Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp.

11-20

Sensors & Transducers

© 2014 by IFSA
http://www.sensorsportal. com

Picking Robot Arm Trajectory Planning Method

1, 2
Zhang Zhiyong, 1 He Dongjian, 2* Tang Jing Lei, 2 Meng Lingshuai
1
College of Mechanical & Electronic Engineering,
Northwest A&F University, Yangling, Shaanxi, 712100, China
*2
College of information Engineering,
Northwest A&F University, Yangling, Shaanxi, 712100, China
1
E-mail: zzy@nwsuaf.edu.cn, jingleitang@nwsuaf.edu.cn

Received: 18 September 2013 /Accepted: 22 November 2013 /Published: 31 January 2014

Abstract: The picking robot arm is scheduled to complete picking tasks in the working space, to overcome the
shaking vibration to improve the picking stability, its movement should follow specific consistence trajectory
points. Usually we should give definite multiple feature picking points, map their inverse kinematics to the joint
space, establish motion equation for the corresponding point in the joint space, then follow these equations
motion for the interpolation on the joint so that we can meet the movement requirements. Trajectory planning is
decisive significance for accuracy and stability of controlling robot arm. The key issue that picking arm
complete picking task will be come true by trajectory planning, namely, robot arm track the desired trajectory.
which based on kinematics and statics picking analysis in a joint space according to the requirements of picking
tasks, and obtain the position and orientation for picking robot arm, study and calculate the theory of trajectory
parameters timely. Copyright © 2014 IFSA Publishing, S. L.

Keywords: Picking robot arm, Trajectory planning, Posture, Interpolation.

1. Introduction joint space quadratic polynomial interpolation

procedures to ensure two adjacent tracks smooth
Trajectory planning can be in the joint space and transition, the realization of Cartesian space using the
in Cartesian space. Carried out in the joint space. It’s straight line segments constitute picking trajectory
joint variables is expressed as time domain function optimization method; Wicaksono, Handy et al.
to describe the trajectory. Study on the joint propose quintic polynomial interpolation function
trajectory interpolation and the moment constraints trajectory to avoid Jacobian by solving the inverse
and dynamic force range have been Conducted [1]. matrix from a Cartesian space to the joint space
These joints are concentrated in only continuity and trajectory planning results involving the conversion
smoothness of tracking interpolation trajectory with of matrix inversion and other complex computing
little consideration of the drive torque and the problems, and ensure joint angle and angular velocity
dynamics on the robot arm constraints [2]. The little continuity and stability [3]. As for the shortest joint
proportion of time spent picking trajectory planning space trajectory optimization problems, Monta M. et
method to determine the feasibility. The movements al. apply a uniform nonperiodic fourth-dimension, B-
of picking robot arm reach the target position with Spline curves based on chaos optimization theory to
the homogeneous transformation matrix, with each ensure that the approximation error is near to zero [4],
joint displacement, velocity, acceleration planning running time on the joint space trajectory was
will fit to Cartesian position and orientation, through optimized. The aboving various planning methods

Article number P_1701 11

 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

were used joint interpolation function, so the position spots in picking space need to remember trajectory
of each control instantaneous of picking the arm were by the demonstration, but that is unfeasible in fact
determined by joint interpolation polynomial [5]. regarding complex the picking environment [6],
moreover the more the spots demonstrated and higher
precision, the lower efficiency is instead. We can
2. Picking Arm Trajectory Planning demonstrates several characteristic points with regard
to the orderly trajectory path whiling the picking,
2.1. Description of Trajectory Planning attain the point the coordinates using the interpolation
algorithm, 2 spots must be demonstrated regarding
The trajectory planning is to control picking arm
the straight path, 3 spots must demonstrated for the
move along certain target trajectory, to guarantee
circular arc path, then transforms these point
track target trajectory with high accuracy control has
coordinates through the reversion kinematics
been the primary content of trajectory control of
algorithm to the arm various joints position and the
picking arm, general method of assigning the target
angle (θ1, θ2, θ3). The following angular position
trajectory the is the demonstration reappearance,
closed-loop control system shown as Fig. 1 enforces
namely to tell picking arm how to do by timing
expected path planning. The whole picking plan may
computation, then to remember this running process,
achieved completely by repetition above continuous
the following control may be the needed to duplicate
interpolation.
this movement at the right moment while picking, all

Fig. 1. Picking-arm trajectory planning control process.

Picking the path planning was carried out in In the path planning, picking point condition is
Cartesian coordinate space, picking arm’s shape indicated with its spot position posture in coordinate
position information, such as the posture of initial system, such as the picking outset position posture,
position, the raising spot, the releasing and the end the termination position posture uses the initial
point position, the speed and the acceleration of 3 station, the end point condition expression
joints must be gain, and guarantee picking arm run on respectively. When anywhere position the picking
fast, accurate and the steady condition, then carries movement is described, not only needs to plot out
out the joint interpolation, then ensure the joint picking initial point and end point, but also should
position, the speed, the acceleration on the entire provide certain node between the two points, which
time-gap continuity and so on. The interpolation also called the path node [7]. The trajectory not only
method is to select different type joint interpolating needs to study the position posture restraint,
function and produces the different path. The task of moreover study time assignment problem on various
path planning are contained the inverse kinematics path node. Namely, movement time between two
solution, the interpolation calculation transformation path node should be known on condition of
equation. stipulation way. In addition to this, the movement of
picking and harvesting should be smooth stable, non-
smooth stable movement not only aggravates joint
2.2. Description of Picking Objects spot attrition, moreover causes various arm vibration
and impact, this kind of vibration also increased
The description of picking object can be possible obviously fluttering [8]. Therefore the description
to draw a series of position posture Pi (i=1, 2, …, n) equation of picking path of motion should be
in the picking coordinate system. This description not continuous, even its first derivative and the second
only conforms to real man-picking the apple, but also time derivative also should be continuous.
advantage to describe and produce the way of picking To sum up, the position posture in relative
path. As for the description of picking movement, it reference system of random picking apple can be
advantage to separate picking the way, concrete described the coordinate system fixed firmly.
picking arm and the picking manipulator, which The picking apple, relative to any point in picking
forms the description method modulation [7]. In this the coordinate system, may be indicated with the
way, it is possible to transform the object from the corresponding position vector P, any direction may
original state movement to the final state, regarding be expressed with the direction cosine. After
the specific transformation of coordinate system from achieving the geometry shape of picking apple and
initial point {T0} to terminating position {Tf}. the picking coordinate system, if the picking

12
 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

coordinate system the position posture is stipulated, capture apple work, requesting to take out the apple
then the posture in the spatial position may be from the tree and put into bucket of robot, the
restructured. The picking robot arm work process is symbolic representation various points position
available to provide from picking manipulator posture along the path movement, make picking arm
position posture point sequence, position of each move and complete the picking task along the dashed
node in arm coordinate system is available to line. Establishes Pi (i=0, l, 2, 3, 4, 5) to picking
describe the homogeneous transformation in the rectangular coordinates system node pass through.
picking coordinate system [9]. Corresponding joint Refer to these points, the position posture is
variable may be computed by inverse kinematics described to series of picking the movement and
solution. Fig. 2 shows the picking robot manipulator action of the picking robot arm shown as Table 1.

Fig. 2. Picking trajectory of picking-arm for apple harvest.

Table 1. The process of picking and putting back for apple harvest.

Knot
P0 P1 P2 P2 P3 P4 P5 P6
point
Moving Moving Moving Putting
Initialization Moved Grasp Lifting Release
object apple Fruit Box Into box

One node Pi corresponds a transformation To

equation, thus solves correspondingly picking robot
arm transformation 0T6. From this obtaining the basic T3  0TB B Pi 1 3TT 1 ,
0
(2)
structure of the picking description: The picking node
Pi correspond the picking arm transform matrix 0T6,
transforming from one to another transformation is to where 3TT is the conversion to robot arm coordinate
achieve by picking arm joints movement. system {T} relative to picking manipulator; BPi and
B
Picking arm's working can be represented by a Pi+1 is respectively homogeneous transformation
series of knot points indicating it's position posture. between the node Pi and Pi+1 relative to the
How to defines and produces a series of points coordinate system {B}.
between the beginning point Pi and the end point
Pi+1, is the most important question on path planning
in the rectangular coordinates system space [10]. 2.3. Picking Space Circular Interpolation
Between two points the simple way is a spatial
Space arc is arbitrary three points determined the
translational motion and rotating around the fix axis.
spatial arc, related theoretical research has carried out,
After the movement time is assigned, may produce
it is difficult to solve the traditional matrix or Euler
movement controlled the linear velocity and the
angle rotation spatial interpolation algorithm on
angular velocity. Movement from the node P0 to the
linear interpolation theory, Hagras Hani designed
close picking apple node P1 or from any point P2 to
quaternion five-axis circular interpolation algorithm
the next point Pi+1 may be expressed from 0T3
3 to achieve a smooth movement from one point to
TT=0TB BPi, namely move from
another point smooth continuous interpolation;
Pervozvanski Anatoli A planned multi-channel multi-
0
T3  0TB B Pi 3TT 1 (1) arc trajectory algorithm by 3 points teaching

13
 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

programming [10]; to gain an accurate interpolation ay1, az} belong to XRYRZR vector can be obtained.
trajectory Sakai Satoru conducted timing non-linear Therefore, the homogeneous coordinate
error compensation in coordinate transformation transformation can be expressed from OX0Y0Z0 to
principle of circular interpolation [11]. Summarizing ORXRYRZR coordinates.
the above literature research methods, they will
convert generally the joint space issues to the flat  nx ox ax x1 
circular arc issue, to calculate picking circular n oy ay y1 
interpolation space generally "three-step" method oR T 
o  y (4)
shown as Fig. 3. Firstly, identify arc plane that is,  nz oz az z1 
 
convert three-dimensional issue to two-dimensional; 0 0 0 1
secondly calculate the interpolation point coordinates
according to two-dimensional interpolation 3) Homogeneous coordinate is oRo inverse
algorithm; finally convert the coordinate values Transformation from coordinate system ORXRYRZR
calculated to the value of the fundamental coordinate to XRYR ZR.
s y s t e m .
 nx ox ax (nx .x1  ny . y1  nx .z1 )
n o a (o .x  o . y  o .z ) 
oT 
oR  y y y x 1 y 1 x 1 
(5)
 nz oz az ( zx .x1  z y . y1  zx .z1 ) 
 
0 0 0 1 

From the above analysis, the process of spatial

circular interpolation process can be designed shown
as Fig. 4.

Fig. 3. Interpolation of joint space circle arc plane.

Through three node non-collinear in space P1, P2,

P3 can be determined the three-point-arcs and circles,
the line of intersection the plane were AB, BC, CA
between basic coordinate plane and three-point arc.
Constructing circular interpolation plane coordinate
system, look the P1 as the original point of the spatial
coordinate OR, P1P3 as the ZR axis, P1P3×P1P3 as YR
axis, XR axis determined by the right-hand rule, so
the arc will be fixed in ORXRYRZR plane. The points
P1, P2, P3 in the coordinate system represented in
ORXRYRZR. The homogeneous transformation is
given in Step 2.
Fig. 4. The flowchart of interpolation for space
1) To solve interpolation point coordinates circle arc plane.
according to plane circular interpolation algorithm;
2) To convert the coordinates value in ORXRYRZR
by the calculation from first step to the coordinate 2.4. Picking Robot Joint Spatial Interpolation
values in OX0Y0Z0. Algorithm
To achieve the second step, we must first solve
the coordinate axes direction cosine from the As for joint space trajectory planning of picking
coordinate system ORXRYRZR to OX0Y0Z0 and the robot arm, a series of positioning points such as the
homogeneous transformation determined from the starting point and end point given posture may be
origin of the coordinate translation. ZR axis and P1P3 selected in the joint coordinate system. It should
vectors are in the same direction: P1P3={x3-x1, y3-y1, select parametric trajectory according to joint
z3-z1}; YR axis same as P1P3×P1P2 vectors: P1P2={x2- interpolation and the functions satisfied interpolation
x1, y2-y1, z2-z1} thus P1P3×P1P2 can be obtained in points constraint, The constraints (functions) can
this way; XR axis determined by ZR×YR; while conclude picking points position and orientation,
unitising unit vectors {nx, ny1, nz}, {ox, oy1, oz} {ax, velocity and acceleration constraints of movement

14
 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

direction (initial point), leaving direction (lifting The equations can be solved as following
points), putting down and back direction of picking
manipulator while gripping apple, etc. Also be the a0   0
continuity constraints of the arm joint displacement, a  0
velocity and acceleration at the time interval. These  1
constraints ensure smooth and stable picking  3
planning. To meet the above constraints, you can a2  2 ( f   0 ) (11)
 t f
apply different types of smooth interpolation function
 2
to represent different joints trajectories planning. a3   3 ( f   0 )
 tf

2.4.1. Cubic Polynomial Interpolation

As for the each arm joint motion, when initial and
On the process of picking arm spatial the termination velocity are both zero, cubic
interpolation, the corresponding posture of the arm polynomial interpolation functions satisfy the
joint angles to the starting point and the termination requirements of smooth and stable movement is be
point can be obtained by inverse kinematics. So the expressed as
picking trajectory description can be represented for
smoothing interpolation function θ(t) to the 3 2
 (t )   0  ( f   0 )t 2  3 ( f   0 )t 3 (12)
termination and starting joint angle. To achieve 2
tf tf
higher stability picking movement, the trajectory of
each joint function θ(t) must be satisfied with
velocity and position constraint of two end points According to Equation (12), equation expression
four timely cases. Endpoints position constraint θ(t) of the joint angular velocity and angular acceleration
represents joint angle of termination posture and can be obtained, therefore
starting posture, assume the terminate joint angle θ(t)
represents value on the terminal time tf when the  6 6
 (t )  t 2 ( f   0 )t  t 2 ( f   0 )t
2
value of the moment, with the staring joint angle θ0  f f
represents value on the starting time t=0, namely  (13)
 (t )  6 (   )  12 (   )t
  (0)   0  t 2f
f 0
t 3f
f 0

 (6)
 (t f )   f
As for the picking arm with rotating joints, the
to ensure smooth and stable joint velocity, joint implementation picking operation of joint movement
velocity of terminating point and the starting point last 3 s by the experiment. According to timely field-
can be considered close to zero, namely control, the movement of shoulder joint must be
smooth namely, picking robot arm can provide with
 (0)  0 following states: Initially, wrist joint can be on
 (7) stationary position θ0=0o; while movement is over,
 (t f )  0 θf=90o, the joint velocity is 0. According to the above
requirements, wrist joint movement may be planned.
According to (6) and (7), the constraints of the cubic meeting the requirements, the cubic polynomial
polynomial only be confirmed interpolation function can used to plan movement of
wrist joint. given, substituting θ0=0o, θf=90o into (11),
θ (t )  a0  a1t  a2t 2  a3t 3 (8) obtain cubic polynomial coefficients

the movement equation for velocity and acceleration a0  0.0, a1  0.0, a2  2.35, a3  -1.63
of joint arm is solved as following
According to (12) and (13), trajectory of the wrist
(t )  a1  2a2t  3a3t 2 joint can be obtained, namely
  (9)
θ (t )  2a2  6a3t
 ( t )  2.35 t 2 -1.63 t 3
to obtain cubic polynomial the coefficients 
 ( t )  4.7 t -4.9 t
2
(14)
we substitute a0, a1, a2, a3 into the (5) and (6) the
( t )  4.7 -9.8 t
given constraints equations 
 0  a0 The curve simulation of cubic polynomial
  a  a t  a t 2  a t 2 interpolation joint trajectory is shown as Fig. 5. It can
 f 0 1 f 2 f 3 f
 (10) be seen that angular displacement curve, an angular
 0  a1
velocity and acceleration curves were hyperbolic sine
0  a1  2a2t f  3a3t 2f
curve line, parabola and straight lines, respectively.

15
 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

Fig. 5. Cubic polynomial interpolation of joint trajectory for picking-arm.

2.4.2. Linear Interpolation with Parabolic Now given 2 parabolic interpolation trajectory
Transition with the same duration time ta between two end
points, it's constant acceleration  which is equal
In the cubic polynomial trajectory planning, the and opposite. This kind of trajectory planning shown
linear interpolation function is chosen for a given as Fig. 8 exist several solutions, so its trajectory is
start and end points shown as Fig.6. On the case of not unique, but all trajectory are symmetrical around
actual picking trajectory control, sometimes joint the position midpoint θh and the time midpoint th.
movement velocity of starting and end points is not
smooth and non-continuous, even acceleration and
the emergence of the rigid impact of the two end
points are so large that picking arms "jitter" occurs.
Naveen Kuma studied a simple linear interpolation
may cause vibrations [12]. So it is necessary to revise
linear function interpolation scheme, the specific
method as following: Set a parabola as two
interpolated end points buffer. Acceleration of the
buffer region will be constant due to the second
derivative is constant. So may ensure the moving
speed a smooth transition between the start and end
points, that the position and velocity of whole picking
trajectory is continuous track. Parabolic function and
the original linear function is smoothly connected to Fig. 8. Multiple solutions and symmetry for picking
picking trajectory shown as Fig. 6. trajectory.

In accordance with control requirement, picking

trajectory ensure smooth, continuous, i.e. average
angular velocity of linear parabolic trajectory
segment must be equal to its velocity of destination
time, so the following formula

h  a ,
ta  (15)
t h t a

where θa-ta corresponds to the joint angle of the

Fig. 6. Linear interpolation trajectory between two picking
points.
parabolic duration time ta. θa can be obtained
according to (15)

1
a  0  ta (16)
2

We can obtain the total joint exercise time tf from

the starting point to the end point, namely, tf =2th

1
h  (0   f ) (17)
2

Fig. 7. Linear trajectory with parabolic transition domain. From (15) to (16)

16
 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

ta 2  t f ta  ( f  0 )  0 (18) According to (18) and (19), compute joints

positions a1 and joint velocity 1 in the transition
where θ0, θf, tf are generally known to determine the end time, exist
appropriate  and ta according to (13), may obtain
the corresponding trajectory. Using identified the  1
 a1  15  (  42  0.59 )  22.3
2  

acceleration value  and (18) calculate the  2 (22)

corresponding ta 1  1ta1  (42  0.59) / s

tf t f 2 4( f   0 ) (19)

2) For the picking trajectory of wrist joint, joint
ta  
2 2 velocity 2 =27°/s2 by measuring experimentally,
calculate
According to (18), ensure ta solvable, the
acceleration value  may be satisfied  27 2  32  4  27(75  15)
3
ta 2  (  )s  1.33s
4(f  0 )  2 2  27 (23)
  (20)   15  ( 1  27  1.332 )  38.88
t 2f
 a 2 2
The (20) gives the new proof by obtaining for
equality, then the linear length of the tracking So, 2  2 t a 2  (27  1.33)  / s  35.91 / s
segment is reduced to zero, two transitional domain is According to the above calculated values, Fig. 9,
the entire trajectory of all, their joint velocity (slope) Fig. 10, Fig. 11 can be plotted as transition linear
in convergence point is equal; the larger acceleration interpolation with parabolic trajectory curve.
 value is, the shorter the length of the transition The linear function with a parabolic interpolation
region is, if the acceleration is infinite, the tracking transition trajectory planning shows the
returns to the case of a simple linear interpolation. corresponding acceleration, constant speed and
The starting angle of shoulder joint θ0 is 15°, the end deceleration movement law of arm joint motors.
angle θf is 75°, the total control time tf=3 s in the Wherein the curve of joint position, velocity and
picking trajectory by measuring experimentally. Then acceleration versus time shown as Fig. 9 - Fig. 11.
design on linear trajectory with parabolic transition The movement of three joint in picking trajectory all
can be obtained. According to (19), we can set up the go through a set of points, scheme of transition
range of the acceleration. If the known conditions region with a linear parabolic trajectory algorithm
meet (20), there exist  ≥ 26.67°/s2, then shown as Fig. 12 can be adopt. Joint movement alone
picking path go through these points, by the joint
1) As for the t picking trajectory of elbow join, angle i , j , k indicate the three adjacent
setting 1 =42°/s2, calculate transition time ta1 trajectory points, and every two adjacent points in
according to (20) picking path connected in the form of a linear
function, and all the points are taken advantage of the
422  32  4  42(75  15) smooth parabola transition.
3 (21)
ta1  (  )s  0.59s
2 2  42

Fig. 9. The displacement trajectory curves with greater variance of acceleration.

17
 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

Fig. 10. The velocity trajectory curves with greater variance of acceleration.

Fig. 11. The acceleration trajectory curves with greater variance of acceleration.

Fig. 12 shows that the parabolic trajectory point P3 (10, 3, 6). In the experiment, set N=0-45,
segments are used to plan the transition domain namely 45 insertion points is planned between P1, P3.
linear function, if the acceleration is sufficiently According to the circular interpolation algorithm, the
large, arm movement joints can not really reach those Matlab simulation experiment showed space circular
picking path points, but the actual trajectory is also interpolation curve shown as Fig. 13.
very close to the ideal trajectory points.

Fig. 12. Line trajectory of multiple line with parabolic

transition domain. Fig. 13. The simulation of space arc interpolation.

3. Experimental Simulations and Analysis 2)The picking arm in the apple-picking work
mainly is continuous operation movement shown as
3.1. Experimental Technique and Steps Table 1, achieving an operating cycle, including the
picking realization and picking playback, needs to
1) Picking robot arm started from relative to the complete the path planning bypassing 6 points, the
base coordinates initial station P1 (10, 3, 4), through planning method is cubic curve interpolation with the
the circular point P2 (11, 3, 5) then arrived the end parabola, its point value may be solved through (15),

18
 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

(17) and (19) interpolation computation, presently trajectory interpolation, if picking arm run trajectory
using the Matlab simulation, the wrist joint angle was err great, mean interval gas of discrete point deep.
carried on cubic interpolation path planning with Therefore it is necessary to reduce the interpolation
parabola, the record was tested normalization discrete distance, to cause to picking the arm
average value of three longest time at the minimum approach expected the trajectory in the small
speed of wrist joint, and timely the displacement, the precision. The interpolation experiment simulation
speed, and acceleration curve on the performance also indicated that, the denser is interpolation spot,
wrist joint picking shown as Fig. 12 according to the more really approaches trajectory curve.
following step. In the very short time, the picking arm trajectory
A. Planning preparation: Suppose the initial is generally regarded as the movement velocity as the
picking position coordinates X0=(0, 0, 0) ， millimeter level of straight line. Seen from Fig. 13, a
X 0  (0, 0, 0) ， X0  (0, 0, 0) . coordinate of picking path point is interpolated, the
arm controller drive the joint motor to complete this
B. Planning start: input endpoint parameter vector. interpolation, to guarantee the stability movement,
C. Confirms whether Xf lives in the planning the time interval cannot be long, the time can be
picking space or not, otherwise returns to the starting considered clock cycle of picking arm controller. At
point phase, discarding parameter on picking the the millisecond level of interpolation time interval,
endpoint, rebuilt picking planning space. the distance between 2 interpolation points is
D. The partition reaches from X0 to Xf0 through approximately proportional to the velocity of
Xf2, again returns to X0, travelling each section last planning path movement. So long as interpolation
2s, the use physics parameter of the picking robot interval is enough small, the precision on trajectory
arm and formula of cubic linear interpolation with movement can be guaranteed, also the stable request
parabola transition satisfy step A to calculate guaranteed correspondingly. If the picking path
planning require high accuracy to ensure the picking
 X f 1  ( 11.2,9.4, 109.6)
 stability, likewise seen from Fig. 13, in order to
 X f 2  (13.2, 10.8, 99.3) guarantee the trajectory moving stability to increases
the precision, 2 characteristic interpolation distance
 X f 1  (9.3,13.4,134.1) is enough small, requiring the time interval of
 interpolation change evenly, Exactly, the picking
 X f 2  ( 15.4,16.2,164.2)
path planning curve reveal the smooth- stability with
the parabola transition.
 X f 1  (13.5, 27.4,104.3) Seen from Fig. 13, using the cubic curve
 
 X f 2  (17.2, 21.1,114.7) trajectory interpolation with parabola may guarantee
the picking manipulator the continuity of speed and
E. Plan picking trajectory in method of cubic the displacement in to pick on the condition of field
linear interpolation with parabola transition, compute picking, namely may complete the apple-picking the
displacement, velocity and acceleration curve at the smooth movement of path planning, specially,
interval of 0.01 s. acceleration continuity reveal the picking
manipulator not to the vibrate and guarantee the
stability of coming picking localization and capture.

4. Conclusions

This paper aims at the picking arm running along

specific path sequence in advance in view of the
most super time, according to path planning request
and the restraint of picking robot arm, on the base of
position posture and object description, proposed
cubic curve interpolation with the parabola transition,
completed trajectory planning of the picking arm.
The simulation experiment result indicated that the
algorithm may realize its smooth path tracking under
on the condition of field picking, and that picking the
Fig. 14. Trajectory planning curves of picking manipulator may not have the big vibration. This
manipulator. algorithm provided the reliable guarantee on the
following apple picking including the localization
and the capture stability.
3.2. Experimental Results and Analysis
As a whole, the picking robot arm achieve space
trajectory curve and the circular arc by discrete

19
 Sensors & Transducers, Vol. 162 , Issue 1, January 2014, pp. 11-20

Acknowledgements [9]. Hagras Hani, Colley Martin, Callaghan Victor, Carr-

West Malcolm, Online Learning and Adaptation of
This project was funded by National Nature Autonomous Mobile Robots for Sustainable
Science Foundation of China (No. 31101075) and Agriculture, Autonomous Robots, 13, 1, 2002,
pp. 37-52.
Technology Innovation Foundation of Northwest [10]. Pervozvanski Anatoli A., Freidovich Leonid B.,
A&F University (No. QN2013051, No. QN2011069). Robust Stabilization of Robotic Manipulators by PID
Controllers, Dynamics and Control, 9, 3, 1999,
pp. 203-214.
References [11]. Sakai Satoru, Iida Michihisa, Osuka Koichi, Umeda
Mikio, Design and control of a heavy material
[1]. Chen C., Ong S., Nee A., Zhou Y., Haptic-based handling manipulator for agricultural robots,
interactive path planning for a virtual robot arm, Autonomous Robots, 25, 3, 2008, pp. 189-198.
International Journal on Interactive Design and [12]. Naveen Kumar, Jin-Hwan Borm, Vikas Panwar and
Manufacturing, 4, 2, 2010, pp. 113-121. Jangbom Chai, Tracking control of redundant robot
[2]. Fellner Gene, Pitts Wesley, Zuss Mark, Beyond the manipulators using RBF neural network and an
sensible world: a discussion of Mark Zuss, the adaptive bound on disturbances, International
practice of theoretical curiosity, Cultural Studies of Journal of Precision Engineering and
Science Education, 7, 3, 2012, pp. 745-754. Manufacturing, 13, 8, 2012, pp. 1377-1386.
[3]. Wicaksono Handy, Khoswanto Andry, Kuswadi, [13]. He Bei, Liu Gang, Ji Ying, Si Yongsheng, Gao Rui,
Behaviors coordination and learning on autonomous Auto Recognition of Navigation Path for Harvest
navigation of physical robot, Telkomnika, 9, 3, 2011, Robot Based on Machine Vision, Computer and
pp. 473-482. Computing Technologies in Agriculture IV,
[4]. Monta M., Kondo N. and Ting K. C., End-Effectors Springer, Boston, pp. 138-149.
for Tomato Harvesting Robot, Artificial Intelligence [14]. Kuffner James, Nishiwaki Koichi, Kagami Satoshi,
Review, 1998, 12, 1, pp. 11-25. Inaba Masayuki, Inoue Hirochika, Motion Planning
[5]. Viola Robert, Walsh Jace, Melka Alex, Womack for Humanoid Robots, Robotics Research, Springer,
Wesley, Murphy Sean, Riboldi-Tunnicliffe Alan, Berlin / Heidelberg, 2005, pp. 365-374.
Rupp Bernhard, First experiences with semi- [15]. McCarthy C. L, Hancock N. H and Raine S. R.,
autonomous robotic harvesting of protein crystals, Applied machine vision of plants: a review with
Journal of Structural and Functional Genomics, 12, implications for field deployment in automated
2, 2011, pp. 77-82. farming operations, Intelligent Service Robotics, 3, 4,
[6]. Yin Yingjie, Hosoe Shigeyuki, Luo Zhiwei, A mixed 2010, pp. 209-217.
logic dynamical modeling formulation and optimal [16]. Viola Robert, Carman Peter, Walsh Jace, Frankel
control of intelligent robots, Optimization and Daniel, Rupp Bernhard, Automated robotic
Engineering, 8, 3, 2007, pp. 321-330. harvesting of protein crystals — addressing a critical
[7]. Katuin J. E., Schreuder A. N., Starks W. M., Doskow bottleneck or instrumentation overkill? Journal of
J., The Use of Industrial Robot Arms for High Structural and Functional Genomics, 8, 4, 2007,
Precision Patient Positioning, in Proceedings of the pp. 145-152.
AIP Conference, 680, 1, 2003, pp. 1138-1141.
[8]. Yan Xuesong, Wu Qinghua, Liu Hammin, An
improved robot path planning algorithm, Telkomnika,
10, 8, 2012, pp. 1948-1955.

___________________

2014 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved.
(http://www.sensorsportal.com)

20