Hardware Implementation for Triaxial Contact-Force Estimation from Stress Tactile Sensor Arrays: An Efficient Design...

Area of Science:

• Robotics and Artificial Intelligence
• Embedded Systems Design
• Sensor Technology

Background:

• Real-time tactile sensing requires efficient algorithms for hardware implementation.
• Existing methods for multiaxial contact-force estimation are often resource-intensive.
• Sparse matrix-vector multiplication offers potential for optimization.

Purpose of the Study:

• To develop and implement a sparse algorithm for tactile sensing on an FPGA.
• To reduce area, power consumption, and data storage for hardware implementation.
• To enable efficient multiaxial contact-force estimation from normal stress tactile sensor arrays.

Main Methods:

• Utilized generalized sparse matrix-vector multiplication for algorithm design.
• Implemented the sparse algorithm on a field-programmable gate-array (FPGA) development platform.
• Employed a high-level description approach for hardware implementation and compared with a non-sparse algorithm.

Main Results:

• The proposed sparse algorithm achieved an average force vector calculation time of 58.68 ms.
• Estimation errors were 12.6% for normal forces and 7.7% for tangential forces on a 10x10 taxel array.
• The hardware implementation demonstrated a 4x reduction in processing elements and required no additional memory.

Conclusions:

• The developed sparse tactile sensing hardware is generalizable, scalable, and efficient.
• This approach significantly reduces resource requirements for tactile sensing systems.
• Enables expanded applications of normal stress sensors in low-power tactile systems.