The specific aim of the senior design project is to design a robot consisting of a cart and a double pendulum to mimic a scenario in which a human need to balance him/herself on a bus or a train. The robot has the classical double pendulum configuration that you have probably seen in your dynamics or control classes. But now sensors and actuators will be added to the joints so that we can test different feedback control law to the robotic system thus gain more insights into human balance. (MAE UC Davis, Jason Moore)
I am going to do my best to document my progress on the Senior Design project as much as possible since I realize that many of my project entries are very minimal. I will be honest and say that I was not initially super excited about the double pendulum project mainly because I was one of the two students who were assigned a project that they did not select. However, after talking with Professor Moore, I am quite intrigued by its potential and how much I will learn about control systems. It should be noted that I am currently taking EME 172 Control Systems so my understanding of controllers is very minimal.
Today we met with our two sponsors Professor Kong and Dr. Moore to understand the scope of the project. We have established that there are two components to this project.
The 2nd requirement is the easier of the two and simply requires that we dumb down the interface so that high school students can play with different parameters and watch the robot attempt to balance itself during perturbations. The 1st requirement is the "fun" one where we will be testing different controller equations and seeing if these controllers will work in keeping the robot balanced.
You may be aware of the double pendulum on a cart problem where the cart can swing up and balance the inverted double pendulum. https://www.youtube.com/watch?v=B6vr1x6KDaY. Ours will be different because instead of balancing our inverted pendulum with a cart we will be using the cart to perturb the pendulum and use the actuators at the pivot points to apply appropriate torques to hopefully keep the pendulum from falling. These actuators act like muscles in the ankles and hips of a human body. In theory models derived from perturbing humans should work on our model. (We'll see if that actually works.)
One of my teammates is having trouble understand exactly what kind of robot we are building. In the previous post I described that we would be making a inverted double pendulum on some sort of a cart that will balance using actuators. I quickly drew this diagram up so that he could understand the scope of the project better. We simply shake the cart and based on the control law the motors will apply torques at the hinges in red to keep the robot inverted (standing).
We are now in the concept generation phase of the project. One of the major problems I am having with this course is that we are trying to design and develop like we would industry. The only difference is that we are given about 1 week for what might take months to do properly especially for a new project. This phase seems fairly rushed to me but I think we have a pretty good idea on some of the concepts we are going to try, Here are some sketches of the carts that the pendulum can go on. The first is an RC type car that we can control wirelessly. The second two are typical linear drive tracks with either a lead screw or timing belt.
Here is the first report that our team submitted.
After spending 28 straight hours in the mechanical engineering computer lab (CAE) the preliminary project report is complete. I realize there is no conclusion or introduction which is bad and hopefully it doesn't hurt our grade too much I am very pleased with how the CAD model and system as a whole turned out. I was in charge of the whole pendulum design.
Two posts in 1 hour!? Yes, I should probably be studying control systems but I figure I'd make another post while I'm still logged in. I just wanted to share some images of our double pendulum arm. It uses 2 geared motors that can put out about 2 Nm of torque. The two grey cylinders at the top are rotary encoders that will track the angle of the motors. The new pendulum can also swing 360 degrees unlike the previous model. This may not be needed for our robot since it is suppose to be able to represent human balance and an upside down human isn't really all that useful for this type of research. I think it will be fun for the high school students so that they can see a swing up demonstration. I also attached the entire assembly with the cart Stanley Tsang made. I do think that our design looks pretty cool especially with all of the belts and pulleys.
We designed a poster for the senior design showcase so that people that come to our project can read some of the background of the project. It was very last minute and stressful getting the solidworks models to render properly.
This is just a video of me testing the PID controller of the double pendulum robot. Its unfortunately the only working video I have of the project. The controller was implemented only on the lower joint. When I push on the pendulum arm the springiness is the result of the actuators (DC motors) engaging using a PID control loop. This was implemented on the Arduino Mega.
Contains an Arduino class for a double pendulum robot.