An intelligent calibrator integrating advanced social memory optimization algorithm and improved radial basis...

Area of Science:

• Robotics
• Artificial Intelligence
• Manufacturing Engineering

Background:

• Industrial robots are crucial for intelligent manufacturing, enhancing productivity and precision.
• Long-term operation leads to wear and errors, reducing absolute positioning accuracy.
• This limits the robots' capability for high-precision tasks.

Purpose of the Study:

• To develop an advanced two-stage calibrator for industrial robots.
• To improve absolute positioning accuracy and precision task performance.
• To address errors caused by long-term operation and wear.

Main Methods:

• A two-stage calibration framework combining Advanced Social Memory Optimization (ASMO) and a Gradient-based Particle Swarm Optimization (GPSONN) neural network.
• ASMO identifies kinematic errors using memory-guided global exploration.
• GPSONN compensates residual nonlinear errors via gradient-corrected swarm refinement.

Main Results:

• The ASMO-GPSONN method demonstrated superior calibration accuracy on two datasets, including a real ABB IRB1100 robot.
• Achieved Root Mean Square Error (RMSE) values of 0.43 mm on dataset D1 and 0.47 mm on dataset D2.
• Outperformed compared algorithms in overall calibration accuracy.

Conclusions:

• The proposed ASMO-GPSONN method is highly effective for industrial robot calibration.
• It successfully improves positioning accuracy and compensates for wear-induced errors.
• The approach offers practical effectiveness for enhancing precision tasks in manufacturing.