Force, Humidity, and Temperature Estimation of a Multi-modal Soft Actuator for Human-Pad Interface

Pressure injuries in long-term care facilities present a significant problem for the well-being of bedridden patients and the overall cost of the healthcare systems. Mitigating risks of pressure injury formation might be possible through monitoring and control of the main extrinsic factors that cause them, including temperature, humidity, and normal and shear loads at the skin-support surface interface. An instrumented soft robotic pad system serving as a support surface is a potential solution. In this work, we present the design of two-degree-of-freedom soft actuators that when combined in a grid form an instrumented soft pad. The actuators have integrated humidity sensor, thermistor, and embedded force sensitive resistor (FSR). We investigate the optimal placement of the embedded sensors to monitor temperature, humidity, and applied normal loads during various actuation modes. We utilize a long short-term memory (LSTM) neural network to obtain estimated values of humidity and temperature at the expected contact interface, and also estimates of the normal loads exerted on the soft actuators under various actuation configurations that affect raw FSR sensor measurements. The developed system can be potentially used to monitor and mitigate pressure injuries risks factors in long-term care patients and enhance the quality of care of those patients.