Walking task space control using time delay estimation based sliding mode of position Controlled NAO biped robot
- PDF / 2,248,653 Bytes
- 10 Pages / 595.276 x 790.866 pts Page_size
- 48 Downloads / 245 Views
Walking task space control using time delay estimation based sliding mode of position Controlled NAO biped robot Yassine Kali1
· Maarouf Saad1
· Jean-François Boland1
· Jonathan Fortin1 · Vincent Girardeau1
Received: 5 May 2020 / Revised: 10 September 2020 / Accepted: 17 September 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract This work proposes a simple technique to implement in real-time a nonlinear robust controller on a humanoid NAO robot that does not have direct drive joints. The key trick consists of designing a time delay estimation based sliding mode controller without any prior knowledge of the robot’s dynamics to alleviate the heavy computations since the on-board processor has low computational power and to deal with the effect of the uncertainties. Then, the calculated torque inputs will be converted to position controller for the servo actuated NAO robot using an appropriate transformer. The proposed method will ensure in addition to high precision tracking a fast convergence during the reaching phase. The proposed control architecture is validated through experimental results on a real NAO robot. Keywords Sliding mode · Time delay estimation · NAO humanoid robotics · Position controlled servo
1 Introduction 1.1 Context and motivation Nowadays, the field of humanoid robots has seen revolutionary growth in terms of design, walking control and practical operations. Humanoid robots represent an exciting and interesting topic of research due to their complexity and to their usability in several and hazardous applications where they assist or replace human. Most of the existing humanoid robots use DC servo motors including internal controller as joint actuators. Most of time, the implemented controller consists of the well-known linear Proportional-Integral-Derivative (PID). In this case, only position control based on the kinematics can be achieved. However, this kind of control is not enough to perform a compliant behavior. Moreover, robustness against perturbations is not ensured. For these reasons, a robust dynamic model-based control must be designed. One of the solutions, to integrate compliance in the humanoid robot walking motion, consists of converting the model-based torque into a model-based position. This solution has been explored and a torque-position transformer
B 1
Yassine Kali [email protected]
based on the identified transfer function of the actuator has been proposed [8,15]. This method has been successfully implemented on the 5-DOF arm of the HONDA ASIMO Humanoid robot. However, the lower body of this humanoid robot was controlled by the Zero Moment Point (ZMP) based stable balance. Therefore, a primary focus of this work is to prove the applicability of this transformer to achieve walking tasks for the lower body of a humanoid robot with position controlled actuators. As all existing nonlinear systems, the NAO robot [1] is subject to a wide range of matched uncertainties caused by unmodelled dynamics, parameters variation, ground imperfec
Data Loading...
