Design of a Biomimetic Tactile Sensor for Material Classification

Dai, Kevin; Wang, Xinyu; Rojas, Allison M.; Harber, Evan; Tian, Yu; Paiva, Nicholas; Gnehm, Joseph; Schindewolf, Evan; Choset, Howie; Webster‐Wood, Victoria A.; Li, Lü

doi:10.1109/icra46639.2022.9811543

Cited by 16 publications

(5 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The sensor performance approaches the performance of rigid piezo-resistive 1D sensors. The sensor remains compact, soft, and 3D-capable—the hallmark features of magnetic force sensors that have fueled the recent advances in robotic tactile sensing in applications such as super-resolution tactile skins [ 18 ] and material classification based on texture perception [ 22 ]. Overall, this work opens the potential for mass-manufacturable compact 3D magnetic force sensors.…”

Section: Discussionmentioning

confidence: 99%

“…For example, Reference [ 21 ] introduced a multimodal sensor (combining piezo-resistive and magnetic) for estimating the contact force direction, location, and joint-level torque. Reference [ 22 ] demonstrated material recognition by engineering grooves in the contact surface, reminiscent of the skin texture.…”

Section: Introductionmentioning

confidence: 99%

“…They can be purchased as accessories for popular cobot grippers (e.g., for the Robotiq 2F-xxx grippers, one of the best-selling gripper families for cobots [ 25 ]). The common sensor chip enabling the advances cited above [ 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 ] is a compact 3D magnetometer: MLX90393 [ 26 ].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Mass-Manufacturable 3D Magnetic Force Sensor for Robotic Grasping and Slip Detection

Signor¹,

Dupré²,

Didden

et al. 2023

Sensors

View full text Add to dashboard Cite

The manipulation of delicate objects remains a key challenge in the development of industrial robotic grippers. Magnetic force sensing solutions, which provide the required sense of touch, have been demonstrated in previous work. The sensors feature a magnet embedded within a deformable elastomer, which is mounted on top of a magnetometer chip. A key drawback of these sensors lies in the manufacturing process, which relies on the manual assembly of the magnet–elastomer transducer, impacting both the repeatability of measurements across sensors and the potential for a cost-effective solution through mass-manufacturing. In this paper, a magnetic force sensor solution is presented with an optimized manufacturing process that will facilitate mass production. The elastomer–magnet transducer was fabricated using injection molding, and the assembly of the transducer unit, on top of the magnetometer chip, was achieved using semiconductor manufacturing techniques. The sensor enables robust differential 3D force sensing within a compact footprint (5 mm × 4.4 mm × 4.6 mm). The measurement repeatability of these sensors was characterized over multiple samples and 300,000 loading cycles. This paper also showcases how the 3D high-speed sensing capabilities of these sensors can enable slip detection in industrial grippers.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Mass-Manufacturable 3D Magnetic Force Sensor for Robotic Grasping and Slip Detection

Signor¹,

Dupré²,

Didden

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Using vibration-based tactile sensing for material or texture classification has been studied extensively in the robotics literature. Many works use extracted frequency information as input to their classification algorithms [2], [8]- [10]. More recent work is able to use deep learning algorithms to learn texture classifiers that operate on time series input [11].…”

Section: Related Workmentioning

confidence: 99%

Fast Image Super-Resolution Based on Multi-Feature Extraction Network

Tang

Meng

2022

2022 4th International Conference on Image, Video and Signal Processing

View full text Add to dashboard Cite

In this work, we use MEMS microphones as vibration sensors to simultaneously classify texture and estimate contact position and velocity. Vibration sensors are an important facet of both human and robotic tactile sensing, providing fast detection of contact and onset of slip. Microphones are an attractive option for implementing vibration sensing as they offer a fast response and can be sampled quickly, are affordable, and occupy a very small footprint. Our prototype sensor uses only a sparse array of distributed MEMS microphones (8-9 mm spacing) embedded under an elastomer. We use transformerbased architectures for data analysis, taking advantage of the microphones' high sampling rate to run our models on timeseries data as opposed to individual snapshots. This approach allows us to obtain 77.3% average accuracy on 4-class texture classification (84.2% when excluding the slowest drag velocity), 1.5 mm median error on contact localization, and 4.5 mm/s median error on contact velocity. We show that the learned texture and localization models are robust to varying velocity and generalize to unseen velocities. We also report that our sensor provides fast contact detection, an important advantage of fast transducers. This investigation illustrates the capabilities one can achieve with a MEMS microphone array alone, leaving valuable sensor real estate available for integration with complementary tactile sensing modalities.

show abstract

“…Multimodal sensory data provides abundant information for decision making and much research effort has been dedicated to developing new sensors for better robot sensing capabilities [7,8,9]. However, somewhat unintuitively, more data modalities do not necessarily promise higher performance for a policy learned from such data.…”

Section: Introductionmentioning

confidence: 99%

Masked Imitation Learning: Discovering Environment-Invariant Modalities in Multimodal Demonstrations

Hao¹,

Wang²,

Cao³

et al. 2022

Preprint

View full text Add to dashboard Cite

Multimodal demonstrations provide robots with an abundance of information to make sense of the world. However, such abundance may not always lead to good performance when it comes to learning sensorimotor control policies from human demonstrations. Extraneous data modalities can lead to state over-specification, where the state contains modalities that are not only useless for decision-making but also can change data distribution across environments. State over-specification leads to issues such as the learned policy not generalizing outside of the training data distribution. In this work, we propose Masked Imitation Learning (MIL) to address state over-specification by selectively using informative modalities. Specifically, we design a masked policy network with a binary mask to block certain modalities. We develop a bi-level optimization algorithm that learns this mask to accurately filter over-specified modalities. We demonstrate empirically that MIL outperforms baseline algorithms in simulated domains including MuJoCo and a robot arm environment using the Robomimic dataset [1], and effectively recovers the environment-invariant modalities on a multimodal dataset collected on a real robot. Our project website presents supplemental details and videos of our results at: https://tinyurl.com/masked-il

show abstract

Design of a Biomimetic Tactile Sensor for Material Classification

Cited by 16 publications

References 37 publications

Mass-Manufacturable 3D Magnetic Force Sensor for Robotic Grasping and Slip Detection

Mass-Manufacturable 3D Magnetic Force Sensor for Robotic Grasping and Slip Detection

Fast Image Super-Resolution Based on Multi-Feature Extraction Network

Masked Imitation Learning: Discovering Environment-Invariant Modalities in Multimodal Demonstrations

Contact Info

Product

Resources

About