Hysteresis Compensation for High-Precision Positioning of a Shape Memory Alloy Actuator using Integrated Iterative-Feedforward and Feedback Inputs

In this paper, we investigate the design of an iteration-based control algorithm combined with feedback control to address the positioning error caused by hysteresis effect in a shape memory alloy (SMA) actuator. SMA actuators offer relatively large strain (up to 8%) and high strength-to-weight ratio, and as a result, they are being exploited in minimally invasive surgery tools and microrobotics. But unfortunately, SMA actuators exhibit significant hysteresis effect which can lead to loss in positioning precision; it can cause as much as 50% error in positioning. To address the prohibitive effect of hysteresis, we study the application of an iterative-feedforward algorithm specifically designed to account for this behavior. We show that one of the major challenges with iterative- feedforward inputs for SMA actuators is the slope of the hysteresis curve at the phase transition zones (from martensite to austenite and vice versa) can be significantly large, and thus the output response can be highly sensitive to small changes in the input. As a result, iterative-feedforward input provides limited performance because at the transition zones, small changes in the input (due to noise, for example) cause the output to change significantly, thus inducing tracking error. To alleviate this problem, and to help improve the performance of the feedforward method, we consider the addition of a feedback input to add robustness. We show: (1) experimental results that demonstrate the efficacy of the method and (2) the tracking precision can be reduced by over ten times compared to just using iterative-feedforward input. In particular, the ultimate tracking error was reduced to 0.15% of the total displacement range - approximately the sensor noise level.