Robust control of a silicone soft robot using neural networks

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Soft robots exhibit complex elastic behaviors, posing challenges for precise position control.
• Accurate modeling of soft robot dynamics is crucial for effective controller design.
• Existing control methods may struggle with the inherent uncertainties and disturbances in soft robot systems.

Purpose of the Study:

• To design and validate robust controllers for regulating the position of a soft robot.
• To investigate the application of artificial neural networks for modeling soft robot actuator-position relationships.
• To compare the performance of integral and sliding mode robust controllers under various disturbances.

Main Methods:

• Utilized an artificial neural network to model the forward kinematics of the soft robot.
• Developed two robust control strategies: integral control and sliding mode control.
• Analyzed controller performance against constant and time-varying external disturbances.
• Conducted experimental tests to validate the precision, convergence speed, and robustness of the proposed controllers.

Main Results:

• The proposed artificial neural network effectively approximated the soft robot's actuator-position dynamics.
• Both integral and sliding mode robust controllers demonstrated effective position regulation.
• Controllers showed significant robustness against both constant and time-varying disturbances.
• Experimental validation confirmed the high precision and fast convergence speed of the developed control methods.

Conclusions:

• Robust controllers, designed using neural network-based modeling, significantly enhance soft robot position control.
• The proposed methods offer a viable solution for precise and stable operation of elastic soft robots.
• The study validates the effectiveness of advanced control techniques in addressing the complexities of soft robotics.