Vous êtes sur la page 1sur 11

DESIGN OF ROBOTIC ARM

Z1
3
Y1
Link 2 Z3
1
4
Link 3
2 X1 Z2 X3 5
Z4, 5, 6
Y3
6
Base
X2
Tool X4, 5, 6
Y2
Y4, 5, 6

Figure 1: Coordinate system geometry for the six-axis robot

Joint 3 is located along a link 20 inches from the base and acts as an elbow. The fourth,
fifth and sixth joints are also concurrent and are located 14 inches from joint three; these
joints act as a wrist. The wrist has a roll axis, followed by a pitch axis, followed by
another roll axis. Having three concurrent joints simplifies the motion of the wrist. For
example, a rotation of joint four does not result in a translation of joints five and six. This
minimizes the complexity of the equations of motion of the wrist.

To develop the articulated arm design of the robot I had to develop equations for
both forward and inverse kinematics. Forward kinematics gives the location of the end
effector in the universe frame. The inverse kinematics gives the joint angles needed in
order for the to the robot arm reach the goal frame. Using derivations from Reference 1,
kinematic equations for our robot geometry were formulated. From this formulation,

1
computer code that modeled the robot kinematics was written in Matlab.

2
Forward Kinematics
Forward kinematics is the mathematics behind the process of taking joint angles
and specified link lengths and calculating the position of the end effector in the universe
frame. To calculate the position of the end effector in the universe frame a series of
matrix calculations must be performed that find the position of the end effector relative to
each robotic joint. The universe coordinate frame (frame U) is considered to be non-
moving and to be located concurrent with the Joint 1 frame which is at the shoulder of the
robot (Figure 1). In our robot design joints one and two are also coincident with one
another. Although joint two is coincident to joint one, its frame is negatively rotated by
90degrees about the x-axis. The matrix that describes the position and orientation of joint
1 in universe is designated as T1toU. Similarly, joint twos rotation and translation into
the joint 1 frame is T2to1. To get to Joint threes frame in the universe frame a translation
was made along the x-axis and z-axis, known as T3to2. Joint four is described in joint
three by a negative rotation about the z-axis and a translation along the x-axis and z-axis,
named T4to3. Joint fives frame has a z-axis rotation, and no translation because it is
concurrent with joint four, it is named T5to4. Joint sixs frame is described in joint fives
frame by a negative z-axis rotation and no translation, named T6to5. Once all the
rotations and translations have been placed in a 4X4 matrix for each joint, they are then
multiplied together in the following order:

ToolinU=T1toU*T2to1*T3to2*T4to3*T5to4*T6to5

This series of matrix multiplications gives the position and orientation of the end-effector
(the tool) in the universe frame. Each rotational joint has a specific joint angle which is
the angle of rotation of the joint axis from its home position. In Figure 1, each joint angle
is designated using the Greek letter followed by the joint number.

Once the forward kinematic equations for our robot were formulated, we wrote
Matlab routines that implemented these equations and plotted a three-dimensional picture
of our robot for a specified set of joint angles.

3
Inverse Kinematics
As described above, forward kinematics is the process of finding the position of
the end effector in the Universe frame if all of the joint angles are known. However, in
order to control the robot one must answer the following question: What sets of joint
angles will place the end effector into a desired position and orientation?

The inverse kinematic equations give the joint angles that will place the end
effector in the goal frame. The inverse kinematic equations are a set of many equations
which must be solved simultaneously. Solution of these equations is complicated by the
existence of multiple solutions and by the possibility of singularities. We used an inverse
kinematic formulation described in Reference 1 in order to solve for the joint angles
sequentially, starting with joint 1. For joints one and three, two possible solutions are
given for each joint angle, yielding Theta1_1, Theta1_2, Theta3_1 and Theta3_2. Using
the atan2 function to find possible joint angle combinations of joints one and three, four
solutions are found for Theta2, Theta2_1 through Theta2_4. Joint fours angle has the
rotation of the goal frame in relation to the universe frame taken into account. Joint four
also takes into consideration all possible combinations of joint two in relation to which
combination of Theta 3 and 1 were used. Joint four has eight possible solutions of
Theta4. The solutions for joints five and six are the same as for joint four in that they also
have eight possible solutions. The equations are the same except for joint five takes
Theta4 into consideration and joint six takes Theta5 into consideration in the atan2
function.

Since there are multiple solutions to the inverse kinematic equations, some
method must be used to select between them. With our formulation, once joint angle
Theta1 and Theta3 are chosen Thetas 2, 4, 5 and 6 are set. For example if the angle for
Theta1_1 and Theta3_2 was used, then Theta2_4, Theta4_4, Theta5_4 and Theta6_4
must be used because of the structure of equations.

Once the inverse kinematic equations were formulated we wrote Matlab routines
that implemented the equations. The inputs to this code were the geometry of the robot

4
and the desired position of the end effector. The output was one or more sets of joint
angles that would put the end effector into the desired position and orientation.

Design Method
The next step in our research was to use our kinematics code to select robot link
lengths that would work for making nano-films. To define the link lengths of the robot
arm we defined a work area of the counters where the solutions used to make the nano-
films will sit. The counter is 24 inches wide. We also determined it would need to go side
to side a total of 40 inches, 20 inches on each side of the base of the robot. The robot
should also be able to reach 2 feet above the counter. We tested a variety of link lengths
and combinations of these lengths. The link length tested for link two was 28 inches and
link three was 20 inches, the next combination was 32 and 16 inches.

End effector in Goal frame

Figure 2: The robot with link lengths 20 and 14 inches is able to reach the Goal frame
with the end effector orientated down.

5
Workspace

Goal frame
End effector

Figure 3: The robot is trying to a point too close to the base and is not able to fold in it
on itself to place end effector in Goal frame.

The third combination was 24 and 24 inches, fourth combination was 19 and 29 inches,
and the fifth combination was 5 and 31 inches. A variety of points were chosen within the
work envelope that the robot would need to be able to reach.

Link Lengths
link 2 link 3 Goal Was Goal Point
(in) (in) Point Reached
28 20 (5,44,18) Yes
(0,5,0) No
(-5,5,2) No
(-7,10,4) Yes
(-5,8,2) Yes

32 16 (5,44,18) No
(0,5,0) No
(-5,5,2) No
(-7,10,4) No
(-5,8,2) No

6
24 24 (5,44,18) Yes
(0,5,0) No
(-5,5,2) Yes
(-7,10,4) Yes
(-5,8,2) Yes

19 29 (5,44,18) Yes
(0,5,0) No
(-5,5,2) No
(-7,10,4) Yes
(-5,8,2) No

5 31 (5,44,18) No
(0,5,0) No
(-5,5,2) No
(-7,10,4) No
(-5,8,2) No

Table 1: Results of experiments to determine the work envelope for a series of different
link lengths

We also tested how close to the base the end of the robot arm could reach, meaning how
much could it fold in on itself. The results of this testing can be seen in Table 1.

From Table 1, one can see that the link length combination that worked best was
28 inches for link two and 20 inches for link three. However, subsequent calculations
showed that the motor torque requirements for such long lengths were excessive. The
motors would have to be very large and expensive in order to hold up and move the arm
around. In order to reduce the required torque we shortened the link lengths but kept the
ratio of link lengths the same. The new link length for link two is 20 inches and link 3 is
14 inches. With these link lengths we can reach within 6 inches of the base on y-axis and
anywhere on the x-axis except between -1 to +1. The point (0,6,0) produces imaginary
numbers for the joint angle Theta3. The farthest straight out the arm can reach is
34inches. We would like the arm to be able to work in a space with a height of 18inches.

7
If the end-effector of the arm reaches out to 18 inches in height the farthest out on the y-
axis it can reach is 28.5 inches. The farthest the arm can reach in the x and the y while
reaching full height of 18 inches is 20 inches in the x and 20 inches in the y. If the arm
was to reach straight out, it could extend to 34 inches, thus it can also reach 34inches
straight to the left and right of the base of the robot. We expect the robot will be doing
most of the work six inches off the counter top, with this height limit the arm can reach
up to 23 inches in the x while reaching 24 inches in the y. Within these exterior limits the
arm can reach in a sphere shape 180 degrees around the base, and 180 degrees vertically
around the base. Each joint has the ability to move 180 degrees. With these link lengths
we are able to reach points throughout our desired work envelope.

Motor Selection
Once the basic geometry of the robot was chosen, it was possible to calculate the
maximum torque required for each joint motor. The torques were calculated by
estimating the weight the motor shaft would have to be holding and multiplying it by the
distance from the center of gravity of the weight back to the motor shaft. The weight
included motors farther up the arm, their casings, the weight of the link arm and bearings.
We assumed the motor casings and the links are made out of aluminum, and using the
density of aluminum multiplied by the volume of material needed. The calculated
maximum torques for each motor are in Table 2.

Using the calculated torque values, motors were chosen so a solid model of the
design could be made in Autodesk Inventor. The motors that were chosen are high torque
stepper motors from Anaheim Automation. The specific motor model numbers are shown
in Table 2. Stepper motors were chosen for the amount of torque they output as well as
their precision. These motors are also open loop control, which means they dont give any
feedback. These motors are also small and light weight which is crucial in keeping the
torque lower for the motors father back in the arm.

8
Joint Maximum Required Motor Motor Maximum
Torque (oz.-in.) Selected Torque Rating
(Anaheim Automation (oz.-in)
Model #)
1 4315 42N3 4365
2 983 34K2 1535
3 109 23L1 126
4 83 17Y4 100
5 59 17L1 63
6 19 14Y1 25
End effector 10.7 GM12 14

Table 2: Torque requirements and motor selections

Design of the End Effector


Once the link lengths and motors were selected, we moved to the task of
designing the individual pieces of the robot in Inventor. For this design work, we chose to
start with the end effector that is used to hold and rotate the nano-film slides. The end
effector must attach to the rest of the robot arm spin separately to agitate the slides while
in solution and being rinsed. An exploded view of the end effector assembly is Figure 2.
The motor housing is a snug fit around the shape of the motor. It has a small hole at the
bottom for the shaft to rotate, and two small holes that align with the motors mounting
holes. The casing cap has two holes for the wires to come out the top. The cap also has a
clearance hole for the excess shaft out the top of the motor. The cap is screwed to the
casing, and the top of the cap as a threaded shaft to screw into the end of the robot, joint
six. Off the bottom of the motor shaft is a shaft enlarger, this attaches to the motor shaft
by a set screw and rotates with it. At the bottom of the rotating shaft is the slide holder.
There is square base where four slides will be able to be attached. The slides are attached
to the square base by covers that cover part of the slide and go over the top of them and
attaches to the square base. The slides are held in place by eight plastic set screws
through the covers. With this design four sets of films will be able to be made at once.

9
Figure 2: Exploded view of the end-effector

There were problems in creating the part that actually held the slides. It could not
have a diameter larger than 1.8 inches at the most. The beaker that holds the solution is 2
inches in diameter, but I have to allow for slop in accuracy of placing the end effector
into the beakers exact center, and wobble while rotating. With this current design of the
slide holder the diameter is 1.6 inches. The reason for having such a large diameter is the
each slide is an inch wide; and all the slides must be placed in a circular pattern so they
all see the same current of the solution moving over them when spun.

Once the pieces are modeled, mechanical drawings were created of the complete
end effector. When creating mechanical drawings, everything must be dimensioned. If
holes are threaded, those must be noted with what kind of threading. The key in creating
mechanical drawings is setting the tolerances. Critical part features must be manufactured
to a tight tolerance. We have given these critical features a tolerance of .005 inches.
An example of this is on the casing of the motor mounting holes, they must be .410
inches apart or the motor will not be able to be mounted.

Conclusions:
This summer research has been very successful. I learned a great deal about the

10
formulation and programming of robot kinematics. Using our Matlab code we were able to
design and analyze the geometry of the entire articulated robotic arm. We then performed
torque calculations and used these calculations to select motors for the robotic joints.
Finally, we used Inventor to design an end-effector that can be used to manufacture up to
four nano-film slides simultaneously.

Future work will focus on the design of the individual pieces of the articulated arm.
The wrist of the robot is made up of joint four, five and six and they will all be concurrent
with each other. The wrist will have roll, pitch and roll angles of rotation. The motors have
been chosen and some have been ordered. A materials list of the current set screws, screws,
Aluminum 6061-T6 and Nylon 6/6 has been ordered from McMaster.
When the pieces of the robot arm are completed, they will be put together and the programming of the
robot will be started. In my continuing research I will now able to finish the drawings in Inventor and
will have a completed six-axis articulated robot arm.

11

Vous aimerez peut-être aussi