Seminar on A Customizable Multimodal Sensing

This seminar will be chaired by A/P Anutosh Chakraborty.

Replicating the human somatosensory system’s complex, distributed sensory network remains a central challenge in advanced prosthetics. Overcoming this challenge is critical to restoring natural tactile feedback and increasing prosthesis adoption. Achieving natural touch requires both accurate tactile sensing and effective haptic stimulation. Specifically, the effectiveness of a functional prosthetic hand depends on high-fidelity, multimodal tactile information. In recent years, there has been a surge in the development of tactile sensors that approach the acuity of the human hand, but many systems still offer limited spatial coverage and resolution. For example, they often sense only the volar surfaces of the fingers and palm, overlooking key regions such as the lateral sidewalls and distal fingertip that support slip detection and precision pinch. In addition, common fabrication approaches (e.g., multilayer lamination and planar microfabrication) can be difficult to conform around joints and tight curvatures. In this talk, we will present a geometry-aware, customizable multimodal sensing system that provides high-resolution pressure and temperature mapping using conformal, free-form modules designed to mimic human skin. We will describe agile fabrication techniques for assembling these multi-module sensors, as well as a neural network-based real-time calibration and denoising strategy to reduce per pixel error and improve signal-to-noise ratio. Furthermore, we will show key results demonstrating up to a 10× improvement in spatial resolution for pressure and temperature sensing.