Our area of focus was the motion model. We believed that through multiple trails and measurements we could provide a more accurate estimate for how far the robot has moved. We tested this by having the robot move a randomn distance forward thirty times. We recorded its starting and ending position by chalking the carpet right in front of the robot (the chalk held perpendicular to the ground and pressed against the front of the robot for consistency). The distance measured was recorded by tape measure from the center of both marks. In order to get the estimated distance, we used the measurment tool in the simulator, measuring from front to front again. The following are the results:
|Trial||Estimate||Actual||Distance Off||Percent Error|
*All measurements are in meters.
*A negative number denotes that the estimate was short by that amount/percentage.
The robot averaged overestimating the distance moved by 0.81%. Accounting for human error, this amount is minor for us to base changes off of.
Other pertinent data:
|Dimension||Estimates (in)||Actual (in)|