Physics-Based Simulation and Machine Learning for the Practical Implementation of EIT-Based Tactile Sensors

Biasi, Niccoló; Gargano, Andrea; Arcarisi, Lucia; Carbonaro, Nicola; Tognetti, Alessandro

doi:10.1109/jsen.2022.3144038

Cited by 10 publications

(3 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, tactile materials can be non-uniform and anisotropic, and the consequences of that on the choice of I-M pattern are unknown. These issues can be tackled by combining physics-based modeling with EIT to handle such effects as irregular geometries of sensors and anisotropic conductivity, as shown in [33]. Noise that depends on electrode position or electrode pair separation may strongly affect the noise performance of some patterns.…”

Section: Discussionmentioning

confidence: 99%

EIT for tactile sensing:considerations regarding the injection-measurement pattern

Smela

2022

Eng. Res. Express

View full text Add to dashboard Cite

Bipolar current injection and voltage measurement (I-M) patterns are frequently used in electrical impedance tomography (EIT) for tactile sensing. In this work, the total set of 36 unique combinations for 16-electrode systems is investigated using simulations. Performance is evaluated on a circular sensor as a function of hyperparameter and target position with respect to critical performance measures for tactile sensing; these include not only peak amplitude and resolution, but also susceptibility to noise and, importantly, the uniformity of performance over the sensor area. The determination of which pattern to employ can therefore be based on the needs of the particular application. The relative performance of the I-M patterns is determined at small hyperparameters by electrode placement symmetry, but at large hyperparameters by sensitivity at the center of the sensor. Patterns with high spatial symmetry should be avoided; these include electrode pairs on opposite sides of the sensor. Patterns with electrodes in adjacent positions, which have been the norm for tactile sensing, should also not generally be used. If performance on all metrics across a wide range of hyperparameters is needed, then placing both the injection and measurement electrodes 3 spaces apart (the 3-3 pattern) can be a good strategy. The use of 13 electrodes instead of 16 is also examined. The absence of symmetry provides greater flexibility in the choice of I-M pattern, and the loss in performance may be small; furthermore, the reduction in data collection time may be advantageous. Beyond avoiding the worst I-M patterns, the most important measure to take is decreasing noise on the signal to permit the use of a smaller hyperparameter; this is of greater importance than selecting from among the remaining I-M patterns.

show abstract

Section: Discussionmentioning

confidence: 99%

EIT for tactile sensing:considerations regarding the injection-measurement pattern

Smela

2022

Eng. Res. Express

View full text Add to dashboard Cite

show abstract

“…Second, neural network (NN) reconstruction may contribute to improved classification accuracy using reconstructed images. NN reconstruction has been reported previously [28], [40], [42]. This approach has been documented to enhance spatial resolution, sensitivity, and localization accuracy.…”

Section: Classification Of Reconstructed Imagesmentioning

confidence: 97%

Development of Pinching Motion Classification Method Using EIT-Based Tactile Sensor

Asahi,

Yoshimoto,

Sato

2024

IEEE Access

View full text Add to dashboard Cite

Fine motor skills have been suggested to be related to human cognitive abilities. To develop an objective method for evaluating fine motor skills, we applied a flexible tactile sensor based on electrical impedance tomography (EIT) and the contact resistance principle to a cylinder designed to mimic the peg used in the Functional Dexterity Test. Six pinching motions were classified to confirm the feasibility of the prototype system. Two types of classification were performed: classification using reconstructed images and classification using measured voltage vectors. The feasibility of the classification method was evaluated using adult subjects, and it was demonstrated that the system can accurately classify various types of pinching motions. The results revealed that utilizing reconstructed images for classification achieved a discrimination accuracy of 79.4%, while employing measured voltage vectors for classification resulted in a discrimination accuracy of 91.4%. These findings underscore the potential for developing an automated finger motion analysis system using EIT-based tactile sensor.

show abstract

“…To improve low spatial resolution, novel reconstruction algorithms and sensor structures have been proposed. Neural network approaches have been proposed to improve the spatial resolution ( Biasi et al., 2022 ) and efficiency of sensing ( Park et al., 2021 ). Adding internal electrodes can improve spatial resolution in the central area; however, there is concern regarding the low spatial resolution at the boundary due to insufficient current flow through the area.…”

Section: Introductionmentioning

confidence: 99%

Optimization of electrode positions for equalizing local spatial performance of a tomographic tactile sensor

2023

View full text Add to dashboard Cite

A tomographic tactile sensor based on the contact resistance of conductors is a high sensitive pressure distribution imaging method and has advantages on the flexibility and scalability of device. While the addition of internal electrodes improves the sensor’s spatial resolution, there still remain variations in resolution that depend on the contact position. In this study, we propose an optimization algorithm for electrode positions that improves entire spatial resolution by compensating for local variations in spatial resolution. Simulation results for sensors with 16 or 64 electrodes show that the proposed algorithm improves performance to 0.81 times and 0.93 times in the worst spatial resolution region of the detection area compared to equally spaced grid electrodes. The proposed methods enable tomographic tactile sensors to detect contact pressure distribution more accurately than the conventional methods, providing high-performance tactile sensing for many applications.

show abstract

Physics-Based Simulation and Machine Learning for the Practical Implementation of EIT-Based Tactile Sensors

Cited by 10 publications

References 47 publications

EIT for tactile sensing:considerations regarding the injection-measurement pattern

EIT for tactile sensing:considerations regarding the injection-measurement pattern

Development of Pinching Motion Classification Method Using EIT-Based Tactile Sensor

Optimization of electrode positions for equalizing local spatial performance of a tomographic tactile sensor

Contact Info

Product

Resources

About