Sensor-less Angle and Stiffness Control of Antagonistic PAM Actuator Using Reference Set

TL;DR

This paper introduces a sensor-less control method for antagonistic PAM actuators that estimates joint angle and stiffness solely from pressure data using an unscented Kalman filter, enabling independent control without encoders or force sensors.

Contribution

It presents a novel sensor-less control approach for PAM actuators that estimates key parameters with a detailed model and validates it through experiments, expanding control capabilities.

Findings

Successful simultaneous control of angle and stiffness.

The method works without encoders or force sensors.

Reference admissible set guides reference value selection.

Abstract

This paper proposes a simultaneous control method for the angle and stiffness of the joint in an antagonistic pneumatic artificial muscle (PAM) actuator system using only pressure measurements, and clarifies the allowable references for the PAM actuator system. To achieve a sensor-less control, the proposed method estimates the joint angle and contraction forces using an unscented Kalman filter that employs a detailed model of the actuator system. Unlike previous control methods, the proposed method does not require any encoder and force sensor to achieve angle and stiffness control of the PAM actuator system. Experimental validations using three control scenarios confirm that the proposed method can control the joint angle and stiffness simultaneously and independently. Moreover, it is shown that a reference admissible set can be used as an indicator to establish reference values by demonstrating that the reference set covers the experimentally obtained trajectories of the angle and stiffness.