English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Fast and precise robot motion is needed in certain applications such aselectronic manufacturing, additive manufacturing and assembly. Most industrialrobot motion controllers allow externally commanded motion profile, but thetrajectory tracking performance is affected by the robot dynamics and jointservo controllers which users have no direct access and little information. Theperformance is further compromised by time delays in transmitting the externalcommand as a setpoint to the inner control loop. This paper presents anapproach of combining neural networks and iterative learning control to improvethe trajectory tracking performance for a multi-axis articulated industrialrobot. For a given desired trajectory, the external command is iterativelyrefined using a high fidelity dynamical simulator to compensate for the robotinner loop dynamics. These desired trajectories and the corresponding refinedinput trajectories are then used to train multi-layer neural networks toemulate the dynamical inverse of the nonlinear inner loop dynamics. We showthat with a sufficiently rich training set, the trained neural networks cangeneralize well to trajectories beyond the training set. In applying thetrained neural networks to the physical robot, the tracking performance stillimproves but not as much as in the simulator. We show that transfer learningcan effectively bridge the gap between simulation and the physical robot. Inthe end, we test the trained neural networks on other robot models insimulation and demonstrate the possibility of a general purpose network.Development and evaluation of this methodology is based on the ABB IRB6640-180industrial robot and ABB RobotStudio software packages.

Industrial Robot Trajectory Tracking Using Multi-Layer Neural Networks Trained by Iterative Learning Control