Control Of Rotary Inverted Pendulum System Driven By A Servo-table With Uncertain Rod Homing Position.

The rotary inverted pendulum (RIP) is a popular mechatronic system used in control education. Commercial arm-driven RIPs, such as the experimental setup produced by Quanser, are widely used for testing various control algorithms by control engineers. However, some potential common issues in industrial drive systems, such as sensor calibration error, Coulomb friction, stiction, gear backlash, and flexible modes have been hidden by such well-built setups. To better educate students on handling those common issues, a more reliable RIP is built-in-house at Purdue for undergraduate control experiments. In addition to the classical rigid-body-based RIP system dynamics, detailed system identification for the servo-table is performed revealing significant effects of friction and complex flexible modes. A novel model-based 2-loop control design via pole-placement is proposed to address the issues caused by rod angle calibration error, due to uncertain homing position, in the outer-loop, while the friction effect is sufficiently attenuated in the inner-loop. Better performance is observed when implementing the proposed controller in both simulations and experiments, compared with using the full-state feedback (FSF) and linear quadratic regulator (LQR).