@@ -76,28 +76,25 @@ At this point we started wondering whether the wires to the engines were creatin
As we saw no discernible difference with the lights turned on or off, we left it on for the remainder of the experiments as it was vastly easier to work with.
Because we know from past experiences that the measured reflected light for the light sensor is affected by the color of the surface, we suspected that either the color of the bathroom floor, or the inconsistency in the color (it was blue-ish with some white) could possibly affect the reliability of our balancing robot.
Because we know from past experiences that the measured reflected light for the light sensor is affected by the color of the surface, we suspected that either the color of the bathroom floor, or the inconsistency in the color (as seen in figure 2) could possibly affect the reliability of our balancing robot.
TODO: Billede af badeværelsesgulv
To remedy this, we tried experimenting on a black surface that we dragged into the bathroom, which can be seen in picture (BILLEDE). A notable problem with this surface, however, was that it was rather rough, and as such could have slightly inconsistent reflection.
To remedy this, we tried experimenting on a black surface that we dragged into the bathroom, which can be seen in figure 2. A notable problem with this surface, however, was that it was rather rough, and as such could have slightly inconsistent reflection.
*Figure 2: Black surface in the bathroom, as well as the black and the grey surface next to each other for comparison.*
The light sensor returns very different measurements of different colors, and the bathroom floor has lots of colors (see picture). Therefore we tried the big black thing which gives a completely different setting. Both the color of the surface (which is now black instead of blue-ish) and the consistency in the color (completely black instead of different blue-ish colors) and how even the surface is, as the black thing is very uneven.
As we now couldn't use the door as a means of a solid standpoint for the robot to lean against when calibrating, we now used a box found in the Zuse building instead, as seen in figure 3.
*Figure 2: Using a box to angle the robot for calibration.*
*Figure 3: Using a box to angle the robot for calibration.*
A problem we discovered with this box was that in some places it was casting a shadow over the surface that we were calibrating the robot on, causing the initial calibration value to be vastly incorrect from what it should be, once the box was removed. Therefore we made a conscious effort to always place the box and robot in a position where the box wasn't shadowing the robot.
We tried to use a box, to still work around the fact that our press effects the robot. But the box was casting a shadow. We moved it so that the light level around the robot would be the same while calibrating, leaning against the box, as while running with the box gone.
Our experiments with this setup once again did not show any noticeable improvements in the balancing ability of the robot, despite the new surface.
The light was off.
Did't work very weel.
After all of these various experiments with change of different factors, we came to the conclusion that this program simply doesn't allow the robot to balance very well under any conditions, and out of all the variables we could change, the (by far) most significant factor was the calibration position/angle.
The conclusion of this experiment is that the biggest factor is the calibrating position/angle.
### Algorithm-discussion
We started by changing the int's to floats - 'cause Ole said so. Aaaaand this will give a more precise adjustment of the PID controller.
