Adaptive control of a dynamic system using genetic-based methods


This paper presents Genetic-based learning Algorithms (GA) for automatically inducing control rules for a typical unstable, multi-output, dynamic system - a simulated pole-cart system. We compare the performance of the genetic method with other learning algorithms for the same task. Our experiments demonstrate that the results using GA as controller are comparable to the existing methods. We also suggest a further enhancement of the genetic learning by applying the recently developed Structured Genetic Algorithm(sGA) which appears to offer impro11ements over the simple genetic algorithm in robustness and speed of optimization.

Publication Title

IEEE International Symposium on Intelligent Control - Proceedings