Mutual-Collision-Avoidance Scheme Synthesized by Neural Networks for Dual Redundant Robot Manipulators Executing...

Area of Science:

• Robotics
• Artificial Intelligence
• Control Systems

Background:

• Collisions between dual robot manipulators can cause task failure and damage.
• Existing motion planning methods may not adequately address real-time collision avoidance in collaborative tasks.

Purpose of the Study:

• To propose and exploit a novel recurrent neural network (RNN)-based mutual-collision-avoidance (MCA) scheme for dual robot manipulators.
• To ensure safe and efficient collaboration between dual robot arms during complex tasks.

Main Methods:

• A recurrent neural network (RNN)-based mutual-collision-avoidance (MCA) scheme.
• Linear variational inequality-based primal-dual neural network for solving the scheme.
• Line-segment-based distance measure algorithm with speed brake control.
• Formulation as a standard quadratic programming problem solved by an RNN.

Main Results:

• The proposed RNN-based MCA scheme effectively avoids collisions in dual robot manipulators during trajectory tracking and cup-stacking tasks.
• The system demonstrates effectiveness, accuracy, and physical realizability through simulations and experiments.
• Dual manipulators learn MCA capabilities through network iteration and online learning.

Conclusions:

• The developed RNN-based MCA scheme is a viable solution for preventing collisions in dual robot manipulator systems.
• The approach enhances safety and reliability in collaborative robotic applications.
• The method is validated for cooperative end-effector tasks, showing practical applicability.