MOTION PLANNING AND CONTROL OF 2-BAR TENSEGRITY SYSTEM

Abstract

A tensegrity is a special mechanical structure which has a special design consisting of bars and tensile elements. Tensegrity systems became popular due to their ability to withstand heavy loads in spite of having a very light weight. Efficiently controlling the dynamics of tensegrity structures is very challenging due to their complex dynamics. This thesis will provide the formulation and implementation of two closed-loop control methods: Model Predictive Control (MPC) and Neural Networks (NNs). Specifically, the method based on NNs consists of imitating the system trajectories generated by MPC. The computation time of MPC is too high for real-time implementation, but this is not the case for NN, which has a much lower computation time.