Purpose:
Model an octopus arm using flexible robotics

This project was my final project for COSMOS UCSD 2018. For this project, my partner, Neelay Joglekar, and I could build any robot we wanted to as long as we could build it with the parts provided and the building of the robot challenged us in some way. We decided to model an octopus arm using a robot. This pushed us because the concept of a flexible robot was new to us and forced us to think differently about its movement. In addition, we had to research the structure of an octopus arm in order to determine the design of our robot.

Overview of Octoarm:
Traditional robotic arms have rigid joints and parts. As a result, they are very useful in industrial settings. However, because they are rigid, they cannot be used close to humans. Moreover, their rigid joints limit their range of motion. Octoarm, on the other hand, is a different type of robotic arm. By using the principles of soft robotics and the structure of an octopus arm, Octoarm is flexible and can move in any direction by just using strings. As a result, it is more versatile and safe than traditional robotic arms. Octoarm has many applications in our daily lives and beyond, whether it be used as a smart camera stand or the limb of another robot.

Key Concepts:
Flexible Robotics: A branch of robotics that involves robots with actuators and parts which are flexible rather than rigid
Biologically-Inspired Robotics: A branch of robotics involving robots that are modeled off of organisms
Omnidirectional Movement: The ability to move in all directions

Hardware Components, Their Purpose, and Why They Were Used:
Hot Glue Gun Stick (backbone): They are very flexible and act like a springs to return robot to an upright position
Fishing Line (tendons): They are thin and strong and built for pulling
Laser Cut Disks (string guides): They guide the strings into position
Servos (pull tendons): They can control their angle when moving, so they are more precise than continuous rotation servos
3D-Printed Pulleys (connected to servos): They allow the servos to pull more string
Raspberry pi (computer): It controls the robot
PWM Driver (drives servos): It allows the Raspberry Pi to control up to 16 servos
PS3 Controller (controls arm): It gives more control over the movement of the arm

Software Components and Why They Were Coded:
Connection to PS3 Controller: It gives more control over the movement of the arm
Use of PWM Driver Library: Allows Raspberry Pi to control PWM Driver
Numpy.minimum and Numpy.maximum: It limits the extension of the arm

Next Steps:
Make the arm move in more complicated ways
Make the design more compact, durable, and precise
Make the arm compatible with more attachments, such as grippers or cameras
Make its movement autonomous
Make a whole octopus

Applications of this Technology and its Benefits:
Stands for security cameras: It will give the cameras the ability to see in all directions
Stands for search and rescue cameras: It will allow rescue teams to search for victims through narrow cracks in rubble
Limbs for personal robots: It will be safer than rigid robotic arms and will allow robots to move in more complex ways
Surgery: By making the design more narrow and long, it will be able to navigate through pathways in patients' bodies, reducing or eliminating the need to make incisions during surgery

The Process:
Day 1
Today, we planned out the design of the arm. At first, we designed a structure that consisted of 4 flexible tubes connected to a larger central flexible tube. There would be a string threaded through each tube which would act like a tendon. Each would be attached to a servo which would control the string's extension. The problem with the use of tubes to hold the strings is that tubes cannot contract, so if a string pulled down on a tube, the arm would form a "c" shape instead of bending in the direction of the pulled string. As a result, we switched to using a design consisting of a central flexible tube with disks placed along its length. The strings would be threaded through the disks instead of tubes. After planning out the design of the arm, we cadded the string pulleys which would pull on the strings. Because the servos could only rotate 180 degrees, we had to design a servo pulley with a large circumference so that it would still be able to pull a string a decent amount. After doing the math, we decided to make the pulleys 8 cm in diameter, which would allow the servos to pull over 10 cm of string, which we believed was good enough for a 20 cm robotic arm. We submitted the design for 3D printing and will get the servos on Day 2. The Flexible pipes will probably arrive on Day 3.

Day 2
We got the flexible piping today, but it would not stay straight nor return to its original shape after being bent. As a result, we had to find something else for the robot's backbone. We could have used a live hinge, but it would not bend as much and it would only move in one plane. After some thinking, we chose to use hot glue sticks, because they are very flexible, can be easily connected to form long backbone, and return to an upright position after being bent. In addition to this, we connected the 3D-printed pulleys to the servos and laser cut both the discs and the 4 walls of the base. Tomorrow, we will fully assemble the arm

Day 3
We began assembling the robot. However, we were not able to finish because we faced a few setbacks. We had to restart building the arm multiple times because every time we made a mistake with the hot glue, the hot glue stick backbone broke. We had trouble assembling the base because the acrylic glue would not cure fast enough and 2 of the 4 base walls broke. Tomorrow, we will finish phase 1 of our project.

Day 4
After some more drilling, laser cutting, and gluing, we finally got our robot assembled and working. At this point, it could lean in any direction. The arm could almost go horizontal. However, while moving it around, the strings sometimes slipped off of the pulleys because the strings were guided into the pulleys at a relatively shallow angle. We fixed this problem by adding laser cut pieces to the base which pushed the servos closer to the holes through which the strings were threaded, therefore sharpening the angle. We will test if this solution works tomorrow.

Day 5
Today, we focused on fixing the problem with the pulley system. Moving the servos forward did not work. After examining the movement of the servos, I realized that the strings were getting stuck on the stringy part of the 3D-printed pulley (see images). To fix this, we used the heated tip of the hot glue gun to smooth out the 3D-printed material. However, the strings were still slipping off of the pulley. Finally, we decided to drill holes right above the pulleys and thread the strings through them. This fixed the problem and the robot was finally up and running. I added in code which reset the arm with the click of a button. Then, I tried to tweak the code to get the robot to curl rather than lean. I reasoned that by making certain buttons control only one servo rather than two, therefore pulling only 1 string and keeping the opposite string stagnant, the robot would curl. However, this logic was flawed and the changes I made did not work. After thinking some more, we realized that the robot was leaning because there was too much friction between the disks and the strings. As a result, only the bottom part of the arm was bending. Tomorrow, we will fix this problem by making the bottom of the arm more rigid.

What I learned:
CAD (Computer Aided Design)
3D-printing & Laser Cutting
Drilling
Flexible robotics
Biologically-Inspired robotics