View Single Post
  #28   Spotlight this post!  
Unread 02-01-2012, 17:08
connor.worley's Avatar
connor.worley connor.worley is offline
Registered User
FRC #0973 (Greybots)
Team Role: Mentor
 
Join Date: Mar 2011
Rookie Year: 2010
Location: Berkeley/San Diego
Posts: 601
connor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond reputeconnor.worley has a reputation beyond repute
Re: Introducing: CowScout

Quote:
Originally Posted by Grim Tuesday View Post
This is very cool. A couple questions:

How is CCWN calculated? Our team has been trying to figure out how to do something like this for a while, but it never got off the ground. It's good to hear that someone else has done it, but it would be quite interesting to see the math that goes into it. If there is a white paper for this, or OPR calculation available somewhere, I would be much obliged if someone pointed me towards it.
OPR and CCWM are calculated using the exact same algorithm. Honestly, I don't fully understand the math behind it; I actually ported some MATLAB code to PHP. Here are some very useful links:
http://www.chiefdelphi.com/forums/sh...0&postcount=19
http://www.chiefdelphi.com/media/papers/download/2321
https://ece.uwaterloo.ca/~dwharder/NumericalAnalysis/04LinearAlgebra/cholesky/#matlab

The difference between CCWM and OPR is that CCWM estimates a contribution to winning margin, rather than score (just feed it different data).

Quote:
We have a dynamically generated graph of the robots performance over time, which is useful, especially in week one regional to see if, maybe, a robot had a problem at the beginning of the competition but it got fixed, and might be a good second pick. Though your system includes this numerically, graphs are a great way to visualize it.
While we do have graphs for specific statistics, we don't have any sort of graph for overall robot preformance. We'll see if we can implement that.

Quote:
A section for comments, either on the way the robot plays, or just comments in general. We've also had sections on our database listing various stats about the robot: Drive train (a stat) and drive strength, as measured qualitatively by our scouters. If such a feature exists and I don't see it, please point me towards it.
We considered comments, but anticipated that too many would be useless (i.e. "this robot is awesome"). Our team isn't sold on pit scouting, and for now we're going to leave it up to teams to do that on their own. Thanks for the feedback!
__________________
Team 973 (2016-???)
Team 5499 (2015-2016)
Team 254 (2014-2015)

Team 1538 (2011-2014)
2014 Driver (25W 17L 1T)
日本語でOK

Last edited by connor.worley : 02-01-2012 at 17:11.