(23/6-27/6)
This week, the first order of business was to fix the overly large sensor holders. Next, we worked out a suitable arrangement for the load sensors that best corresponds to the pressure points of a human body laying down. Since we had 16 individual load sensors connected into four sets of four, we divided the body into four sections that can be distinguished by colour.
The arrangement is vertically symmetrical (down the y-axis).
From top to bottom: sensors by the head, shoulders, upper back, elbows, tailbone, wrists, calves and ankles.
We tested users of different heights ranging from ~150-180cm, and noticed some variation in sensor position. For example, some sensors would be at the thighs of taller users instead of the thighs of a the calves user etc.
We also tested the sensors to see if they could sense human body weight accurately (we used metal weights earlier and tested that they were accurate to 0.1kg) but found that at times they fluctuated despite the person laying down not moving. We go on to fix this problem in week 5.
Technical issues aside, we also considered the comfort level of users when lying down on the mattress, as when we tried them out ourselves we could feel the hard metal from the sensors. We first attempted to use stuffing to fill the gaps between the sensors, but quickly found that it was impractical as the stuffing moved around too much once the mattress was placed over.
After, we also tried to stuff the stuffing into the mattress cover with the foam but once again the stuffing was too flimsy and could not stay in place, making the mattress lumpy.
We settled on buying another piece of foam such that the sensor arrangement would be placed between the two pieces.
Finally, we improved our code by including a function that would store our calibration factors into the EEPROM so we did not have to recalibrate and re-tare the sensors every time we set the arrangement up.