Test Results

Rabbit Pelt Testing

Two setup configurations were tested with a rabbit pelt. We compared two types of spring, according to its spring constant or its "strength," as well as the preload length or how long the spring was pre-stretched before trigger.

Rabbit Pelt Procedure:

• Choose spring and load system at Hole #1, mounted vertically

• Press with rabbit pelt (wrapped around stuffed animal) to trigger

• Remove spring, and photograph

• Tape hair sample to paper to document the trial

• Repeat at Holes #2 and 3

Humidity Test

• Device was held over boiling water in the steam stack.

• Rabbit pelt testing procedure was followed.

Results:

Counting the number of hairs was not feasible at home due to clumping. Matlab image processing was used to quantify the amount of hair sampled:

• As shown above, the hairs were taped together then scanned.

• Each sample box was cropped to the same size to keep the area consistent.

• Each picture was turned greyscale, so that each pixel was assigned a value 0-255.

• All pixels below 230 were considered "hair area." This cutoff value was verified by viewing the binary image.

• Results of the rabbit pelt test are graphed to see how each setup performed.

Conclusions

• The spring preload plays an important role in the spring's ability to sample hairs.

• The differences between types of springs is negligible. This is good, since biologists may not have the specifications of hardware springs.

• Device isolates slower in humid conditions, however samples were still fully isolated and sufficient hairs were collected.

Raccoon Testing

We were fortunate enough to be able to test our design on euthanized raccoons supplied by San Diego Wild Life Removal Service. The same test procedure was used on frozen raccoons as was used in the rabbit pelt testing. However, here we also warmed up the skin of the animal with a hair drier to better mimic that of a live one. Lab coats, masks ang gloves were worn by team members handling the raccoons.

The same Matlab analysis was done with scans from this round of testing, as well as hair follicle counts. The DNA present in the hair follicles is higher quality than that of the hair strands. Researchers gave us the target of at least 10 hair strands and 3 hair follicles per sample, both thresholds were reached. For the hair follicle test, we used a non frozen, fresh raccoon that was brought into the office as the workers were leaving for the day. Due to time constraints, we were not able to re-test the different springs and preloads with the non-frozen animal, only able to test for the presence of follicles present in the samples. A 4th hole was also added below to test a larger preload.

Here we can see that we have met our goal of 3 hair follicles sampled with the idea spring and preload setup. We can also see that we have sampled a sufficient hair quantity with a variety of different spring and preload setups. This shows how robust our solution is, where at worse case of optimization, we still reached our benchmark and greatly exceeded it with our optimal setup. The graph below shows the results of the Matlab image processing.

From our rabbit pelt testing, we predicted that the strongest spring at the largest extension would be the most effective. However we found that hole 2 was more effective than 3 with the strong spring. We are unsure which is the best case as the two different tests had different results. The raccoon had greater variability in hair density across the animal, which may be a source of error. Another source of error is overlap of hairs in the scans. If there is a high amount of overlap of hairs in the scans, the percent coverage will be lower than it actually is. More testing will need to be done to be sure, however both setups greatly exceeded 10 hairs sampled.

This testing assumes that biological mechanism of the hair releasing from the dead animal is the same as that of a live animal. Live animal testing is scheduled for later this summer with our prototypes to replicate these results. This future testing will also record how animals react/interact with our device.