Rodney brooks had invented behavior based robotics in the 1980s in response to the former symbolic AI which is using a world model. A world model needs symbolic grounding which means, that the outside world is mirrored into the system. Then the planner solves a problem in the symbolic system and the result is send back to the outside world.
Behavior based robotics hasn't an internal world model. Reactive systems aren't planning into the future and they have no idea what the outside world look like. This allows to bypass the grounding problem. If no world model is there, it can't become out of sync with the reality.
But, even behavior based robotics which are often line following EV3 kits and light seeking wheeled robots need an input signal in form of sensory data. Is the raw data of the light sensor the world model for behavior based robots?
 Wikipedia: Symbol grounding problem, https://en.wikipedia.org/wiki/Symbol_grounding_problem