QArm Lab Procedure

Trajectory Generation
Setup
1. It is recommended that you run this lab individually.

2. Find an object that is roughly cuboidal in shape, with a length dimension under 5 cm and weight
under 100 grams (for example, a piece of Styrofoam). This will be used during the pick and place
task.

3. Move the QArm manipulator to the home position and turn ON the unit using the power switch
located on the rear side of the base. Once powered, the manipulator should hold this position.

4. Launch MATLAB and browse to the working directory for Lab 5 – Trajectory Generation.

Navigate
1. Open the Simulink model Navigate.slx. You will use this model to implement a simple trajectory
generator. the inverse kinematics of the Quanser QArm manipulator.

2. Replace the Inverse Kinematics and Forward Kinematics functions with the completed versions from
previous labs.

3. Open the function called cubicSpline. This function contains the incomplete code for the four
coefficients given an initial and final setpoint. Complete lines 4, 5, 6 and 7. Note that each of the four
coefficients is a 3x1 vector, representing the x, y and z splines. Base your result on equation 6 from
the Trajectory Generation Concept Review for zero speeds, that is, 𝑥̇ 0 = 𝑥̇𝑓 = 0. Close the function
when complete.

4. Open the function timing. Set line 4 to keep the time-constrained between 0 and 𝑇 .
HINT: Use the mod function in MATLAB. Try mod(5,2) and mod(9,5) in the command window.

5. Complete line 7 to set the 4x1 time vector to output 1, 𝑡, 𝑡 2 and 𝑡 3 .

6. Note that the two position waypoints [0.25 0.25 0.10]𝑇 and [−0.25 0.25 0.10]𝑇 are similar to the joint
space waypoints you used in the Low Level Control lab application. Note that 𝑇 is set to 5 seconds,
and it will take 5 seconds to traverse between the points (10 seconds for a complete cycle).

7. Prior to running the model, open the model’s Configuration Parameters and verify that they are
configured as follows:

i. Solver type: Fixed-step

ii. Solver: ode4 (Runge-Kutta)

iii. Fixed-step size (fundamental sample time): 500 Hz

8. Ensure that the workspace around the manipulator is clear of obstacles and stay well outside the
reach of the manipulator.

9. Build and deploy the model using the Monitor & Tune action. Once started, the model will actuate
the manipulator to traverse between the two waypoints. Does the manipulator follow a circular arc
between the two points or a straight path? Take notes.
1
 10. Open the Tracking scope. Identify the desired setpoints, the desired task space trajectory being
passed to the inverse kinematics module, as well as the measured position out of the forward
kinematics module. Comment on the performance and take a screenshot of one complete cycle.

11. Stop the model.

Welding
1. Open the Simulink model Welding.slx. Copy over the cubicSpline function that you completed in the
Navigate.slx model and copy over your qarmForwardKinematics and qarmInverseKinematics
functions from previous labs. The function waypointNavigator has been completed for you.
Comment on what this function is accomplishing.

2. Right-click on a blank space in your model and open the Model Properties. Browse to the callbacks
section, and then to the InitFcn section. Set and apply the, A, B, C, D and default setpoints to the
following,

𝐴 = [0.25 0.25 0.1]𝑇

𝐵 = [0.25 0.50 0.1]𝑇

𝐶 = [−0.25 0.50 0.1]𝑇

𝐷 = [−0.25 0.25 0.1]𝑇

𝑑𝑒𝑓𝑎𝑢𝑙𝑡 = [0.25 0.25 0.2]𝑇

3. Set the waypoints to the correct order, that is, default, followed by A, then, B, C, D, back to A. Press
Apply and Okay to close the Model Properties menu.

4. Prior to running the model, open the model’s Configuration Parameters and verify that they are
configured as expected in the Navigate section.

5. Build and deploy the model using the Monitor & Tune action. Once started, the model will actuate
the manipulator to traverse between the waypoints. Does the application weld as expected from the
Lead Through laboratory? Take notes on the speed of welding between A to B, and B to C etc.

6. Repeat this experiment with the duration set to 2s. Is the manipulator able to weld correctly? Why?
Take notes.

7. Stop the model after a few cycles.

Assembly
1. Open the Simulink model Assembly.slx. Copy over the cubicSpline, qarmForwardKinematics and
qarmInverseKinematics functions from the Welding.slx lab. You will notice that the model is the
same.

2. Open the Model Properties menu to access the callbacks tab and InitFcn section. Set the positions
A, B, C and D to the same ones from the welding application. Set default to,
𝑑𝑒𝑓𝑎𝑢𝑙𝑡 = [0.00 0.375 0.2]𝑇

3. Set the waypoints to be default, then A, then A again (representing the time it would take for the
manipulator to take action at this location), back to default, then B, then B again, then default, C, C,
default, D, D. Press Okay.

2
 4. Build and deploy the model using the Monitor & Tune action. Once started, the model will actuate
the manipulator to traverse between the waypoints. Does the manipulator move as expected?

5. Take notes on any differences and/or any screenshots as you see fit.

6. Stop the model.

Pick and Place

1. Open the Simulink model PickAndPlace.slx. Copy over the cubicSpline, qarmForwardKinematics and
qarmInverseKinematics functions from the Welding.slx lab. You will notice that the model is still the
same, except for an added function actuateGripper.

2. Open the Model Properties menu to access the callbacks tab and InitFcn section. Set the pick
position to a desired positive x and negative y value (pick x and y values such that the combined
magnitude √𝑥 2 + 𝑦 2 is larger than 0.25 m to avoid proximity and collision with the base) and a z value
of 0.03 m (this is to avoid contact with a table).

3. Set the pick_high position to be the same as the pick position, but with a z component of 0.15 m
instead.

4. Set the place position to a desired positive x value and positive y value with a z component of 0.15
m as well.

5. Set the waypoints in sequence to pick_high, pick, pick_high, home, place, and then home again.
Why is this particular sequence selected?

6. In the model, open the function actuateGripper. This implements a simple state machine to check if
you are close to a desired setpoint and correspondingly open or close the gripper. At line 10, set the
threshold to 0.01 m. This implies that this state machine will consider the manipulator ‘at’ the setpoint
if within 1 cm of it. You may want to relax this threshold later if your manipulator fails to identify arrival.

7. At line 13, replace condition with code that correctly implements the logic to actuate the gripper to
the close position when close to the pick setpoint. When this condition is true, you want
gripper_state at line 14 to be 1.

8. At line 18, replace condition with code that correctly implements the logic to actuate the gripper to
the open position when close to the place setpoint. When this condition is true, you want
gripper_state at line 19 to be 0.

9. With a completely clear workspace (a dry run without any object), build and deploy the model using
the Monitor & Tune action button. Once started, the model will actuate the manipulator to
traverse between the waypoints. Does the manipulator move and actuate the gripper as expected?

10. Repeat the previous step by placing the object you selected at the pick position before the
manipulator arrives there. Does it pick the object and then drop it off at the place position correctly?

11. Stop the model, and close all models

12. Power OFF the manipulator using the switch at the rear end of the base and bring it back to rest
position. Close MATLAB.