25-11-2008 23:05

JVN

Re: paper: Using a WOT for Competition Robot Design

Hi All,
Just a quick whitepaper I wrote up on using a Weighted Objectives Table.

This paper is a response to this request:
http://www.chiefdelphi.com/forums/sh...6&postcount=37

It may be an interesting read for designers. This is based on my experiences on 20, 229, & 148 and the things I learned during my 4 years at Clarkson. Hopefully it can provide another tool to those looking to bring some more formal engineering decision making into their build season.

-John

25-11-2008 23:22

Brandon Holley

Re: paper: Using a WOT for Competition Robot Design

Thanks JVN

26-11-2008 02:16

=Martin=Taylor=

Re: paper: Using a WOT for Competition Robot Design

Okey, so I've had this crazy idea for a long time, and have always wondered if it were possible.

Perhaps some of you 1337 engineers out there can ponder over this...

So we get the same battery every year right?

Doesn't this mean we essentially get X amount of energy in the kit?

Most FIRST games involved moving mass from one place to another (Doing work on the game pieces).

So wouldn't it be possible to calculate the max possible points you could score with the battery?

Of course this will be an astronomical number, that doesn't account for all those annoying "real world limitations." Then you just need to look at statistics of "top robots" and gauge their efficiency to the max possible. This will give a scale factor that could be used to figure out the max possible points achievable in any given game.

Design your robot to score X points = Winnage.

If only the WOT could be used to design your robot for you…. Hmmmm…

26-11-2008 09:54

kramarczyk

Re: paper: Using a WOT for Competition Robot Design

That is the intent stated in section 8.3 of the 2008 rules.
The variation is present in starting conditions...
Did you start with your clippard tanks charged?
Do you have a mass up high that you can lower to do work for some functionality?
Did you start with your shooter springs compressed?

Looking at the energy required to complete a task is a good way to estimate it's feasibility. These two concepts go a long way in this area.
Work Energy = force * distance... i.e. to lift something = Mass * Gravity * Height (example)
Kinetic Energy = 0.5 * Mass * Velocity^2 (example)

26-11-2008 09:57

kramarczyk

Re: paper: Using a WOT for Competition Robot Design

John,

Have you ever looked into using a Pairwise comparison to help along the criteria weighting?

26-11-2008 14:02

rutzman

Re: paper: Using a WOT for Competition Robot Design

Thanks, John. I think this will be very useful for us(2614) once build season hits.

26-11-2008 14:59

billbo911

Re: paper: Using a WOT for Competition Robot Design

Now that I've had a chance to read through this paper, I must say, I'm impressed. It is easily understandable and straight forward to implement. It allows a quantification of the design process. One of the best things I like about how you spelled out the process is that you also showed how easy it was to modify it to fit the particular circumstances of almost any design situation!!
Thanks again!

29-11-2008 10:44

gblake

Re: paper: Using a WOT for Competition Robot Design

One note of caution:

Applying an analysis method based on the assumption that linearly combining parameters yields a correct result is appropriate in many situations, but not all.

Some parts of designing a FIRST machine often also involve making binary (or N-ary) decisions, or are best described non-linearly. When these decisions and their non-linear effects on the design's merits are present, they are often poorly expressed in a WOT approach. This often shows up when the merits of different combinations of items are assessed (at all levels of abstraction). Combining the "best" claw" with the "best drive train" might yield a mediocre combination when compared to lower scoring blends (in a WOT score sheet) that are able to achieve synergies not encoded into the WOT's weights.

In short, the WOT is a good analysis tool; and is one that you should always use... with your eyes "open". It will help you develop a good design; but it is not the only approach you should use, nor is it guaranteed to determine a best design. Leave room in your conversations for alternative methods.

Thanks JVN, for demystifying this method!

Blake

29-11-2008 12:29

JVN

Re: paper: Using a WOT for Competition Robot Design

My college roommate and mentor extraordinaire adds the following comment:

"If you WANT to fudge your results, the WOT steps also work in reverse! Step 1 - Decide which design YOU want to win..."

29-11-2008 12:47

Daniel_LaFleur

Re: paper: Using a WOT for Competition Robot Design

Your college roomate was wise.

We use WOT not only in the design phase but in our scouting during the regional as well.

By understanding the capabilities of our own robot, we can determine the capabilities needed from our teammate robots to further succeed, and we can apply that weighted scale to our scouting data.

29-11-2008 14:58

Tom Bottiglieri

Re: paper: Using a WOT for Competition Robot Design

For PLTW students: this is in your POE curriculum as a 'Design Matrix'. As stated above, its a good tool, but must be looked at from a totally objective point of view.

30-11-2008 01:11

Raul

Re: paper: Using a WOT for Competition Robot Design

John,

You continue to impress me with how helpful you are to the FIRST community. This is a good addition to the library of white papers that is sure to help many new teams.

As an engineer at Motorola, we were trained to use this same method. We call it the Pugh Methodology, which uses a Pugh decision matirx or chart. You can google "Pugh Matrix" for much valuable information.

We use this methodolgy every year to help us rationalize if to use Crab drive or not, along with some other design decisions. I use the word "rationalize" because you have to be careful with how you weigh the different criteria. I have found that those numbers have to be tweaked carefully based on experience. You also have to be careful because sometimes you leave out an important criteria that could make a big difference in the effectiveness of your design. And as you said, you have to be careful about fudging the weights

The other thing we found is that some criteria can be broken down further into subcriteria. Using your gripper example - "Speed of Grab" could be broken down into "from floor", "from human" and "from the rack".

Another thing we found is that you need to score a very specific design and not just general concepts. Again using your example - not all roller claw designs would be scored the same. The beauty of this methodology is that even once a design concept is chosen, it helps the team decide where to make design tradeoffs. Again in your example, you should be willing to add more cost to the roller claw to make it easier for the driver and reduce the precision required. In affect, your detail design goals should be to improve the total score on the design by making the right design tradeoffs.