Applying the constraints and substituting the boundary values
{frst segment cubic gives the following equations: * ‘= au + bon #1920p = :
7.65) are solved to compute the from coefficients
nis :
BP cabir fe: the second Segment is
(1) 91.77 +7.361- 2.2277 + 0.1607 _
es tetard cubic polynomial constraints from Table 7.1 give
93(8) = yy + 845, + 64a, + S12a,5 = Hz,
93(10) = ayy + 1045, +100a5p + 1000455 = =2n
43(8) = as, +1605) +192035 =
© 43010) = a3, +204; +300435
The fourconsans for the thifd cubic are oe as: Soluion As in Example 7.2, the starting and final velocities are rere, thal is
qi=q'=0 (uy
The average velocities for the segments |, 2 and 3 are computed using
Eq, (7.14) andl the data from Table 7.1, Thus
(hI
According to Eq. (7.14), the via point velocities are computed from
a if sipntig’”') + signi’)
ie
= 93
‘ (ya eg) if signtg’) = signtg) )
AL via point 2
oP <0 because from Eq, (7.72) signig') #sigmig?) (7.74)
and at via point 3
q° =O since sign(g*) « signtg*) (15)
Note that in this case the given path points pasitions result in zero velocity at both,
via points. The three cubic polynomials for the three segments of the trajectory
are computed using the position and time constraints fram Table 7.1 and velocity:
constraints from Eqs (7.71), (7.4) and (7.75), The three poymomials are as im
Eq. (7.61).
Applying the given and computed constraints for the first polynomial. gives
four equations as:
FAO) = ay =O
PUBI= aay +34, +97 + 2Tayy = 32
Asa, =0 (1.76)
A= 4) +6ay + 27a, =O
which give the four coefficients af P(r) as:
tp =a, 20
2
ay = 0.35
Applying the constraints for the second joint yields the four equations as
isoe Eq. (7.651):
PAD @ ay, + 3a) +9ayy + 2ayy mw Bed D
PABY © ay, + Bay) +fbayy +5 12a, = —/2
PW = ay) +603, +27, o
(hTTy
(778)
A(B)= ay, + 16ety +1922, =0 Trajectory Planning 275|
Solution Asin Example 7.2, the starting and final velocities are zero, that is
qgi=qi=0 (7.71)
The average velocities for the segments 1, 2 and 3 are computed using
Eg. (7.14) and the data from Table 7.1. Thus
3x
—-0
2
—2_2 5 (7.72)
> 242
gia q a a one 2 _ 5a
oT 3 2
According to Eq. (7.14), the via point velocities are computed from
qi 0 if sign(g!') # sign(q’) (173)
SG +g!) if sign(g’™) = signq’) ~
At via point 2
gg =0 because from Eq. (7.72) sign(g') # sign(q?) (7.74)
and at via point 3
g° =0. since sign(q?) # sign(g*) (7.75)
Note that in this case the given path points positions result in zero velocity at both
via points. The three cubic polynomials for the three segments of the trajectory
are computed using the position and time constraints from Table 7.1 and velocity gq =~ because trom bq. (/./2) sign(q*)# sign(q-) (#74)
and at via point 3
gq’ =0 since sign(q”) # sign(q?) (7.73)
Note that in this case the given path points positions result in zero velocity at both
via points. The three cubic polynomials for the three segments of the trajectory
are computed using the position and time constraints from Table 7.1 and velocity
constraints from Eqs (7.71), (7.74) and (7.75). The three polynomials are as in
Eq. (7.61).
Applying the given and computed constraints for the first polynomial, gives
four equations as:
P(0) = ayy = 0
PB) = ap +34), + Pay + 27a); = 3/2
A(O)= a, =0 (7.76)
P(3) = a), + 6a); +27a,; = 0
which give the four coefficients of P(t) as:
ayy = a4, =0
ayy = 1.57 (1.77)
ayy = 0.35
Applying the constraints for the second joint yields the four equations as
(see Eq. (7.65)):
Pi(3) = ayy +3cyy +9093 + 27453 = 30/2
PA8) = ayy + 8a, + O43, + 512a,, = -#/2
P,3) = ay, + Oay + 2Tay = 0
P,(8) = dy, + 16a, +192a,, =0
(7.78) 276) Robotics and Control
which are solved t give four coefficients of the second segment cubic Pfr) as;
(7.79)
Similarly, the solution for the four coefficients of the third polynomial Pts) is
7.80)
Substituting the computed values of the coefficients from Eqs (7.77), (7.79)
and (7.80) into three cubie polynomials (Eq, (7.61 )) gives the thre cubies farthe
three segments as:
Ris Ls7e -0.35"
P(t) = 3.79 + 7.241 = Loon? +0100 aly
Pur) LAR 10? = 4712 + $3.00? — 1.960"
‘These cubic for the three segments are plotted and the time-history of position,
velocity and acceleration are obtained as illustrated in Pig. 7.12, From these plats
itiseasy wobserve that the continuity of position and velocity is maintained us im
Example 7.2 and the acceleration is still discontinuous. Comparing these profiles
with profiles in Fig. 7.11 reveals the effect of change in boundary conditions om
the motion of the end-effector,
Example 7.4 Trajectory with a trapezoidal velocity profile
‘The motion of a jolntof 3-DOP arm is constrained by an actuator that can produce
2 maximum acceleration of 0.35 radés* and maximum velocity of 15 rad/s. If
Irapecoidlal velocity profile is assumed, determine the trajectory if the joint moves
by wrrad in 10s.
Solution A trapezoidal velocity profile implies a constant aceeleration in the
Mart phase, a cruise velocity and a constant deceleration in the arrival phase. The
trajectory will be as shown in Fig, 7,
Assuming that the trajectory is symumetic and the traversal is with maximum
Possible acceleration, the parameters for the trajectory arc:
gist
i,
vee 82
5 1s
5 rad/s G = O35 radis?
the inid-point of the trapezoidal trajectory
fete
: 1.83)
*
tye tede 276| Robotics and Control
which are solved to give four coefficients of the second segment cubie P4(f) as:
dy = 4.79
7.24
-1.66 (7.79)
az3 = 0.10
Similarly, the solution for the four coefficients of the third polynomial P3() is
ayy = 138x107
471.2 (7.80)
Substituting the computed values of the coefficients from Eqs (7.77), (7.79)
and (7.80) into three cubic polynomials (Eq. (7.61)) gives the three cubics for the
three segments as:
P(t) = 1.57? = 0.35
Pi(1) = 4.79 + 7.241 — 1.667 +0.10° (781)
P,(i) = 1.38 x 10° —471.2¢+53.0107 — 1.9677
These cubic for the three segments are plotted and the time-history of position,
velocity and acceleration are obtained as illustrated in Fig. 7.12. From these plots
itis easy to Observe that the continuity of position and velocity is maintained as in
Example 7.2 and the acceleration is still discontinuous, Comparing these profiles
with profiles in Fig. 7.11 reveals the effect of change in boundary conditions on
the motion of the end-effector.
Example 7.4 Trajectory with a trapezoidal velocity profile
an nradice Example 7.4 Trajectory with a trapezoidal velocity profile
The motion of a joint of 3-DOF arm is constrained by an actuator that can produce
a maximum acceleration of 0.35 rad/s* and maximum velocity of 15 rad/s. If
trapezoidal velocity profile is assumed, determine the trajectory if the joint moves
by rad in 10s.
Solution A trapezoidal velocity profile implies a constant acceleration in the
start phase, a cruise velocity and a constant deceleration in the arrival phase. The
trajectory will be as shown in Fig. 7.3.
Assuming that the trajectory is symmetric and the traversal is with maximum
possible acceleration, the parameters for the trajectory are:
q =0 qe=n
q =0 q® =0
1,=0 t,=10s
vo=I5 rad/s g° = 035 rad/s?
From Eq. (7.16), at the mid-point of the trapezoidal trajectory
gtg _n
g ” - = ;
b, = =5s5
(7.82)
(7.83) Substituting the values from Eq. (7.71)
1, = 101 | (0? x035—4@=0) _;
ge 35
‘The motion time law for the trapezoidal trajectory is, therefore,
substituting the computed and given parameters in Eqs (7.22), (7.23) at
0.1750? Osrsl
g(t) = { -0.175+035¢ 1<1s9
314-0.175(10-1)? 9<1s10
0.35t Osrsl
wo-| 035 1<1s9
0.35(10—r) 9<1£10
035 Ostsl
won| 0 1. = 13.0 deg /s
13-05% 0156 ver’
a 30H (7.88)
ro H- #0 = 500 dex /s
The second blend duration is computed from:
2B 12
where 7? = —55 (for deceleration). Thus,
tp = S089 L033.
° 55
and (7.89)
Also (7.90)
and gt = Ev __15-30____so7g aegis (79)
3p 0.092
MOS
— -5.078+5.0 _ yogiars
55
2-1 .33)—1 0.00142) = 1.8345
2 2 (7.92)
- ‘ (0.00142) = 2.99 s
The parameters for the trajectory are tabulated in Tables 7.2 and 7.3 with ne oe
2 2
Similarly, for the third segment:
4 523
- = 5.
=f 24 = 2OTR*5.0 _ o.00142
l
b3 = 2-5 (0.33)~ 5 (0.00142 = 1,834
%8 cL 3! ) . (7.92)
fy = -1 7 (0.00142) = 2.99 s
The parameters for the trajectory are tabulated in Tables 7.2 and 7.3 with
blend duration of cach parabolic segment at cach path point and linear velocity
and duration of linear velocity for each segment, respectively.
Table 7.2 Blend duration for parabolic blends at each path point
Path Point ins)
1 0.156
2 0.330
3 0.001
4 0.092
Table 7.3 Linear velocily and its duration for each segment
Segment q t
(deg/s) fs)
1-2 13.00 2.68
2-3 3.00 1.83
34 -5.08 2.99 280] Robotics and Control
Example 7.6 Trajectory planning for a 3-DOF RAP arm
Forthe 3
DOF manipulator arm discussed i
ample 44,
ig 47 design linear
trajectory with parabolic blemds. The initial & final position of the end-effector is
expressed by homogencows transformation matrices T, & T, given below and
time tiken for traversal is § seconds,
0354 866 0354 0,106
A612 O500 G6I2 0.184
q, (7:93)
0.707 oO O707 = G.212
o a 0 l
O53 0.766 0.272 0.027
0.694 0643 0324 GN32
= (7.94)
0.423 00.906 0.081
0 0 o ol
Solusion The forward kinematic moxdel for the 3-D0
in Example 4.4 and is given hy Eq. (4.67) as:
RP are was obtained
“SOS: AOS) Pin fia fis Hie
CSS, dy fa fy Ph m9
0 Gat, ‘a faye
oO oO 1 o oO O 1
‘The inverse kinematics of the arm wus also carried out in Example 4.4 and
solutions for the three joint variables were obtained as (sce Eqs (4.71), (4.73)
andl (4.74):
8) = Atami, — 44)
Atan2[2y ya + Foy oe)
Vata
For the Position of the end-effector specified by T,, Eq. (7.93), the
feasible solution obtained in Example 44 is(sce Bg. (4.75)):
% = 0°
= 45"
dj = 0.30 m
‘The joint variables corresponding to final position of the end-effector as
specified hy TF, are:
(7:96)
(797)
(7.98)
Now, to plan linear tajectory with parabolic blends for the three links, the
default acceleration for euch link should satisfy the constraint in Bg. (7.21). 280| Robotics and Control
Example 7.6 Trajectory planning for a 3-DOF RRP arm
Por the 3-DOF manipulator arm discussed in Example 4.4, Fig. 4.7 design a linear
trajectory with parabolic blends. The initial & final position of the end-effector is
expressed by homogencous transformation matrices T, & T, given below and
time taken for traversal is 5 seconds.
0.354 0.866 0.354 -0.106
0.612 0.500 0.612 0.184
T= (7.93)
0.707 0 0.707 O.212
0 0 0 1
0.583 -0.766 0.272 0.027
0.694 0.643 0.324 0.032
—0.423 0 0.906 0.091
0 0 0 |
Solution The forward kinematic model for the 3-DOF RRP arm was obtained
in Example 4.4 and is given by Eq. (4.67) as:
CC, -S, CS, —d3C)S; moo Mis Mia
SC, C, -SS, —d5)S; hy hm Ba fh
°7, = " 2 1 192 sez] _ |r 2 3 Ma (7.95)
y 0 C, aC, Bi 2 3a Ma
0 0 0 1 0 Oo 0 1
The inverse kinematics of the arm was also carried out in Example 4.4 and
solutions for the three joint variables were obtained as (see Eqs (4.71), (4.73)
and (4.74):
(7.94) eye 8p e824 3h 142 "nz 3 4
oT = SG CSS, -d35)Sp =| "2 3 Me (7.95)
Sy 0 C asc, Bi 2 a Ma
0 0 0 | o 0 oO 1
The inverse kinematics of the arm was also carried out in Example 4.4 and
solutions for the three joint variables were obtained as (see Eqs (4.71), (4.73)
and (4,74):
8, = Atan2(-rhy, — 44)
By = Atan2(ty ri try 5751)
dy=tViythatra
For the initial position of the end-effector specified by 7,, Eq. (7.93), the
feasible solution obtained in Example 4.4 is (see Eq. (4.75)):
(7.96)
& = -60°
a= 45° (7.97)
d5 = 0.30 m
The joint variables corresponding to final position of the end-effector as
Specified by T, are:
a = 50°
8 = -25° (7.98)
dj =0.10m
Now, to plan linear trajectory with parabolic blends for the three links, the
default acceleration for each link should satisfy the constraint in Eq. (7.21). Trajectory Planning 261]
lle seal (7.99)
‘That is, for Vink .
a 2 SECON 176 oe choose 4! = 200 dep! (7.100)
Similarly, for link 2 and link 3
Ge 2-112 archoose jy = 15.0 degis?
Gy 201032 or choose g,° = 0.05 mis? (inl)
‘The trajectory (1ime-history) for each link is computed using the Eqs (7.20),
(7.22), (7.23) and (7.24),
Bor first tink:
1 1
digs t 7.102)
25 20 prey
‘Therefore,
60 + 10s" Osrs 163
q(t)=)-B6574 326 Lakers 247
A3<1s 50
ay Osis Led
qi =)326 L6d<1s 337 (7.103)
MKS) AITers S80
0 Osts 163
qinsyo Led 0.032 or choose g;° = —0.05 m/s?
The trajectory (time-history) for each link is computed using the Eqs (7.20),
(7.22), (7.23) and (7.24).
For first link
Therefore,
-60+ 1017 O<1< 163
gq, (t)=5-86.57+ 32.6 L63
