An article published in Advanced Intelligent Systems introduces a novel soft tactile sensor designed to enhance the sensory capabilities of robotic grippers. The new sensor, highlighted in “High‐Load Capable Soft Tactile Sensors: Incorporating Magnetorheological Elastomer for Accurate Contact Detection and Classification of Asymmetric Mechanical Components,” employs a multilayered structure and deep learning (DL) model to overcome limitations of existing tactile sensors. By integrating advanced materials and modern technologies, the sensor aims to meet real-world application demands, offering enhanced measurement capabilities and durability.
Design and Composition
The proposed sensor features a sandwich-like multilayered structure, combining a 15 wt% iron magnetorheological elastomer, a flexible printable circuit board with three-dimensional Hall sensors (TLE493D; Infineon), and permanent magnets. This innovative design allows the sensor to be compact yet highly efficient in detecting and classifying contact forces. Furthermore, the incorporation of a multilayer perceptron network enhances the sensor’s ability to classify loading states accurately.
This sensor is capable of performing multiple functions simultaneously: measuring normal forces up to 3.73 kgf, identifying the precise location of force application with 2.5 mm interval subdivisions, and differentiating between wide (approximately 8 mm) and narrow (approximately 2 mm) contact surface areas. Such capabilities make it a significant advancement over traditional tactile sensors, which often suffer from large sizes and limited force measurement ranges.
Experimental Validation
To validate the sensor’s performance, it was experimentally tested by attaching it to a robotic gripper. The experiments involved training a deep learning model to enhance the sensor’s performance, demonstrating its high accuracy, resolution, and extensive measurement range. These tests confirmed that the sensor could effectively improve the functionality and precision of robotic grippers in real-world environments.
Soft tactile sensors are designed to be flexible and are commonly attached to robotic grippers to provide human-like sensory capabilities. However, the development presented in Advanced Intelligent Systems stands out due to its compact design and comprehensive functionality, addressing many of the limitations faced by existing tactile sensors.
Previous tactile sensors often struggled with bulkiness and a narrow force measurement range. The introduction of magnetorheological elastomers and advanced circuit board technologies in the new sensor marks a significant improvement. While past designs were mostly limited in practical applications, this new sensor’s ability to measure forces accurately and classify contact types broadens its usability in various robotics fields.
The innovative use of deep learning models in tactile sensing is not entirely new, but the specific implementation in this sensor, along with its multilayered structure, sets a new benchmark. Previous sensors generally lacked the integrated approach of combining material science with advanced machine learning, which this new design successfully achieves.
The sensor’s potential for improving robotic grippers lies in its high accuracy and large measurement range. By addressing the limitations of traditional sensors, it paves the way for more sophisticated and sensitive robotic applications. Future developments could see further miniaturization and integration, making these sensors even more practical for a wider range of uses. The ability to detect and classify different contact types with high precision will undoubtedly enhance the versatility and efficiency of robotic systems, particularly in fields requiring delicate handling and precise force measurement.
