Adaptive traffic signal control based on bio-neural network

Urban traffic management is one of the major concerns for big cities around the world, due to its negative impacts on society. Several approaches of traffic signal control based on artificial intelligence techniques or on control theory were proposed as alternatives to mitigate this problem. However, it is a challenge to reach a good solution, as the urban traffic is a complex and dynamic ecosystem. On this scenario, this paper proposes an adaptive biologically-inspired neural network that receives the system state and is able to change the behavior of the control scheme as well as the order of semaphore phases, instead of prefixed cycle-based ones. Proposed adaptive control was evaluated on a single intersection scenario. Despite analyzing the control of a single intersection, the model proposed is modular, allowing the control of multiple intersections. The analyses conducted herein showed that the model is robust to different initial conditions and has fast adaptation between system equilibrium states. Simulations with SUMO showed a better performance than a cycle-based traffic responsive control method regarding reactivity and capacity tests, in which the relevance of the constant monitoring and acting became evident. (C) 2017 The Authors. Published by Elsevier B.V.