RoboSub 2016 Recap

entire-team-and-sub-img_0219

 

2016 Recap:

Looking back on our first year or RoboSub, there’s a lot to talk about. The entire team is incredibly proud of everything we have accomplished. We learned an incredible amount, not only about the technical aspect of designing an AUV, but also about the far more applicable skills of creatively solving a problem, troubleshooting that solution, and managing time and resources over a large-scale, year-long project. I think the best thing that I could start with to sum up our experience is this graph of the Dunning-Kruger effect:

 

screenshot_799

Ultimately, this graph came to represent the most valuable lesson we learned. The best thing about this project is how incredibly over our heads we were when we started. RoboSub is by far the most complicated project any of us have worked on. On that graph of the Dunning-Kruger effect, we very quickly climbed to that area of baseless confidence. We put together a schedule which had us finishing with plenty of time to spare, and almost immediately fell behind. Everything which we expected to be easy was not. Every piece of software took longer to write than we expected, and designing took far more iterations than we expected. We thought that we accounted for this in our schedule – we certainly did not. Needless to say, it did not take us long to fall down the steep slope of the Dunning-Kruger effect.

Another one of the most valuable lessons we learned was about effective communication. We had one of the smallest teams in RoboSub; however, we still had different people working on different pieces of software and hardware at the same time. Everything had to work together nicely at the end of the day. It took us a long time to get that right. Actually, we didn’t really ever get that right. Throughout the course of the year, we got much better at communicating effectively what we were working on, discussing how to set everything up for seamless integration, and spending the time to make sure that we weren’t repeating each other’s work. However, we still broke our code repository (http://github.com/beaverauv) at the competition while we were in the process of putting the robot in the water. It took us a day to fix it; however, it is safe to say that we have learned from that experience and we have since completely overhauled our revision management.

The last thing that needs to be briefly mentioned is how much technical knowledge we gained. We went from having worked in some basic programming (Arduino, C/C++, a bit of Java, etc.), to working with professional tools (mainly ROS) in a complex system. We learned how to manage code that runs off dozens of files, not just one or two. We learned how to design a robot that can be manufactured within a budget and with machines on hand. I think we are all very proud of what we learned last year, and are excited to see what we learn going forwards into the 2017 competition.

2017 Preview:

We are just starting to work towards the 2017 competition. Here’s the main list of improvements we are making to this year’s iteration:

  • Add front-facing lights and make everything salt-water compatible so that Prospero can be used as a research vessel as well as a competition vessel.
  • Implement the sonar system to interact with the acoustic pingers, including a custom-designed signal processing PCB (printed circuit board).
  • Use a convolutional neural network (a type of machine learning) to complete far more advanced vision tracking.
  • Redesign Prospero to be more modular and functional by adding a center console in the middle of the hull, and removable external battery enclosures.
  • Add a pneumatic system to operate a torpedo and dropper system

We are very excited to work towards these goals and more in the 2016-2017 school year.

October 27th, 2016

Posted In: Uncategorized

Leave a Reply

Your email address will not be published. Required fields are marked *