Control of inverted pendulum robot that can change the center of gravity by use of self-evolving adaptive controller
A. Tomohisa Ashida, B. Tetsuya Minatohara, C. Hiroyuki Inoue · International Conference on Electrical Machines and Systems · 2016
To adapt to a changing environment, to use multiple models is effective. The authors have also proposed two multiple models, i.e., Self-Organizing Adaptive Controller (SOAC), and Self-Evolving Adaptive Controller (SEAC). SOAC has a fixed structure and therefore, an incremental learning to adapt a novel environment is difficult. On the other hand, the SEAC has an evolving structure and therefore, incremental learning is very easy. In addition, the network structure tends to be smart. This is because the number of modules is one at first and the new module is increased as needed. Before now, simulation study regarding the SEAC has been proposed but there is no report dealing with using actual equipment. Thus, in this paper, we report control experiments using the SEAC with actual equipment developed by the authors.