Procedural Generation of Traffic Signs

Procedurally-generated virtual urban worlds typically miss plausible signaling objects on the road network, unless they were manually inserted. We present a solution to the problem of procedurally populating a given urban road network with plausible traffic signs. Our tagged graph approach analyzes the road network using a rule-based reasoning mechanism that represents relevant traffic rules, in order to identify potential sign locations. Eventually, a context-based reduction step helps choose the most suitable candidates, taking into account a variety of real-world rules, and determines their actual place and orientation. We discuss the performance and validation of our approach, and conclude that its generality and flexibility make it a very convenient extension to many procedural urban environment applications.