iAPF: an improved artificial potential field framework for asymmetric dual-arm manipulation with real-time inter-arm...

Area of Science:

• Robotics
• Computer Vision
• Control Systems

Background:

• Dual-arm manipulators require sophisticated motion planning for complex tasks.
• Existing methods often struggle with real-time collision avoidance and human-like manipulation.

Purpose of the Study:

• To develop a robust vision-based motion planning framework for dual-arm manipulators.
• To enhance manipulation precision, ensure inter-arm collision avoidance, and enable asymmetric task execution.

Main Methods:

• A hybrid control strategy combining improved Artificial Potential Field (iAPF) for linear velocity and Proportional-Derivative (PD) control for angular velocity.
• A priority-based state machine for human-like asymmetric dual-arm control.
• Real-time vision system (YOLOv8 OBB) and computationally efficient continuous distance calculation for collision monitoring.

Main Results:

• The framework achieves asymptotic convergence to desired configurations, validated by Lyapunov stability analysis.
• Real-time vision system demonstrates high detection accuracy (0.99/0.97) at 20 FPS.
• Comparative tests show smoother trajectories, optimized spatial separation, and effective collision prevention compared to traditional APF.

Conclusions:

• The proposed vision-based framework offers stabilized motion planning for dual-arm robots.
• It significantly improves safety and efficiency in industrial tasks like component sorting.
• The method enables precise, human-like manipulation while preventing collisions.