My experience with GPS is that placement may "jump" as much as 50m or more when stationary, so while it's good for things like the Darpa, I wouldn't use it for close quarter navigation, but that's just me... At what initial distance do you expect to color detect the cone?
If it's a regular traffic cone, it will likely be fluorescent, so just turn off the sun and use an ultraviolet beam to detect it, it will glow like on fire
Seriously, the fluorescent effect should help you with the sun as well, and a color filtered receiver, matched to the cone, shouldn't be hard to make (eg. using yellow and red filters as used in front of stage light, to make up the right color).
A pulsed (color filtered, UV/Blue LEDs or what experiments reveal as useful) emitter will help filter out ambient light.
Assuming it stands on plain ground, you know at which height to scan for it.