Tesi etd-11072022-120156 |
Link copiato negli appunti
Tipo di tesi
Tesi di laurea magistrale
Autore
DEL MEDICO, RICCARDO
URN
etd-11072022-120156
Titolo
Design of a whole-body roll balancing control for a two-wheeled humanoid robot
Dipartimento
INGEGNERIA DELL'INFORMAZIONE
Corso di studi
INGEGNERIA ROBOTICA E DELL'AUTOMAZIONE
Relatori
relatore Prof. Bicchi, Antonio
supervisore Ing. Zambella, Grazia
supervisore Ing. Zambella, Grazia
Parole chiave
- balancing
- control
- humanoid robot
- lqr
Data inizio appello
24/11/2022
Consultabilità
Tesi non consultabile
Riassunto
The aim of this work is to find out a stabilizer controller, which acts on the
not actuated roll degree of freedom, for the robot AlterEgo, in order to make
a two wheeled humanoid robot able to stands just on one wheel, balancing its
position through the use of the arms.
Indeed, an humanoid robot is able to move its arms, which in this case behave
five degrees of freedom, and that will be used in order to reach the goal, more or
less like a tightrope walker, or more simply a human being, would do to balance
himself only on one foot.
The workflow begins from a feasibility study on a simplified model, which looks
like an inverted pendulum with two arms, where a controller is elaborated in
MATLAB and tested in Simulink and then in Gazebo; afterwards, a similar LQR
controller is implemented on a realistic model of the robot, on which the study
is more in-depth, including an estimate of the region of asymptotic stability;
the actual simulation is carried out in the ROS and Gazebo environment, in the
optics of trying some experiments on the real robot, that seems possible from the
results obtained in Simulink and confirmed by the Gazebo simulation, with due care.
not actuated roll degree of freedom, for the robot AlterEgo, in order to make
a two wheeled humanoid robot able to stands just on one wheel, balancing its
position through the use of the arms.
Indeed, an humanoid robot is able to move its arms, which in this case behave
five degrees of freedom, and that will be used in order to reach the goal, more or
less like a tightrope walker, or more simply a human being, would do to balance
himself only on one foot.
The workflow begins from a feasibility study on a simplified model, which looks
like an inverted pendulum with two arms, where a controller is elaborated in
MATLAB and tested in Simulink and then in Gazebo; afterwards, a similar LQR
controller is implemented on a realistic model of the robot, on which the study
is more in-depth, including an estimate of the region of asymptotic stability;
the actual simulation is carried out in the ROS and Gazebo environment, in the
optics of trying some experiments on the real robot, that seems possible from the
results obtained in Simulink and confirmed by the Gazebo simulation, with due care.
File
Nome file | Dimensione |
---|---|
Tesi non consultabile. |