A Tandem Robotic Arm Inverse Kinematic Solution Based on an Improved Particle Swarm Algorithm.

Area of Science:

• Robotics
• Artificial Intelligence
• Computational Mechanics

Background:

• Robot inverse kinematics is crucial for control and path planning, with traditional methods facing limitations.
• Intelligent algorithms offer advantages by directly solving forward kinematics equations, reducing computational steps.
• Particle Swarm Algorithm (PSO) is a widely used intelligent algorithm known for its simplicity and performance.

Purpose of the Study:

• To propose an improved Particle Swarm Algorithm (PSO) for solving robot inverse kinematics problems.
• To enhance the search ability of PSO through adaptive weight adjustment and modified velocity factors.
• To utilize an exponential product form (POE) modeling method based on spinor theory for improved kinematic description.

Main Methods:

• Developed an adaptive weight adjustment strategy for PSO to improve global and local search capabilities.
• Introduced a condition setting based on limit joints and a position coefficient k in the velocity factor to optimize running time.
• Employed the exponential product form (POE) modeling method based on spinor theory, contrasting it with the traditional Denavit-Hartenberg (DH) method.

Main Results:

• The improved PSO algorithm demonstrated superior accuracy in position and orientation compared to traditional PSO and Quantum PSO (QPSO).
• Achieved near-zero position error (0) and minimal orientation error (1.29 × 10⁻⁸) for the proposed algorithm.
• Outperformed other algorithms in computation time, with faster and more stable convergence observed across different robotic arm models.

Conclusions:

• The proposed improved PSO algorithm offers significant advantages in accuracy, speed, and convergence for robot inverse kinematics.
• The spinor-based POE modeling method effectively avoids singularities associated with local coordinate systems.
• The algorithm shows high applicability and potential for solving complex multi-arm inverse kinematics problems.