Final Progress Report
As the 16-264 Instructor, Chris Atkeson is the creator of Baymax, we wanted to provide him with a robot that could imitate Baymax's waddling. We had the whole semester to build it so we spent a lot of time on thinking about how to approach it and modeling on computer software.
We did not have that much time to fix stability issues with the robot but we have built a strong robot that others can use. We want people to read this website and continue or build similarly to carry on the mission of making Baymax waddle.
We have created this documentation to help you guys understand what tools we have used and what problems we came across so that it can be much easier for you.
Team: Deniz Sokullu, Taichi Akiyama, Bobby Li, Max Stein, Priscilla Jenq
Design Process
Demonstration of the MIT Toddler Project:
https://www.youtube.com/watch?v=goqWX7bC-ZY
The Article:
http://groups.csail.mit.edu/robotics-center/public_papers/Tedrake05.pdf
https://www.youtube.com/watch?v=goqWX7bC-ZY
The Article:
http://groups.csail.mit.edu/robotics-center/public_papers/Tedrake05.pdf
Mit Toddler
Flaws in the First Model
The first prototype we made did not go as planned for many reasons. We decided to use MIT's toddler project as an inspiration so make Baymax waddle. The first problem we came across was that we thought we had no access to machine shops and we couldnt create curve feet for the robot we had in mind. We created slices of the feet and attached them together with screws but it didn't work well.
We also planned to make the parts without realizing where we could place the servos that would move the legs. The robot we built requires 4 servos to tilt on the x,y axis. After we realized that this version of the robot was too flawed to build on, we decided to start with a large scale robot that could actually fit inside a baymax suit.
Planning ahead for the Final Model
For the final model, we found access to a machine shop and created perfect feet for the robot. And to make sure that the curvature and the size of the feet were correct, Bobby emailed the MIT professor that created this robot to ask for his Matlab simulation that simulated the robots movement given certain parameters about the feet.
Also, we started by including everything, even the servos and all the off the shelf parts in our Solidworks model. This way we got rid of most of the problems that could happen by human errors and measurement. We also worked around the servos to ensure that their placement would be optimal and checked the center of gravity to make sure that the robot would not tip over.
We seperated the team into people who worked on the software and 3D design and people who spent time printing, cutting crafting the model. We made sure that there was a constant back-and-forth communication to ensure there were no misunderstandings.
The robot consisted of the feet, legs and the electronics/battery box. We ordered off the shelf parts to attach 2 servos together on each leg, and used some large screws to lock them in place. The legs were made out of plywood but the connection to the axle was 3D printed to minimize damage on the plywood.
The electronics box held a Arbotix-M Robocontroller and a 12V battery box, that could power the robot while mobile. Also, the board has a cool Big Hero 6 logo which is the key point of this project.
Screenshots from Solidworks
Actual final built model
Problems we faced
After assembling the model, we had a hard time getting the servos to respond to us. We used a roboplus to talk to the servos but whenever we uploaded our code via Arduino, we had a hard time controlling the leg. But if torque is given, the legs successfully moved, example in the video below.
We want the upcoming team to solve this problem and continue to find solutions to future problems such as:
- Placing potentiometers to the legs to understand the balance of the robot. If this value can be recorded, the robot will have an easy time not falling down.
- Communicating with the servos. We thought everything was going fine until this happened. We got many errors and did not know what to do. We even used new Robocontroller Boards to test movement but either something in our code or the servos was wrong and we had a hard time moving forward after this point.
Here is a link of everything we bought:
Here is the link to the Solidworks Assembly of the robot:
We will return the robot to Chris Atkeson so it will be available for the next group to pick up.
Hope the next team carries this forward,
Team Baymax Walking
No comments:
Post a Comment