Neuromorphic Twins for Bioelectronic Medicine
Bioelectronic medicine treats chronic diseases by sensing, processing, and modulating the electronic signals produced in the nervous system. While electronic circuits have been used for several years in this domain, the progress in microelectronic technology is now allowing increasingly accurate and targeted solutions for therapeutic benefits. To fully exploit this approach it is crucial to understand what aspects of the nerve signals are important, what is the effect of the stimulation, and what circuit designs can best achieve the desired result. Neuromorphic electronic circuits represent a promising design style for achieving this goal: their ultra-low power characteristics and biologically plausible time constants make them the ideal candidate for building optimal interfaces to real neural processing systems, enabling real-time closed-loop interactions with the biological tissue. Here I will highlight the main features of neuromorphic circuits that are ideally suited for interfacing with the nervous system and show how they can be used to build closed-loop hybrid artificial and biological neural processing systems, also showing examples of applications that follow this approach.
Towards Flexible, Printable Neuromorphic Perception for Soft Robotics
Human-like, next generation soft-robotics manipulators will require all of the following: (1) physically soft and flexible, (2) multi-functional sensing (e.g. pressure, temperature, etc.), (3) fabrication integrated into the structure (instead of separately fabricated sensors and structures which are then manually integrated), (4) with cognition (understanding) of the sensing data.
We will present the progress of our efforts to accomplish all of that. We show the use of electrically active polymeric materials, used to fabricate physically soft and flexible pressure and temperature sensors, fabricated using additive manufacturing (so-called 3D printing) techniques. The sensing-to-perception is accomplished via biologically inspired neuromorphic data processing. Our current efforts concentrate on integration of all of the aforementioned in soft finger-like manipulators intended to use in soft robotics.
Neuromorphic Control
Motivated by the growing availability of event-based sensors and event-based actuators, we ask the question of how to close the loop from events to events. We propose to address those design questions through a mixed-feedback methodological framework inspired from the architecture of biophysical neural circuits. This architecture suggests that much of the classical theory of feedback control can be leveraged to designing event-based controllers that combine the reliability of the digital with the tuneability of the analog.
Biomimetic artificial touch for brain-controlled bionic hands
Somatosensory feedback from the hand is essential to support object interactions as evidenced by the deficits observed when it absent. With this in mind, we are developing approaches to restore tactile feedback for brain-controlled bionic hands via intracortical microstimulation (ICMS) delivered to somatosensory cortex (S1). To this end, we build sensory encoding algorithms that convert the output of sensors on the hand into patterns of ICMS delivered to S1. We show that biomimetic algorithms, which attempt to reproduce natural patterns of S1 activity, evoke sensations that are more natural, more informative about interaction forces, and less disruptive to decoders of intended movements from signals in motor cortex.
