[B.Johnston]

This is going to be a long post so please bear with me.

I agree with you and adam above that there is no substitute for good design.

I believe that good design includes the analysis of the functional requirements of the task you are performing.

This includes appropriately matching component properties in a system.

I think this is where my emphasis on fault tolerance and redundancy is being misunderstood.

For me the meaning of fault tolerance is the ability to recover (perhaps not fully) from an unanticipated occurrance.

For example if you have a tank track system and your tread is stripped off your robot because opponents have discovered that when you are pushed sideways they can be broken.

A fault tolerant design would, perhaps allow for you to drive with no track on(rather than being immobilized because your drive is sitting on a skid plate or bogies).

An example of redundancy does not neccessarily include the use of a second system which fully duplicates the function of the primary system either.

An arm mechanism driven by 2 #25 chains operating in parallel to support the functionality needed rather than a single heavier #35 chain would provide a redundant system capable of withstanding the failure of one chain and sprocket system (accidental entanglement with another robot) if done properly.

Here full function may not be possible but some function would remain.

Neither would be an example of poor design, they are both designed to mitigate the outcome of failures that while not wholly unexpected are somewhat probable.

To design a system so "strong" that it never breaks is not inherently good design.

If we apply enough force to a wheel we can break traction and start it spinning.

This force needed would be our design target (plus a moderate margin) for the strength of the components in the system.

To use components which far exceed this constraint is not good design.

You would carry a combination of penalties in weight, cost, bulk, parasitic losses etc that while nothing broke, something was affected along the way.

A function that couldn't be included, a speed that couldn't be reached, an incline that couldn't be climbed etc.

While I don't advocate either redundacy or fault tolerance as a substitute for good design they can be used as part of a strategy which maximizes the outcome despite the circumstances.

B.T.W. The 2006 sprocket that fractured was a 19 tooth #35 martin sprocket capable of absorbing about twice the power that our double 2.5 cim/andymark 2 speed drive could deliver. It fractured at an inclusion near the keyway due to an internal flaw in its structure. The other sister drive trains were good enough to win three regionals that year.