M.Sc Student | Lior Alpert |
---|---|

Subject | Minimum Energy Control of Redundant Robots |

Department | Department of Autonomous Systems and Robotics |

Supervisor | Full Professor Halevi Yoram |

Full Thesis text |

A manipulator is said to be kinematically redundant when more Degrees-Of-Freedom (DOFs) are available than required to execute a given end-effector primary task. In kinematically redundant manipulators, there are infinitely many solutions to the inverse kinematics problem, which allows judicious use of the redundancy to improve various performance criteria. The redundancy can be used in avoiding collision with obstacles and improving the accuracy. Another important use of the redundancy can be reducing the amount of energy that the system consumed while executing a given task. Minimization of the energy is very important when dealing with mobile robots that have finite amount of energy (e.g. batteries or solar panels).

We focus on minimizing the invested energy of the system by using optimal control theory. The optimization is carried out under the constraint of exact tracking of the given end-effect trajectory, while also considering the displacement limitation. Since optimal control problems are always two point boundary value (TPBVP) and have to be solved offline, the entire desired end-effector path must be known in advance. For applications where only a finite future of the path is known, a finite-horizon solution is proposed. By using this method, the solution is getting closer to online solution and can take into account changes in the path.

In this thesis we show that by adding redundant degrees of freedom to a linear system, the energy consumption can be reduced. In the example that is used, we show that by adding two redundant DOFs, up to 70% of the energy can be saved. This large amount of energy might be worth the mechanical complication due to the addition of the redundant DOFs.