Cellular Neural Networks Based Local Traffic Signals Control at a Junction/intersection

In this paper, the core objective is to show how far a local and ultra-fast optimal control of the traffic signals is possible by involving a computing paradigm based on cellular neural networks (CNN). The first important issue is a systematic reformulation of the basic problem to make it solvable by a CNN processor. CNN has the advantage of being extremely fast while solving even nonlinear complex problems. CNN can further be easily implemented on embedded hardware platforms (especially FPGA and GPU). A further advantage of a CNN based computing is that one can consider the full model in all its nonlinearity and complexity; CNN does cope with both aspects robustly. In the core, we are building a novel neuro-computing based adaptive local traffic control concept that will be extended in a next step for an “area traffic control”, where the proactively optimized coordinated control of many junctions either in an urban context or on the highways will be in the focus of the CNN based optimization.