Laboratory Notebook

Week 1:

Discussed the basic components of the project. The group decided that the four main 
components to this project are communication, graphing, coding and data analyzation, 
and controls. We will start off with using a 3D printed base, an Arduino Uno, ultrasonic 
sensors, and two XBee's as the basic components.

----------------------------------------------------------------------------------------------------------------------------------

Week 2:

Discussed about using an XBee for communication from the Arduino Uno to a computer. A 
software called ROS could be used to perform the graphs to display what the sensors have 
detected. An example can be seen in Figure 1 below.

Figure 1. Example of an ROS map.

----------------------------------------------------------------------------------------------------------------------------------


Week 3:

Began to code the Arduino to send serial commands to MatLab in order to analyze and 
plot the data. A data plot sample can be seen in Figure 2 which shows a set of data that 
has been detected and plotted without any filtering. This is done by having the Arduino 
print serial commands to the computer, which will then be read by a serial object created 
in MatLab. All of the math has been done on the Arduino, which determined the X and Y 
coordinates of the surfaces detected by the ultrasonic sensor. This information is printed 
serially with a tag as the first digit which corresponds to the X or Y value of a detected 
object.  The information is sent in ASCII for each piece of data sent, and must be 
converted. After the conversion, points are created and then updates onto the plot after 
reading values twenty-five serial readings. The test rig can be seen in Figure 3.

Note: All the codes are posted onto the documentation portion of this blog.

Figure 2. Plotted points from the ultrasonic sensor.

Figure 3. Test Rig.


Changes that have been made:

The group decided to try to use the ROS software described earlier in order to plot the 
data. The software has been used before with projects that are similar to ours. After 
downloading the software, the group realized that Linux terminal commands would need 
to be used to create the environment and to plot points. Because of this, the group 
decided to use MatLab instead since the members are familiar with this software.


Plans for next week:

The plan for next week is to put the ultrasonic onto a mobile robot and attempt to read both 
the robots location and any objects detected by the ultrasonic sensor. Problems that may 
occur are latency, processor speeds, and serial communication speeds. Research will be 
conducted on STP and optimization of signal processing. A chassis for the mobile robot 
has already been made and can be seen in Figure 4.

Note: Documentation on the chassis can be seen here on the main website under models.

Figure 4. Mobile robot 3D printed chassis.

----------------------------------------------------------------------------------------------------------------------------------

Week 4:

A GUI was made in MatLab in order to display 3 plots: unfiltered data, filtered data, and a 
grayscale image of the data in order to see the probability of a wall being in an area based 
on how many times a wall has been detected in that location. This GUI can be seen in the Figure 5.


Plans for next week:

An encoder will be made in order to localize the mobile robot through a room. This part will 
be designed in CREO Parametric and will be 3D printed by a MakerBot. 

Along with the encoder, an algorithm will have to made in order to let the mobile robot 
know not to move to a location that it has already been to. This could be done by making 
landmarks of areas of high concentration of points which would correlate to the mobile 
robot being sure that there is an enclosed will at that location. The mobile robot will then 
stay as far away from this landmark as it attempts to maneuver and map the rest of the 
room until all areas have been marked with these landmark. This concept will be 
integrated into code once the mobile robot is moving and working.

Figure 5. MatLab GUI for graphs of unfiltered data, filtered data, and grayscale image of the plot.

----------------------------------------------------------------------------------------------------------------------------------


Week 5:

An encoder was made for localizing the robot as seen in Figure 6. An HC-020K 

photoelectric encoder will be used to count the ticks from the encoder. This encoder was 
made to have only two different states per revolution to make processing the current 
location easier and less computationally intensive. The part shown below will be 3D printed using a MakerBot.


Figure 6. Motor encoder with one tick per revolution.


Plans for next week:

All the components will be put together and a basic code for localization and object 
detection will be put together on Arduino IDE. The Xbee's will also be configured in order to 
send information from the Arduino to another XBee which is connected to a computer for 
data analyzation and graphing.

----------------------------------------------------------------------------------------------------------------------------------

Week 6:

All components have been put together and function. Values calculated with the ultrasonic 
sensor can be successfully processed and sent over the XBee to another XBee which 
communicates serially with MatLab. The test rig can be seen in Figure 7.


Figure 7. In Red: Xbee, Green: Ultrasonic sensor mounted onto a micro servo, Pink: 3D printed encoder
 Light Blue: Photoelectric encoder, Yellow: L293D Motor Driver, Dark Blue: Arduino.

In order to localize the direction that the mobile robot is facing, a control systems would 

need to be created, or a compass could be used. The HMC6352 Compass Module might 
be used in order to accomplish this task. A control algorithm will also have to made in 
order to attempt to move the robot straight. For the first step, the mobile robot will move 
forward five rotations, the servo will sweep the ultrasonic sensor, the location of the robot 
will be added to any object detected by the ultrasonic sensor, and all this information will 
be sent through the XBee.

----------------------------------------------------------------------------------------------------------------------------------

Week 7:

In order to minimize space, the circuitry done on the breadboard was soldered onto a 
prototype board. On this board, the L293D motor driver and a quad-NAND gate was 
placed. The L293D motor driver contains a quad h-bridge allowing for two motors to be run 
on the one IC. The two PWM pins were lazy chained together and router to the other side 
of the board along with the four input and output pins. The NAND gate was used in order to 
connect the three digital sensors together into one connection to the Arduino. The digital 
sensors gave a high voltage if there was not an object within 10 cm from the sensor, and a 
low voltage if there was an object within 10 cm. If the wires were to all be connected 
together into one input, without the use of a NAND gate, all the sensors would have to 
detect an object in order for there to be a low voltage going into the Arduino. Using the 
quad-NAND gate IC, De Morgan's law was used with the NAND gate in to give a high 
voltage if any one sensor saw an object, and give a low voltage if all of the sensor did not 
detect any object. The prototype board can be seen in Figure 8.


Figure 8. Prototype board containing the L293D motor driver and quad-NAND gate.

----------------------------------------------------------------------------------------------------------------------------------

Week 8:

The final design for the base has been made and 3D printed using a MarkerBot. The base 
can be seen in Figure 9. This base allowed for the prototype board, motors, photoelectric 
encoder, wheels, Arduino, digital sensors, servo, ultrasonic sensor, compass, xbee, and 
battery pack to all fit onto the 11 cm by 9 cm base. The 3D printed encoder was 
also redesigned to have four ticks per revolution to allow for more accuracy, while still not 
being computationally intensive. 


Figure 9. Final 3D printed base containing all components.

When testing the robot, all systems worked and data was able to be plotted on the first 
test. The first test consisted of having the robot elevated on the platform in order to make 
sure the mobile robot is not moving. The Matlab code was made to determine the current 
location of the robot using (x,y) coordinates and displaying these points on the graph. The 
mobile was turned by hand in order to mimic the way the robot would move on the ground. 
The data was successfully plotted as seen in Figure 10.


 Figure 10. Data displaying the (x,y) location of the mobile robot in two dimensional space.

----------------------------------------------------------------------------------------------------------------------------------

Week 9 & 10:


When testing out the functionality of the mobile robot all the components, initial tests 
proved to work. After some time, the Xbee’s encountered some hindrance and ultimately 
failed to be able to communicate information. The Xbee’s would be able to detect that 
they were on the same networks, but would not communicate. This could have been due 
to a bad connection, a change in the configuration, or a problem with the Arduino code. In 
order to ensure that this problem was not due to a bad connection, the connection to the 
Xbee was wire wrapped. After testing the Xbee again, they were still not able to 
communicate. The configurations were checked using the XCTU software along with the 
Arduino code to make sure the proper procedures were done correctly in order to send 
data. After checking these, the Xbee’s would still not communicate. Due to this problem, 
the robot was not fully functional and was not able to be optimized. The preliminary testing 
for mapping and localization did prove to work, but there is still work to be done in respect 
to finalizing the system integration.

----------------------------------------------------------------------------------------------------------------------------------

Future Work:

Ultimately, the purpose of this project was to create a fully functional two-dimensional 
autonomous mapping robot that would be useful for educational purposes. Final testing 
and optimization still needs to be performed in order to show that this can be fully 
functional. After this, a kit can be made to allow for customizable components and 
projects that the user desires. This product will be marketed for all ages ranging from 
middle and high school students, up to college students. This kit only costs $279.85 
compared to other robotics kits like Lego NXT and the Blackfin Surveyor Robot, which 
range around $530 to $560. While being at half the cost when compared to other robotics 
kit, this kit is customizable and can be used for real world applications. This will give the 
opportunity for students to become engaged with robotics, enabling them to learn and take 
on more difficult concepts in order to solve complex problems.



No comments:

Post a Comment