ADOB: A Field-Friendly Control Framework for Reliable Robotic Systems via Complementary Integration of Robust and...

Area of Science:

• Robotics
• Control Systems Engineering
• Mechatronics

Background:

• Robotic systems face challenges with limited computation, environmental uncertainty, and dynamic changes.
• Model-based control is often impractical due to model uncertainty and identification costs.
• Existing robust and adaptive control methods can suffer from interference between disturbance rejection and parameter estimation.

Purpose of the Study:

• To develop an Adaptive Disturbance Observer (ADOB) that integrates disturbance observation and online parameter adaptation.
• To overcome the functional interference issues found in combined Disturbance Observers (DOB) and Parameter Adaptation Algorithms (PAA).
• To enhance the reliability and performance of robotic control systems operating under uncertain conditions.

Main Methods:

• Proposed an Adaptive Disturbance Observer (ADOB) integrating a DOB with a Recursive Least Squares (RLS)-based PAA.
• Implemented a dual-filtering structure to separate disturbance rejection and parameter identification processes.
• Utilized hyperstability theory for stability analysis, incorporating a smoothing mechanism for parameter variation.

Main Results:

• Demonstrated reduced model uncertainty and tracking errors in experimental robotic systems.
• The ADOB showed improved performance compared to conventional DOB methods.
• Successfully separated disturbance compensation and parameter estimation, mitigating mutual masking.

Conclusions:

• The proposed ADOB effectively integrates online parameter adaptation with disturbance observation for robust robotic control.
• The dual-filtering approach ensures stability and improves performance in uncertain dynamic environments.
• The ADOB offers a promising solution for enhancing the reliability of practical robotic systems.