.png){kind=icon}
.png){kind=icon}
Return
Imitation-based Path Planning and Nonlinear Model Predictive Control of a Multi-Section Continuum Robots
.png){kind=icon}
PRE
AI
DOI：10.1007/s10846-023-01811-8.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Flexible robots have exhibited impressive features in working in congested environments due to their compliance behavior and morphological structure. However, designing motion planning techniques and robust control strategies that actively control their deformations are challenging in many applications. Thus, this article presents the learning by Demonstration (LbD) approach for planning the spatial point-to-point motions of a multi-section continuum robot. Via teleoperation, the human demonstrations are captured by moving the flexible interface with similar kinematics of the active robot in front of the Motion Capture System (MCS). Meanwhile, a Nonlinear Model Predictive Control (NMPC) scheme is proposed based on the robot's kinematic model to follow the reference trajectories while respecting the constraints imposed by the cable lengths and control actions. The simulation results prove the efficiency of the LbD approach in reproducing and generalizing the spatial motions of the robot's tip and avoiding obstacles and external disturbances. On the other hand, the numerical simulations show the performance of NMPC scheme in terms of trajectory tracking and avoiding static and dynamic obstacles. Additionally, its robustness is analyzed by comparing it to the Pseudo-Inverse Jacobian Kinematic Control (PIJKC) while considering the constraints of cable lengths. Finally, the stability of NMPC is evaluated against input perturbations using the Monte Carlo simulations.
Keywords:
Soft continuum robots
Learning by demonstration
Model predictive control
Obstacle avoidance
Monte-Carlo
Journal
J
IF:
2.8
Papers:
3.8K
Citations:
6.9K