Engineering Failure Analysis

[mikeleslie]

A significant part of engineering is learning from failures. Many engineering standards exist as a response to a failure of some type. I would like to start a thread where teams share present specific robot failures (failure to control mentors and bad dancing by the drive team would not qualify). I would also like for those teams to share their solution or ask the community for help. If, for example your bot took a strong hit and something unexpectedly bent or broke, show some pictures, tell us about it but be specific. The legitimacy of the hit does not matter (in this thread), but the response of the system does. I believe that we could all learn a lot from the "good work" of others. Remember: if it never breaks, it must not be close enough to the edge.
I know that much of CD kind of get's to this, so what I'm asking if for a more structured "failure analysis" of the problem, and ultimately a "root cause analysis" I think this would be great all around, as not only do we all get to learn from them, but we get to teach a real world process at the same time.

So here’s some simple rules to follow:

• Quantify it. Don’t say “fast” think about roughly how fast
• Specify the type of material if you can. (6061 and not just aluminum)
• Include Pictures or drawings of the failed part, and in particular a close-up of the actual failure, like in CSI
• Don’t be judgmental. It doesn’t matter whether the other guy was going too fast, or they were already yellow carded. Focus on the failure.

The goals of this thread would be to begin to develop some standards and practices for durability built from real world examples so that teams can have a common point of reference for “I built it like this”.

I think as a community we could all benefit from this type of discussion.

Mike Leslie, Lead Build Mentor 1189