Vision
The robot is equipped with a camera that allows it to interact with objects in its environment. The robot’s vision software recognizes objects by their shape and color, using methods that take into account changes in scale and lighting.
Given constraints of the environment (i.e. flat floors), the robot can
perform simple trigonometry to determine the object location with respect
to the camera. Using the robot’s dead-reckoning-based floor position, the
object position can then be transformed into world coordinates.
Our accuracy is limited primarily by the resolution of the camera. At a
distance of several meters, objects positions can be determined to within
a few centerimeters.

The vision software places a marker to show the
image coordinates of the base of the cone.
|