Evolution of Design

Define

The suburban area of Linden is largely cut off from the rest of Columbus due to I-79. This denies Linden residents easy access to centers of employment and basic healthcare and welfare services. We needed to create a safe, cheap, and fast transportation system to connect Linden to Easton and Polaris so that the standard of living for Linden residents can increase. The transportation system had to be an autonomous, electric-powered AEV.

---

Represent

There are many criteria associated with this project, especially the route that must be taken.

Above is the route that the AEV must traverse, labelled with the distances between stops. There is also a set amount of actions/stops the AEV must complete.

Above is the same route, but with the necessary actions of the AEV shown. It must stop a total of four times, and go to the station as well as go back to where it started from.

---

Plan

Iteration 1: Prototypes

Design #1

Desc – Plan designed for simplicity, and to test the efficiency of horizontally orienting the battery and board behind the motors as opposed to vertically.

Pros/Cons – Design 1 is simple and straight-forward and light-weight. On the other hand, the combination of propellers and uneven weight distribution towards the front of the AEV harms maneuverability.

Design #2

Desc – Plan designed to test the feasibility of a belt driven AEV. It would be easier to stop and start the AEV using this design.

Pros/Cons – This plan was designed to be more complex but improve on every other aspect of the reference AEV. The design is appealing, light-weight, efficient, and more easily controllable due to belt-based movement.

Design #3

Desc – Designed for aerodynamic efficiency, it will be powered by twin propellers on one end of the AEV.

Pros/Cons – Design 3 was more appealing to the eye and rather simple; however, even though the weight was more evenly distributed, the design was more clunky than the other designs and therefore spatial efficiency suffers.

Design #4

Desc – This prototype is designed for aerodynamics, visual appeal, and stability. The plane shaped front should carve through the air and the symmetrical design should be stable. Coupled together, this design is sleek and aesthetically pleasing.

Pros/Cons – This design improves on design 3’s lack of spatial efficiency and is light-weight as a result, although the propeller-based movement does not meet the team’s standard for control.

 

Development Decision

Although more expensive, the team developed design 2 because the benefits to efficiency, control, and weight greatly outweighed the cost. We believed belt-based movement would allow us to control the AEV more precisely and therefore reduce testing and development costs via reduced trial-and-error. The team also kept developing runner-up design 4 as a back-up in case design 2 didn’t work out.

Concept Screening Matrix

Success Criteria	Reference	Design 1 (Vincent)	Design 2 (Mitchell)	Design 3 (Alec)	Design 4 (Ryan)
Aesthetic	0	0	+	0	+
Efficiency	0	0	+	0	0
Control	0	–	+	0	0
Weight	0	+	+	+	+
Cost	0	0	–	0	0
Sum +’s	0	1	4	1	2
Sum 0’s	5	3	0	4	3
Sum –‘s	0	1	1	0	0
Net Score	0	0	3	1	2
Continue?	No	No	Yes	No	Back-up

 

Concept Scoring Matrix

		Reference		Design 1 (Vincent)		Design 2 (Mitchell)		Design 3 (Alec)		Design 4 (Ryan)
Success Criteria	Weight	Rating	Weighted Score	Rating	Weighted Score	Rating	Weighted Score	Rating	Weighted Score	Rating	Weighted Score
Aesthetic	15%	1	.15	3	.45	5	.75	4	.60	4	.60
Efficiency	20%	3	.60	3	.60	5	1.00	3	.60	3	.60
Control	30%	2	.60	2	.60	5	1.50	2	.60	2	.60
Weight	15%	3	.45	3	.45	5	.75	3	.45	4	.60
Cost	20%	3	.60	4	.80	2	1.00	4	.80	4	.80
Total Score			2.40		2.90		5.00		3.05		3.20
Continue?		No		No		Develop		No		Develop (as back-up)

Focus Areas

For our AEV, we looked for a design that was visually appealing, energy efficient, easy to control, and inexpensive. Our final design should rate the highest in all four of the above mentioned categories. At this point, no analysis of the above 4 designs had been done, analysis had only been done to the reference AEV. To establish a baseline of which to compare the AEVs, this reference AEV should be evaluated on the above categories and then all 4 designs’ rankings compared to the reference AEV’s rankings. After doing Preliminary R&D Activity 3, we collected data just about the energy efficiency and the ease of control. The aesthetics of the reference AEV were judged without data needed.

Visually Appealing:

The reference AEV is just a skeleton and so it doesn’t look that appealing at all. Our final design should make consumers excited to travel on the AEV and like looking at it. While having a visual appeal, it should also be aerodynamic, contrary to the reference AEV.

Energy Efficient:

Executing the code from Preliminary R&D Activity 3 showed that the power usage of the reference AEV, is at a max of about 6-7 watts as seen below, but since there is nothing to compare it to, no conclusions could drawn.

Ease of Control:

The reference AEV is powered by two fans. During testing on the rails on Preliminary R&D day 3, the AEV took a while to get up to a reasonable speed, and then stopping at that speed also took a long time. Because the ‘brake’ command doesn’t actually break the AEV (it just cuts power from the motors), any fan based AEV will most likely have the same issues as the reference AEV. In order to get the stopping and starting of the AEV just as needed, many test runs will be needed to see how the Arduino should be coded.

Cost:

Since the reference AEV is bare bones without any extra parts, it will likely be the least expensive compared to the other prototypes that almost all use some sort of 3D printing or extra parts.

Iteration 2: Yellow Submarine Mk 1 (offset wheels and enclosed capsule)

This design was used throughout aR&D 1 tests and ran very well and never broke; however, this design had two major design flaws.

Pros:

1. Ran like it was supposed to do (via the code)
2. Minimal sliding
3. Sleek looking
4. Our team won the grant, meaning all of our printing is free as the grant covers the price. This means that structurally it has the same exact cost as the reference AEV which is very minimal

Cons:

1. The first major flaw was the fact that having the wheel to wrap the belt around (top orange wheel) pushed out the wheel that runs on the track (larger black wheel). In doing so the wheels became offset, as seen above; therefore, whenever the AEV was put on the track it was leaning and wasn’t oriented straight. One way to fix this problem would be to just push out the other track wheel. The reflectance sensors prohibit that solution as when that was tried, the sensors were too far away from the wheel and didn’t pick up any data. Without the reflectance sensors the AEV wouldn’t be able to tell its position or how far it went, thus, any solution must keep the wheel close against the reflectance sensors so they work.
2. The second flaw was the fact that the orange capsule that holds all the hardware is fully enclosed. That means between every test everything would have to be unscrewed to get to the Arduino, and after pushing the start button, the whole thing reassembled quickly in order to get the AEV on the track before it started moving. Adding a block of code that started the AEV after it went to a negative position allowed for the assembly of AEV to be done as slow as needed, put on the track, and then pulled backwards to that negative position until the AEV started. While this worked fine for the aR&D tests, for the actual final test, the AEV can not be touched after it is put on the track, so it wouldn’t be allowed to be pulled backwards when we conduct the final test.

Iteration 3: Yellow Submarine Mk 2 (enclosed capsule)

In order to fix the offset wheels, this iteration uses a smaller wheel attached to the track wheel in order to minimize the offset. This solution works so much so that the offset nor its effects can not be seen when the AEV is on running on the track, meaning the offset is now negligible.

Pros:

1. Ran like it was supposed to do (via the code)
2. Sleek looking
3. Our team won the grant, meaning all of our printing is free as the grant covers the price. This means that structurally it has the same exact cost as the reference AEV which is very minimal

Cons:

1. The same problem with the enclosed capsule is still present. A new design was submitted that should fix this problem by having access holes in the capsule where the start button for the Arduino can be pushed.
2. Since the wheels are now aligned there is no extra friction of pushing the wheels out against the track, this resulted in a visibly greater sliding distance.

Iteration 4: Yellow Submarine Mk 3 ( The Red Submarine)

 

 

This new design fixes all of the previous designs’ drawbacks and did so without adding any additional drawbacks. As visible in the above images, there is a slot for the start button on the Arduino to be pushed, as well as a slot for an metal attachment for the magnetic connection with the passenger car.

 

Focus Areas 2

For our AEV, we decided on the Yellow Submarine design, whose iterations are shown above. Since we have now completed aR&D 1 and aR&D 2, we can evaluate most of our focus areas and see how the AEV is doing in those categories. For our AEV, we are looking for a design that is visually appealing, energy efficient, easy to control, and inexpensive. Our final design should seek to achieve a maximum in all of the above categories.

Visually Appealing:

The successive iterations of the Yellow Submarine keep making it sleeker and more appealing. From Mk 2 to Mk 3 the AEV  does get a little bulkier, but allows for all of the internals to be safely enclosed and hidden. Mk 3 also allows for the firm attachment of our magnetic attachment site to the AEV without risk of it falling off. A final appealing aspect of Mk 3 is the new paint job, which, as it is zooming along the rails above the ground will surely make heads turn in awe.

Energy Efficient:

The advanced R&D tests were to find the optimal power level at which the AEV runs the most cost-effective both with and without the passenger car. Looking at the two tables below (first is without the passenger car, second is with), it is clear that the ideal power levels of 35% and 45%, are among the power levels with the lowest energy consumption and while they don’t minimize the energy consumption, they strike the perfect balance between time taken and energy consumed.

 

Ease of Control:

An auxiliary part to the advanced R&D tests was finding the sliding distance for each power level, with and without the passenger car. This distance could then be factored into future Arduino codes in order to compensate for the amount the AEV will slide after the motor power is shut off. The first graph below shows that for the selected power level of 35%, the AEV slide about 59 marks. While that is still a lot of distance covered while sliding, the AEV did this consistently; therefore while the AEV does still slide, it can easily be compensated for due to its consistency. Looking at the second graph below shows the same exact thing except for the loaded AEV power level of 45%. This power level produced a reduced slide distance of about 48 marks despite the increased power level due to the added weight of the passenger car. Once again while not the best sliding distance, it was consistent making the ease of control of the AEV still extremely high.

 

Cost:

The main point for the Advanced R&D tests was to find the most cost effective power level to run the AEV motor at when it didn’t have a load and when it had the passenger car. In that way, the resource cost for the AEV could be determined as opposed to the hardware costs (Arduino, 3D printed shell, wheels, etc.). The two major resources that factor into the budget of the AEV are the energy and time costs. In order to find the most cost-efficient power level to run the AEV at, there needed to be a balance between energy consumed and time used. Both the tables below detail how this balance was found. Scaling up our test trials to the full length of the track (energy converted to total Joules and the average speed converted to total time using the distance of the track) the total energy cost and total time cost for each power level was found. When added together, the total resource cost for the power levels was determined. The lowest cost in each table corresponded to the most cost effective power level; therefore, 35% was determined for the unloaded AEV and 45% was determined for the AEV with the passenger car. Finding the most cost effective power level to run the motor of the AEV at both down and back up the track significantly reduces the cost of the AEV and is a huge step to making it as optimal as possible.

 

 

Iteration 5: Yellow Submarine Mk 4 (The Final One)

This final design incorporates all previous design adaptations along with two strategies that reduce slipping of the drive wheel and the slide distance, the new passenger car attachment method and the latex covering on the drive wheel (from aR&D 3)

Focus Areas 3

As the last design, the Yellow Submarine Mk 4 ranks the highest in all 4 of the focus areas as explained below.

Visually Appealing:

The only change in appearance is the addition of the attachment arm and the purple latex wheel covering. Both of these additions don’t alter the appearance of the AEV too much, if at all, meaning Mk 4 is just as aesthetically pleasing as ever.

Energy Efficient:

Taking previous design and Arduino code energy efficiency, this design takes efficiency one step further. The two new additions both eliminated the the slipping of the drive wheel, as seen below, which caused a waste of energy, meaning using both of them together decrease the energy used and thus increase AEV energy efficiency.

As seen above, both the tire and attachment strategies eliminate the bouncing/slipping of the drive wheel, thus conserving energy.

Ease of Control:

Both of the added strategies not only eliminate the drive wheel slipping, but also decrease the slide distance of the AEV.

The latex tire decreased the slide distance by almost 22 marks while the new attachment decreased the slide distance by about 3 marks. Using both of these strategies together means the slide distance for the final AEV was decreased by 25 marks. The slide distance reduction means that when the Arduino was coded for a distance, it essentially went that distance with almost no variation due to the reduction in slide distance. That means the control of this design skyrockets, making it the easiest AEV design to control out of all the other iterations.

Cost:

The two biggest factors impacting the cost of the AEV are the time and energy costs from the final runs. As shown above, this design reduces the energy consumption of the AEV and because the slipping of the drive wheel is eliminated, the AEV doesn’t waste time slipping around, thus decreasing the time consumed as well. Because both the time and energy consumption are minimized in this design, the Yellow Submarine Mk 4 is the cheapest, most cost-effective iteration of the AEV.

 

---

Implement

Performance Test #1:

The first performance test consisted of going from the start line, stopping at the first sensor, waiting 7 seconds for the gate to open, and then go through the gate (the first two green arrows in the image below). Yellow Submarine Mk 3 was used in this performance test and worked just as intended and passed all requirements for this test with no mistakes.

This code actually caused the AEV to travel all the way to the other end of the track, essentially to the exact position it needed to go to pick up the passenger car. That means, this test data is a pretty good estimate of the data for the un-encumbered (no load) AEV.

Performance Test #2:

The second performance test consisted of going from the start line, all the way to the passenger car, hooking onto it and pulling it backwards a little bit (all of the green arrows below). Yellow Submarine Mk 3 was used in this performance test and completed it in full, passing all requirements with no mistakes.

Final Performance Test:

The final test consisted of three separate trials with the AEV running the full length of the track each time (both the green and blue arrows below).

Trial #1:

The AEV perfectly completed the course with an extremely low energy usage of 136.7 joules and a fast time of 42.3 seconds.

The most notable thing about this graph is how the energy increases much more around the 25 second mark, when the passenger car is attached and power increases to 45%, just as predicted.

Trial #2:

Despite using the same exact code as trial #1, the AEV went to far at the first gate, tripping the second sensor, and then went too short at the station, not attaching to the passenger car. One the way back, the AEV ran perfectly, resulting in an overall penalty of -4 points. The AEV completed this trial in the fastest time out of all the trails at 40.6 seconds with an energy usage 145.5 joules.

Trial #3:

After re-coding the Arduino to fix the the problems from trial #2, the AEV was ran again. Overall, the AEV ran just as it was supposed to; however, after stopping at the gate the first time it failed to run again after the 7 seconds for some unknown reason, meaning it needed to be tapped a little to start moving again (-2 points as the penalty). This resulted in an overly large increase in the energy and time used, with 180.8 joules being consumed in 46.9 seconds.

 

 

---

Evaluate

Performance Tests #1 and #2:

Both of these performance tests showed the large impact of the AEV sliding on the accuracy of the AEV. The more it slid, the harder it was to code for tests, as changing the speed would then change the slide distance which would then cause the distance and the speed to have to be reevaluated in a positive feedback loop. The other main issue was that during performance test #2 when the AEV reversed with the passenger car, the drive wheel seemed to slip and not fully catch on the track. This caused the AEV to barely move backwards and during an actual final test, would cause the time and energy costs to sky rocket. These two issues were addressed in advanced R&D 3.

Final Performance Test Future Ideas:

The first trial of the final test was a success with the AEV completing the course with a perfect and low energy and time consumption. The next two tests were not perfect runs despite the same code. This indicates that the difference in batteries (the first trial was on one day while the other two trials were the following day) played a large part in the performance of the AEV. Knowing that, it suggests that a fresh fully charged battery should be used for each run and/or that the AEV should recharge its battery as it runs. This battery replenishing could be done through solar panels or mini-wind turbines on the outside of the AEV and could be implemented in the future after the public response to the AEV is evaluated (to see if the AEV will actually be used and ridden on), but won’t be part of the Yellow Submarine final design.

 

---