研究目的
To develop a soft and flexible sensor for real-time surface shape reconstruction using fiber Bragg gratings (FBGs) embedded in an elastomeric substrate, overcoming limitations of existing sensors like bulkiness, wiring complexity, and lack of flexibility.
研究成果
The developed sensor successfully reconstructs surface shapes in real-time with high accuracy (RMSe of 1.17 mm for node displacements) and reliability (wavelength fluctuations below 0.01 nm over 1000 cycles). It outperforms electronics-based sensors in terms of simplicity and performance, with potential applications in soft robotics and wearable devices. Future work should focus on optimizing fiber layouts for complex deformations and improving reconstruction speed.
研究不足
The sensor has limited flexibility and stretchability due to the rigidity of the optical fiber, making it unsuitable for very complex deformations. It is sensitive to pressure, which can cause incorrect shape reconstruction, and local temperature changes are difficult to compensate. The learning-based approach requires extensive training data and may have reduced accuracy for distal nodes in the sensor.
1：Experimental Design and Method Selection:
The study uses finite element analysis (FEA) to design and validate the sensor parameters, and machine learning (artificial neural networks) for shape reconstruction from strain data.
2：Sample Selection and Data Sources:
A silicone rubber plate with embedded optical fiber and FBGs is fabricated. Data for training and validation are generated from FEA simulations and experimental setups with infrared tracking cameras.
3：List of Experimental Equipment and Materials:
Optical fiber with FBGs, silicone rubber (Ecoflex 0030, ELASTOSIL E41), optical interrogator, infrared cameras, linear actuators, and computational tools like Abaqus and MATLAB.
4：Experimental Procedures and Operational Workflow:
Fabrication involves molding silicone, embedding fiber with FBGs, and applying protective layers. Experiments include applying deformations, measuring strains with an interrogator, and tracking positions with cameras for ground truth. Data is used to train neural networks for real-time shape reconstruction.
5：Data Analysis Methods:
Strain data from FBGs is processed using wavelength shifts, and neural networks map strains to surface displacements. Accuracy is evaluated using mean-squared error and root-mean-square error compared to ground truth.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
