Design of a Wheeled Inverted Pendulum as a platform for learning based control

Recently, a robust system as a platform to study multi learning based control methods was developed in our research. Learning based intelligent control is a challenging research topic. It is a kind of intelligent control approaches by learning a mapping between world states and actions from demonstrations. To carry out experiments, many platforms had been presented in past research. But most of them are either too complex or too expensive that researchers have to pay more attention on platforms rather than research. Thus, a low cost but robust and multi-purpose platform is quite helpful and valuable for such a research topic. Wheeled Inverted Pendulum(WIP), with clear model and dynamically stable static unstable characteristic, may be an ideal choice. This system uses two paralleled wheels to bear weight, keep balance and make movements at the same time. However, In this paper, we present an implementation of the design of a low cost, high quality WIP with special consideration for learning based controlling research with experiments illustrated.