Name: ______________________________ EE411 Lesson 3: Concept Generation & Evaluation This lesson assumes that you have completed KPR 3. Concept Generation Enhancing Creativity o Certain types of puzzles are known to enhance creativity. http://www.thecourse.us/students/lateral_thinking.htm o Allow time for brainstorming o Use your background research and apply SCAMPER (Substitute Combine Adapt Modify Putto-other-use Eliminate Rearrange) o Make curiosity a habit. The more you know about other things, the more fodder you have for new ideas. Team Brainstorming Methods o 5 Basic Rules for Brainstorming Suspend criticism or judgment Wild ideas are encouraged Quantity over quality Build upon and modify the ideas of others All ideas are recorded o The Nominal Group Technique (NGT) 1. The leader/facilitator should read the problem statement out loud. 2. Each member should restate the problem in his/her own words and the group should discuss any differences. This is to make sure that everyone in the group fully understands and agrees on the problem. 3. As individuals, silently generate and record ideas. 4. Collect ideas in a round-robin fashion. Each person presents one idea at a time until all ideas are exhausted. The facilitator should clarify ideas and record them where all can view them. 5. The facilitator then leads a discussion to clarify and rephrase ideas. Similar ideas should be grouped or combined. 6. The group should then prioritize ideas to explore, either by vote or consensus. The outcome is a set of prioritized ideas to further discuss and pursue. Page 1 of 7 A common brainstorming technique for engineering systems is to break down the system into its functions or characteristics, and then to brainstorm ways to achieve each function or characteristic. There are 2 tools particularly mentioned by the text that you might find useful for your project: Concept tables— The column headings represent functions or characteristics of the system, and the items beneath them are options for achieving that characteristic or function. Here’s the example from the KPR again: Pet Incentive Ball Squeaker Treat Mechanical rabbit Exercise Mechanism Swim current Treadmill Exercise wheel Pulling weight Ball Chase Track Power AC Power Battery Dog powered Solar power Operation Location Indoors Outdoors Portable Note that rows have no correlation (the mechanical rabbit isn’t particularly associated with “pulling weight” or “solar power”). After discussion, the group would then circle the idea(s) to try to combine in a design to represent options for further consideration. For example, this might represent one option: Pet Incentive Ball Squeaker Treat Mechanical rabbit Power Operation Location AC Power Battery Dog powered Solar power Indoors Outdoors Portable Exercise Mechanism Power Operation Location Swim current Treadmill Exercise wheel Pulling weight Ball Chase Track AC Power Battery Dog powered Solar power Indoors Outdoors Portable Exercise Mechanism Swim current Treadmill Exercise wheel Pulling weight Ball Chase Track And this would represent another: Pet Incentive Ball Squeaker Treat Mechanical rabbit Concept Fans or Function-Means Trees A related tool is the “concept fan,” (aka “function-means tree”) which is particularly good for decomposing systems with hierarchical design decisions. In these charts, circles represent functions or characteristics of the system, and rectangles represent means. (This is slightly different from what our textbook says—the truth is that the shape isn’t critical--- whether it’s a function of a means is usually clear from context). Here is the start for a concept fan for a smart phone biometric ID app: Page 2 of 7 Droid Platform iPhone Near IR Spectrum Visible Iris Take Picture RED Algorithm Biometric App Neural Network Biometric Capacitance Read print Optical Fingerprint Image correlation Voice User Interface Matching algorithm Phase matching Keyboard Minutiae matching Don’t marry your first design idea. Keep an open mind and try to come up with several different ideas before you move on to the concept evaluation stage of design. Page 3 of 7 Concept Evaluation Every design decision point needs to be justified. Initial evaluation o Don’t rush to prune the design space. o Discard only those concepts that are completely unfeasible. Strengths and weaknesses analysis o A simple chart of strengths and weaknesses between approaches o Useful when comparing apples and oranges o Example: NIR vs. visible spectrum for iris recognition: Spectrum NIR Strengths Weaknesses Potentially covert Can’t use phone camera Could be done in the dark with a peripheral light source Can use phone’s camera Dependent on ambient lighting or built-in flash Better established— freeware more readily available Visible Decision matrices o A more analytical approach—particularly appropriate for choosing between similar options. o Can be applied at the “micro” scale or the “macro” scale o To establish scoring system, need to consider the performance sweet spot: Point of diminishing returns. Set to 100 Benefit Minimum for consideration. Set to 0 on metric. Quality Page 4 of 7 o Example: battery choice for a small robot Requires: >5 V & ~100 mA Objectives: low cost, low weight, long life Relevant engineering requirements: “Total system should weigh less than 1 kg” “Total system should cost less than $200” “Battery life should be > 5 h” Statistics for the options (assuming rechargeable batteries): Battery Voltage Capacity/ Weight/ Cost/ Necessary Tot. battery battery battery Config. Weight AAA 1.5 V 850 mAh 13 g $2.75 4 batteries in series AA 1.5 V 2500 mAh 30 g $4.40 4 batteries in series C 1.5 V 4500 mAh 50 g $6.79 4 batteries in series D 1.5 V 9000 mAh 80 g $11.79 4 batteries in series 9V 9V 250 mAh 42 g $10.88 2 batteries in parallel* *2 in parallel needed to make minimal lifetime from requirements Tot. Cost Tot. Life 1) Determining weightings for objectives Light Low Cost Long Life Light 1 Low Cost 1/5 1 Long Life 1/3 3 1 2) Establish metrics for each objective Light? Weight Score Range >500 g 0 <50 g Would be half of total system allotted weight. Even though we expect the batteries to be a significant part of the system weight, we still need to leave room for other components. Would be < 5% of total system allotted weight--- negligible to total weight Interpolates between min and max scores with straight line. 100 AAA weight Justification 50-500g Option “Light” Score avg AA C Page 5 of 7 D 9V Long Life? Could use similar model but upper limit harder to define since battery lifetime directly determines system lifetime. Alternate model (particularly if there is no upper limit to a desired quantity)—score options relative to “Best in Class” However, this also assumes that you’ve included all of the realistic options. Option Lifetime Lifetime/Best Lifetime AAA AA C D 9V Low cost? A similar system is used for rating cost relative to best in class, except it needs to be flipped so that low cost = high score. Option Cost Best Cost/Cost AAA AA C D 9V 3) Score the options. AAA The Decision Matrix AA C D 9V Light Low Cost Long Life Score 4) Review ___ is the winner—largely based on cost and despite relatively short lifetime. Does the result feel right? Cost played a big role, such that even though long life was prioritized above weight, one of the shortest lifetime options was chosen. Is that okay with the customer? Are there other options that should have been considered? For example, would there be a benefit to looking at adding more batteries in parallel to boost lifetime (at expense of cost and weight)? Page 6 of 7 Pugh concept selection A more qualitative method for comparing concepts—better for apples & oranges than the decision matrix. 1. 2. 3. 4. 5. Generate lots of concepts. Choose the one that looks best to the group—set as baseline. Rank other concepts against it (+1 = better, 0 = equal, -1 = worse) Compute weighted scores for other concepts. Use results to generate a new concept set and repeat until the superior concept is clear Example: pet exercise eq. Baseline: Existing treadmill modification with sidewalls for safety and treat dispenser Option 2: Ball shooter that dispenses treat upon ball return Option 3: Outdoor track with electric, erratically moving, squeaking rabbit Option 4: Doggie TRX with squeaking lure and treat reward Great Dane appeal Capability for autonomy Canine energy dissipation Cost Space it takes up Score 3 2 2 3 2 Baseline Concept --- Option 2 ---- Option 3 Option 4 0 +1 0 +1 +1 +1 -1 +1 -1 +1 +1 -1 -1 +1 +1 Based on these results, a new set of options would be considered: Is there a way to make the track more consumer friendly? What could we do with the Doggie TRX and ball shooter options to increase dane appeal or energy dissipation? Is there a way to make the treadmill autonomous New Baseline: TRX with erratically moving, squeaking rabbit lure & treat dispenser Option 2: Ball shooter that launches squeaking rabbit Option 3: Squeaking rabbit robot that runs around house—track idea without the track. Option 4: Treadmill that can sense dog’s presence and position, and autostart and stop Then you would do a new chart for this and continue until a clear idea emerges… Page 7 of 7