Implementation of ANN-Based Auto-Adjustable for a Pneumatic Servo System Embedded on FPGA

Area of Science:

• Robotics
• Artificial Intelligence
• Control Systems

Background:

• Pneumatic robot manipulators offer advantages like clean operation and high power-to-weight ratios for industrial use.
• However, pneumatic actuators exhibit non-linear characteristics, posing control challenges.
• Existing methods struggle to effectively manage these non-linearities in complex robotic systems.

Purpose of the Study:

• To propose and evaluate an intelligent controller for a 3-degrees-of-freedom pneumatic robot.
• To minimize the non-linear behavior inherent in pneumatic actuators.
• To enhance the precision and responsiveness of pneumatic robotic systems in industrial settings.

Main Methods:

• An intelligent controller was embedded within a programmable logic device.
• The controller utilized neural networks with three neurons per degree of freedom to adjust controller gains.
• A continuous learning process tuned gain values to minimize mean square error.

Main Results:

• The proposed intelligent controller demonstrated a more appropriate system behavior during transitive time.
• The controller effectively minimized non-linearities in the pneumatic robot's air behavior.
• A minimum mean error of ±1.2 mm was achieved, indicating high precision.

Conclusions:

• The intelligent controller successfully addresses the non-linearities of pneumatic actuators.
• This AI-driven approach enhances the performance and accuracy of pneumatic robot manipulators.
• The findings support the use of intelligent control in industrial applications requiring precise pneumatic actuation.