search

deeper-blue / sdp-robot

The robot code for Deeper Blue, an assistive chess-playing robot. Developed for the System Design Project, a 3rd year course offered at the School of Informatics, University of Edinburgh.
2 stars 0 forks source link

Decrease Arch error #35

Closed pilif0 closed 5 years ago

pilif0 commented 5 years ago

Based on Demo 2 QA (move 50 cm), there is an average error of +0.59 cm in the actual distance moved and -0.04 cm (~1 degree) in the perceived distance moved. The perceived distance is close enough for our purposes. But the +0.63 cm difference between reality and how far the motor thinks it has moved should be improved.

The data would suggest that the main problem is in converting the linear distance into rotation of the motor, because the desired rotations were carried out but the distance was overshot. Therefore the problem is probably in the conversion calculation (angle = distance / circumference * 360), probably with the wheel circumference constant (as that is the only variable in the equation). The test distance was 50 cm and with wheel circumference of 12.5 cm, resulting in 1440 degrees of movement (4 rotations). The actual distance of 50.63 cm would instead correspond to a circumference of about 12.66 cm over the same 1440 degrees.

Changing the arch wheel circumference constant might lead to a decrease in the error.

pilif0 commented 5 years ago

The actual circumference calculation above is wrong.

To compute the actual circumference, I divide the actual distance travelled by the actual rotations taken. The actual distance is 50.59 cm, while the actual rotations are 3.9968. However that still results in circumference of about 12.66 cm (difference in result removed by rounding).

pilif0 commented 5 years ago

This change resulted in average distance travelled being 50.01 cm with st. dev. of 0.03

This is more than enough for our purposes.