Trajectory and Energy Optimization Method for Industrial Robot Arm

The multi-axis energy analysis module is crucial in automated factories, especially when used with robotic arms. We developed a hybrid dynamic model that integrates the dynamics of the multi-axis robot arm with machine learning for error correction, considering the angular velocity and acceleration of each joint. This model provides a mathematical representation of the robot's physical behavior, utilizing angles, velocities, accelerations, and loads for each axis. By incorporating an Artificial Neural Network (ANN) with experimental torque values, we estimate the energy consumption for a given trajectory. This hybrid model is used as a surrogate for offline path optimization with the Particle Swarm Optimization (PSO) algorithm, formulated as a bi-objective problem focusing on motion time and energy consumption. The PSO algorithm identifies optimal trajectory, velocity, and acceleration settings for the robotic arm to efficiently perform pick-and-place tasks. Experimental evaluations show that this optimization technique can reduce energy consumption by over 10% while maintaining motion speed.