Impact of control strategy on trajectory tracking performance in open PLC-based robot controllers: A comparison of PID and model-based control

摘要

This paper explores the impact of control strategy on trajectory tracking within open-architecture, Programmable Logic Controller (PLC)-based robot controllers, where direct access to low-level control loops enables the implementation of custom control schemes. This study experimentally evaluates the performance trade-offs between a standard model-free Proportional-Integral-Derivative (PID) control and a model-based Computed Torque Control (CTC) approach on a SCARA robot, representing prevalent examples of each approach. Trajectory tracking performance and CPU load were assessed across different robot paths, trajectories, velocities, and control frequencies. The goal was to quantify the performance gains achievable with model-based control, assess the associated computational cost and potential benefits of increasing control frequency with a simpler scheme. Results confirm that, at a given PLC frequency, model-based CTC consistently outperforms model-free PID, achieving significantly reduced tracking errors even with increasing velocity, without substantially increasing the PLC's computational load. The factor of reduction in tracking error, indicating the performance gain relative to PID, reached values up to 20 for maximum error and 40 for RMS error, with greater improvements observed for joints exhibiting stronger non-linearities. Furthermore, increasing the PLC frequency with PID control achieved, in some instances, comparable trajectory tracking accuracy as CTC operating at a lower frequency. This demonstrates that increasing the PLC frequency remains a valuable strategy for improving performance while retaining a simple scheme. These results offer insights for optimizing robot control design in open controllers, quantifying the significant performance gains achievable through model-based control and the benefits of increasing PLC frequency in simpler control schemes. • Compares PID and model-based control in open, PLC-based robot controllers. • Quantifies substantial margin of improvement in tracking using model-based control. • Identifies key trade-offs between control complexity and frequency. • Demonstrates that model-based control does not increase computational load.