Development of modular tactile sensor based on optical sensing mechanism

Abstract

As the Internet of Things (IoT), which refers to the hyperconnectedness of people, things, and data, has recently emerged as a keyword in the Fourth Industrial Revolution, tactile sensors have been actively researched for their potential applications in various industries such as human-computer interaction, healthcare, and robotics. Tactile sensors are fabricated in arrays to provide visual and physical feedback to users through pressure distribution information, or to be utilized as electronic skin for robots. However, conventional methods for measuring the information of sensor arrays require multi-channel measurement boards or additional integrated circuit chips such as a large number of amplifiers and switching devices, which increases the volume of the measurement board. In addition, wiring must be considered according to the array shape, which adds to the complexity of the design and creates difficulties in manufacturing to meet user needs. To solve these problems, this study proposes a mechanism that can distinguish information from multiple sensors connected to a microcontroller with a fixed number of channels. By doing so, we developed a modularized tactile sensor that can reconfigure various arrays through inter-sensor connections without degrading the performance of the array. The proposed concept is a sensor system based on an optical mechanism that obtains information through sequential sampling, and for this purpose, sequentially controls light-emitting diodes with a fixed number of lines through addressable light-emitting diodes or shift registers, and utilizes a light sensor to obtain tactile information of the corresponding module. The functional modularity of the proposed mechanism was experimentally characterized. Two tactile sensors were fabricated: a pressure sensor and a force sensor with increased spatial resolution by applying machine learning. The developed pressure sensor has a pressure measurement range of $200 kPa$ and a response time of $0.15 s$ to a step loading. In addition, the response was the asdasdme when the sensors were connected and in single-sensor and multi-sensor loading situations, and the durability of the sensor was confirmed through 1000 repeatability experiments. In the case of the force sensor, a machine learning model was implemented to predict the force applied to locations separated by $2.5 mm$ from light emitting diodes and photodiodes distributed at $2.5 cm$ intervals, and showed a prediction error of $0.51 N$ for the test set. As a result, it showed a spatial resolution of $2.5 mm$ through signal processing from taxels spaced $2.5 cm$ apart. The developed modularized sensor is expected to be used as a customized tactile sensor device in the future.