Pre-Kickoff Decision Matrix

Hello CD,
(Direct questions bolded so you dont need to read a wall of text if you dont want to)

My team is trying to be more organized and on time this year, as well as still teach the importance of the design process as it is taught in our school’s PLTW classes. Last year, our team made a last minute decision matrix type document that we went through during our Kickoff meeting. We used that to help decide on the size of our robot, drivetrain, as well as wheel type/size, etc. I was wondering if any other team has created a decision matrix to help with these types of decisions. If so, what do you usually include? I want to try to have this document set up and ready to go before kickoff. I currently have included criteria such as cost, time needed to build/program, team experience, etc. If anyone has any suggestions, or has done this before, id love to hear them.

Yes i have searched CD for this before posting this. I did not find much, and what i did find wasnt particularly helpful. If you have a thread that i may not have seen, feel free to send it.

My PMs are always open, so feel free to contact me that way as well.

Kaitlyn

2 Likes

We used one last year to decide what we wanted to do in terms of scoring and scoring location, but I don’t remember it too well, so I’ll use an example from one of our off-season projects this year. It’s for drivetrain specifically but should work well enough.

We started by listing criteria/aspects, and ranked them according to what everyone thought:

image

After this, we weighted everything with respect to importance:

image

Which gave us our eventual winner.

In terms of application for a competition robot, a more overarching matrix would probably be a better idea. For a pre-kickoff one ranking drivetrain, cost, auto capabilities (not exactly what it will do but how much time you’ll want to spend, for example), etc is more important since you won’t know how scoring or rps will work. Weighting is also really important, especially for larger decisions.

hopefully this was somewhat helpful
2 Likes

Here’s a weighted objectives table in Google Sheets. You can enter your own weighting factors in the box to the left, and the designs to evaluate in the boxes to the right. We haven’t used this in 2018 or 2019, but the categories in there are the ones that we last used.

1 Like

Good stuff here. Another thing to consider is risk, which will count as negative points towards each design option. For each subsystem:

  • What things can go wrong? Make a list and answer the next two questions for each:
  • How likely is it that this thing will go wrong?
  • How damaging is it if this thing goes wrong?
3 Likes

As far as a drivetrain selection matrix goes, do you see many teams change their drivetrain year after year? Do you see many elite teams change their drivetrain every year? How many teams can build top performing tank, mechanum and swerve drives?

Shouldn’t your drivetrain be the one part of your robot that you have the least concerns about?

Should you even use a decision matrix to pick a drivetrain?

Build a drivetrain that you know how to build well and stick with it.

1 Like

We still go over everything for the teaching part. We usually know we are going to do Tank. But, we did just get new swerve modulas, and have been practicing. So based on this year’s game, swerve is an option for us. Either way, it’s necessary that we teach the new kids the process that must be gone through in order to find the best and most efficient robot, and use of our time. This isn’t purely deciding on robot factors.

To keep it simple and short on time we’re doing a S.W.O.T. Analysis. (Strength, Weaknesses, Opportunities and Threats) this allows for people who come up with ideas to present their idea the way they see it and keep it simple as well as fast.

3 Likes

Be careful. I’ve seen some very poor decisions made because a group agreed upon weights for a weighted objective table and then disagreed with the outcome. If that’s the case, you run into a few scenarios (1) making a bad decision because the “data” told you to (2) manipulating the “data” to serve your purpose (3) ignoring the results of the table because you disagree. Those are all bad outcomes.

The best decisions are ones which are made based on a discussion, leading to something close to a consensus. Definitely do consider all of those factors you would in a table (speed, time, whatever it is that fits the decision being made). And you can even record them in a table. But assigning numbers to things is tough and can yield bad results.

Some math-y engineers may disagree with me, but I’ve seen it fail so many times. The SWOT as previously mentioned is a great mechanism for creating a meaningful conversation!

1 Like

I remember a class I took at work that asked several teams to come up with criteria in order to choose a mouse disposal device. Some of the devices were:

The old fashioned spring trap.
A live trap.
A trap that was similar to the spring trap but you didn’t have to “see” the results.
A trap that could kill multiple mice.
Poison.
A cat.

The class got 2-3 different “winners”. It was because of the criteria each team decided on and the weights that was assigned to each criteria.

There was only one time where all the teams came up with the same winner. They were all from the same department and the all had identical criteria.

1 Like

Yeah this is the worst especially when the ideas that aren’t the greatest are pushed up front because the way it’s perceived. This is what led to some bad designs on our robot for 2019.

Perhaps the questions being asked should be “What roles can our team play using the drivetrain type that we are proficient at building?” and “What scoring methods and mechanisms work well with the drivetrain type that we are proficient at building?”

One of the advantages of decision making methods such as a decision matrix is to replace emotion, preconceptions and wishful thinking with objectivity to drive the decision making process. They require the people making the decisions to be willing to let go of their emotion, preconceptions and wishful thinking in order for such methods to work.

I have seen several groups of (objective oriented) middle school students use decision matrices very effectively in FLL.

I agree. SWOT analysis is a much easier way to organize and discuss options. I’m not a fan of decision matrix approach - too much bias in the numbers. I have seen a decision matrix used at work to evaluate options that was not simply driven by low cost. But even then, the pro / con discussion was more important than a mathematical outcome.

It’s been a while since I’ve seen the PLTW design process, but I recall it had the fundamental steps. But you also need to consider an FRC robot is more of a prototype than a complete engineering design. Design iteration doesn’t end.

The “on time” part sounds like your key consideration.** Part of our discussion includes the following construction questions:

  • Have we built and used this type of design before?
  • Can we design and build it quickly? (Some have described this as “fail early.”)
  • Are there COTS options?

My bottom line recommendation is to use what you know and add something new. You just don’t want a majority of the robot to be new stuff during the build season.

David

** I can’t remember if PLTW had those engineering bummers… it needs to be on time and on budget.

1 Like

We do something like @BordomBeThyName posted on kickoff. Based it off of 148’s resources: https://www.robowranglers148.com/resources.html

We ran into problems doing a drivetrain pick after kickoff a couple of years ago when west coast drive (6 wheel) and mecanum ended up with identical weighted scores… just with different strengths. So we ended up doing a vote between the two set ups.

An additional practice that we started doing a couple of years ago was to make a “matrix” of sorts for scoring in the game: points per action, estimated time to complete, and difficulty level. We use the points per second, and the difficulty level to build a sequence of events we think we can accomplish in 90s (accounts for lost time to defense)
image
Sometimes our time estimates are WAY off.

5 Likes

I absolutely agree that the decision can’t be left just up to the numbers, but assigning weights beforehand will quickly help you identify which options should go into the final consideration. I’ve found that there are many times where the second or third numerically ranked options are the right decision - and not just in robot design, or even design, but also (perhaps even more so) in hiring decisions. Once you’re down to a few candidates, SWOT is effective and efficient. If you have a dozen or more candidates and a dozen or more people in on the decision, having something numeric to thin the herd is a good thing. One of the things I picked up from my office’s hiring process is that once you have your numeric ranking, look for “natural breaks” in the data. If you’ve decided up front that you were going to thin to two options, but then #2 and #3 are separated by one point and #4 is five points behind them, add #3 to your final consideration.

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.