Synergistic Prosthetic Control
Synergistic adaptation toward control is my main research interest. Conceptually, it is an artificial intelligence (e.g. neural network) and a natural intelligence (e.g.human) working and learning together to solve problems. This co-adaptation potentially increases overall adaptation to new environments and challenges through. Specifically, the artificial intelligence can adapt (assist) more or less according to the user's current needs and abilities. Unfortunately this is not a free lunch - the costs of co-adaptation include increased system complexity and potential instability. Additionally, both intelligences must share common goals to develop learning synergies. Prior experience has been robot control using neural states. However, the learning paradigm is not intrinsically limited to that particular configuration. Potential applications include:
- brain-machine interfaces for prosthetic control
- functional electrical stimulation for restoration of natural systems (e.g. muscles)
- robot assisted patient rehabilitation
Closed LOop Neural prosthesis for vestibular disorderS (CLONS)
The CLONS project is developing suitable technological solutions to increase the quality of life of people affected by vestibular disorders. Specifically, an innovative closed-loop sensory neural prosthesis is being developed for testing in animal models. Furthermore, these technologies are currently being implemented in a clinical population (humans). This prosthesis restores vestibular information by electrically stimulating nerves connected to the semicircular canals. The stimulation patterns will be based on information provided by inertial sensors mounted to the user's head. This ambitious, inter-disciplinary project has addressed the following challenges:
- intra-vestibular interfaces able to create connections with neurons innervating the vestibular periphery
- implantable circuitry that is powerful and robust yet small and biocompatible
- an artificial vestibular system that can be attached to a user's head
- machine learning to discover effective vestibular stimulation strategies
Brain-Spinal Interfaces (BSI)
Brain-machine interfaces typically aim to replace an aspect of body function by connecting the user's nervous system with external prosthetics. The NEUWalk Project is pursuing brain-spinal interfaces that restore body function by providing a new pathway between the brain and spinal cord which bypasses an injury site. This strategy will exploit existing, functional spinal circuitry and also provide physiological benefits for the patient. Recent spinal cord injury therapy breakthroughs using a novel, multi-pronged neurorehabilation approach showed that rats with completely transected spinal cords can walk again after a combination of traditional physical therapy, pharmacology, and epidural electrical stimulation. We aim to integrate a BSI with neurorehabilitation to facilitate patients gaining voluntary control of therapy and eventually locomotion.
31 May, 2012: The Science paper is PUBLISHED!!! Great story from a great team, really should change the way we think about SCI rehab. Nice web-documentary about it here at www.project-rewalk.com
26 May, 2012: Khoa submitted late breaking research to EMBC - nice way to get accurate template subtraction without the physiological response to the masker.
10 May, 2012: Co-authored five abstracts to SfN, looking forward to getting to the conference again after about 7 years away. See you in New Orleans!
24 April, 2012: Just had the Center for Neuroprosthetics demonstrations for the press and some sponsors.
19 April, 2012: Gave three lectures on my favorite topic of invasive BMI, one day after arriving in my new flat in Lausanne. Went over well
1 March, 2012: Moved to the Center for Neuroprosthetics at the EPFL!! site is a bit out of date - working on it
30 Sept, 2011: New paid position available:
Motion Capture Data Tracking for Biomechanics research.
8 Sept, 2011: Just finished recording the "Diva" cohort of rats. It's an amazing dataset - stay tuned.
9 June, 2011: Both EMBC papers accepted for presentation!
26 February, 2011: Site is more up to date, publications and CV updated. Few new links. Time for a movie
25 February, 2011: The NER paper was accepted! See you in Cancun :)
13 February, 2011: This site is in the middle of a long overdue updating. Getting information up to date, tweaking the style, and getting compliant with standards. It is going to take some more days
3 December, 2010: Gave a talk on state-based BMI to the Bio-Robotics Network Zurich (BiRoNZ) , I still love this stuff. Talk was very well received
19 October, 2010: Gave a lecture on Rehabilitation of Vestibular Function in in the Rehab Engineering II course starting from Silvestro's slides. Talk was good, but no questions
10 Novemeber, 2009: Gave a half-day workshop on Reinforcement Learning at ETH had participants from 4 or 5 different labs. I've been told it was quite good!
9 October, 2009: Gave a half-day workshop on Reinforcement Learning at Scuola Superiore Sant'Anna. The topics ranged from Associative to Q Learning. Hands on programming session to escape a fiery grid-world seemed to be a hit.