High-Resolution Contact Localization and Three-Axis Force Estimation with a Sparse Strain-Node Tactile Interface...

Area of Science:

• Robotics
• Materials Science
• Sensor Technology

Background:

• High-resolution tactile sensing is vital for human-robot interaction and manipulation.
• Current methods face limitations in sensing area, channel density, and wiring costs.
• Sparse strain readout presents challenges in accurately estimating location and multi-axis forces due to signal coupling and nonlinearity.

Purpose of the Study:

• To develop a cost-effective and scalable sparse tactile interface for precise contact localization and three-axis force estimation.
• To overcome the limitations of dense sensor arrays and extensive calibration procedures.
• To enable robust joint estimation of contact location (x,y) and forces (Fx,Fy,Fz) using a minimal number of strain channels.

Main Methods:

• A sparse strain-node tactile interface device (SSTID) with an optimized three-module layout using particle swarm optimization.
• A strain-node contact-state decoding framework (SCDF) utilizing a lightweight multilayer perceptron.
• A two-stage sim-to-real training strategy involving Finite Element Method (FEM) pretraining and few-shot real-data adaptation.

Main Results:

• The SSTID successfully enabled contact localization (x,y) and three-axis force estimation (Fx,Fy,Fz) with only nine strain channels.
• The SCDF demonstrated accurate contact-state decoding across the full workspace.
• The proposed sim-to-real strategy facilitated effective model training and adaptation.

Conclusions:

• The developed SSTID offers a low-cost and scalable solution for high-resolution tactile sensing.
• The optimized design and decoding framework effectively address challenges in sparse strain readout.
• This approach supports broader implementation of advanced tactile sensing in robotics and manipulation.