Affiliations: [a] The Centre for Transformative Innovation (CTI), Swinburne University, Victoria 3122, Australia
| [b] Institute for Intelligent Systems Research and Innovation (IISRI), Deakin University, Victoria 3216, Australia
Correspondence:
[*]
Corresponding author. Sahar Araghi, The Centre for Transformative Innovation (CTI), Swinburne University,
Victoria 3122, Australia. E-mail: saraghi@swin.edu.au.
Abstract: This paper compares the performance of three different types of traffic signal timing controllers that all of them apply fuzzy logic systems in their controlling structure. The comparisons are done between an interval type-2 adaptive neuro-fuzzy inference system (IT2-ANFIS) traffic signal controller, a type-1 ANFIS (T1-ANFIS), and a fixed-fuzzy logic, and three different fixed-time controllers as benchmarks. Obtained results demonstrate the superiority of the fuzzy controller over traditional controllers. The traffic network used in experiments is a network of nine intersections and all controllers are designed in distributed form. Each intersection has its own controller while it considers its neighbor intersections’ traffic jam in determining signal time for each traffic phase. The average performance of IT2-ANFIS, T1-ANFIS, and fixed-fuzzy controllers against the fixed-time controller are approximately 35%, 30%, and 8%, respectively.
