Force/Velocity Control of a Pneumatic Gantry Robot for Contour Tracking With Neural Network Compensation

Abstract

In this paper, the application of a pneumatic gantry robot to contour tracking is examined. A hybrid controller is structured to control the contact force and the tangential velocity, simultaneously. A previous study provided controller tuning and model validation results for a fixed gain PI-based force/velocity controller. Performance was limited by system lag and Coulomb friction. New results demonstrate that even with perfect friction compensation, the limiting factor is the system lag. A neural network (NN) compensator was subsequently developed to counter both effects. Results for straight and curved edged workpieces are presented to demonstrate the effectiveness of the NN compensator and the capabilities of a pneumatic gantry robot.

Research of a Self-adaptive Robot Impedance Control Method for Robot-Environment Interaction

“…In [18], the designed neural network established the contact force and position tracking control for the robot-environment interaction, with the robot's end-effector position error and its derivative, the contact force error as the inputs and the robot joint driving torque as the outputs, large numbers of inputs result in difficulty for system programs development, in contrast, the proposed neural network in this paper is dedicated for fitting contact force-stiffness relationship consisted of only one input and the rule self-tuned fuzzy controller for adjusting impedance model parameters are convenient for engineering practice. …”

Performance Test of Full Range Suspension Pneumatic Lifting System

“…Kingview monitoring system software can communicate with PLC [6]. It can process, export, store and display data collected by PLC in the working process of pneumatic lifting system.…”

“…Because the signal is a follow-up signal, it is not conducive for the detailed analysis of the dynamic performance of the response process of the system [6][7][8]. This section will analyse the response characteristics of the system by the experiment about charging and discharging of the pneumatic lifting system.…”

Control of a Pneumatic Gantry Robot With Adaptive Neural Network Compensation

“…In previous work, two NN compensators were evaluated by simulation [11]. Tracking performance was shown to improve significantly.…”

“…This paper sets out to validate the previous simulation results by experiment. Specifically, the NN recommended in [11] was implemented as a compensator for a fixed gain PID controller, as applied to the position control of the x-axis of the gantry robot. The NN is considered "adaptive" because it is trained on-line.…”