After every match, the most important thing our team does is debrief. We go over what went well, but more importantly, the big question is what can be improved. Data Analytics is our advanced off-season project that works to answer that question. We can track faults in our robot by logging data from its many sensors during matches. For example, if our robot stops working during a match, taking a quick look at the logs can help us understand which specific subsystem malfunctioned, giving actionable data that our pit crew can execute. A robot generates a lot of data, but not all of it is important in diagnosis. We went around and spoke to every one of our robot subteams asking what data was important to log. Additionally, by logging how each aspect of the robot performs during a match, respective subteams can address faults in their design. If something does not show up as working correctly, we can view these issues before they later affect the robot as a cascading error. To figure out how to
create our logging system, we did research into multiple different preexisting logging libraries, including readdressing our Log Manager application from last year. We decided to adapt from Team 581’s Dog Log and integrate it into our current Log Manager. Through spikes in data which are analyzed after the robot moves, we can figure out what went wrong and when. We have finished the advanced project and tested it on our robot looking at results using Advantage Scope.
