Bioinspired Robotic Arm Planning by Tau-Jerk Theory and Recurrent Multilayered ANN

Abstract

This work presents a planning model control for a 6-axis robot manipulator simulation assembling task. This work’s purpose is to plan trajectories for locking cable harnesses in palettes using nylon ties. This work is motivated by two biologically inspired approaches. The general 𝜏τ - J erk theory for trajectory tracking and a recurrent bi-layer Hopfield artificial neural networks (HANN) for visual feedback of multiple palette’s elements. Equidistant Cartesian points describing free-collision paths between the robot and target positions are generated. Nonlinear regression-based 3th grade polynomials are obtained by multidimensional least squares as assembling trajectories. The Cartesian paths between robot and target position are chosen based on optimization with derivatives, where the path’s height is a criteria to minimize a route. This work validated the proposed method through computer simulations, which showed feasibility and effectiveness for assembling tasks.