| title | layout | permalink | published |
|---|---|---|---|
About Us |
page |
/about |
true |
We're a team of three sophomores at Olin College of Engineering, in Needham MA. Project Memory was created for the Spring 2017 iteration of Computational Robotics, as taught by Paul Ruvolo.
In order to choose our final project, we had to hit the following requirements:
- Involves a robot, most likely a modified Neato with a Raspberry Pi and camera module. We will be expected to deploy our module on a robot by the end of the project.
- The project should involve (in part) learning about a robotics algorithm. This is not only about creating a great demo, you must structure your project so that you learn about a robotics algorithm in a project-based manner.
- The project work should primarily be computational. Some amount of electrical / mechanical work is fine, but it should be a fairly minimal portion of the project.
We formed as a team due to the fact that our intended individual learning goals for the project had significant overlap. You can read more about this on the fourth slide of our project proposal, but for a brief summary, we were very focused on learning about recurrent neural networks and creating solid documentation.
This led us to banding together to create a line following program for the Neatos, built on recurrent neural networks. This is a particularly interesting problem because although there are many line followers out there, none of them can handle sharp turns and correcting when they lose track of the line. We wanted to see if we could solve this problem by adding "memory" to our neural network via LSTM RNNs.
Lauren Gulland: I'm studying Robotics Engineering at Olin with a focus on system integration and 3D printing.
Nathan Yee: I'm a floppy orange noodle ༼ つ ◕_◕ ༽つ
Eric Miller: I'm a total nerddddddd