I+ I. Eric Eisenstadt, Ph.D. DARPA Defense Sciences Office. Direct Brain-Machine Interface. Science and Technology Symposium April 2004

Transcription

1 ------~~ ~ Direct Brain-Machine Interface Eric Eisenstadt, Ph.D. DARPA Defense Sciences Office Science and Technology Symposium April 2004 I+ I Defence Research and Recherche et developpement Development Canada pour Ia defense Canada Approved lor Public Distribution i\, Case #44239

2 D RPA Accomplish ents Taurus Launch Vehicle

3 D RPA Role in Science and Technologv ~ ~ C) 0-0 c:.s:: 0 ~ Service S& T ~ (I) 0 c: (I) - 0 UJ NEAR MID FAR

4 so RP Role in cience and Technolouv ~ > C) 0-0 c:.c: (J ~ Service S& T ~ (I) (J c: (I) - (J C/) FAR

5 1 ternal Technologv Flow DARPA OFFICES NEAR MID FAR

6 DARPA Strategic Vision DEfENSE SCIENCES OFFICE Strategic Thrusts Detection, precision ID, tracking, and destruction of elusive surface targets Location and characterization of underground structures Force multipliers for urban area operations Networked manned & unmanned attack operations Assured use of space Cognitive systems Bio-Revolution Robust, secure self-forming networks Enduring Foundations Materials Microsystems Maintain the technological superiority of the U.S. military and prevent technological surprise... ttigh:-payoff research that bridges the gap between fundamental discoveries and their military use~ Inform a ti 0 n Tech n 0 I og iesd fi>r Public Distribution A, Ca,:d44239

7 Biology... DARPA's Future Historical Stren Brain Technology Cell & Tissue roteomics Behavior Performance... ''Deploy at "ntain at Enhanced tern Performan "Using Biology to Enhance se Systems" The Biosilico interface Energy Transduction New Materials Tools Biocomputati

8 Human Assisted Neural Devices Use brain activity to command, control, actuate and communicate with the world directly through brain integration with prosthetics and peripheral devices Da1a Acquisitiof1 Bo)(,, ANN -.--, and,/t-~- Realtime Linear., 1 Predictions Client Fit Via Server ::1 - Closed loop demonstration of arm reach and grasp of food ~f \ Visual - LAN Open loop demonstration of human control of gripping force Long-term compatibility Non-invasive correlates \ Feedback \. -:-.:.::::::--._..._ Loop <> \.. Robot Arm + Gripper Approved for Public Distribution A. Ca>~ #44239

9 DEfENSE SCIENCES OFfiCE Human Assisted Neural Devices Learning to Control a Brain-Machine Interface for Reaching and Grasping by Primates 1,4 1, _.. _~.2 id tucd' Jose M. Carmena, Mkhall A. Lebedev, Roy E. Crist, Joseph E. 0 uvn ny, Dav M. San, Dragan F. Dlmltrav 1 ' 3, Parag G. Padu, Craig S. Henriquez,.., Miguel A. L. Nlcolells l,l..t,s- 1 Ot!partment of Neurobiology, Duke Univerl.ity, Durham. North Carolina. United States cj America, l Department of Biomedical Engineering, Duke University. Durham. North Carolina. United States of America. 3 Division of Neurosurgery. Duke University. Durham, North Carolina. United States of America. 4 Center for Neuroengineering, Duke University, Durham, North Carofina, United States of America. S Department of Psychological and Brain Sciences, Duke University, Durham, North Carofina. United States of America ~' Surface EMGs of arm muscles recorded in task 1 for pole control (left) and brain control without arm movements (right). Top plots show the X-coordinate of the cursor. G POLE CONTROl BRAIN OONTROl!_;~.::~M w ~~ Fllfi"ofli >e~~~~ ~~~~±~±d.:~!~--..._ -_-_ -_ -_--_-_--~--~--t1 Plots below display EMGs of wrist flexors, wrist extensors, and biceps. EMG modulations were absent in brain control. ApJ ~ ~f~~4f~~-, _] :::::~: I ~. :~~;:;; : ;~:~ ~,; :' j. ::r ' "::k:: = :.:1 0 ~ ~ ~ 0 ~ ~ ~ Time{$) Tme {$)

10

11

12

13 ~ Example of extending the frontiers in multielement ~~;electrical recording of spatially extended neuronal activity ~-~~m I,J\-,:-~.. :.:."' '~ 11 ij_~ ~~(,'i.i,..';.\' Today: silicon microelectrode arrays for in-vivo probing of brain cortex (J. Donoghue; Brown) Tomorrow: Carbon nanotube arrays with superior spatial resolution endowed by superior electrical/mechanical properties (J. Xu, Brown)

14 CAL TECH Record the intended movement activity from a reach area in the parietal cortex, decode this signal, and use it to move an animated limb on a computer screen, and later a robot limb. Virtual Reach Experiment (X,Y)

15 CAL TECH :/'' ~ ~"'"",~... ~ ~' :?~1~'~, / Using the parietal cortex rather than motor cortex is novel. Useful features of parietal cortex activity are: Detennlne desired reach Generate control signal ~Neural signal Processed neural signal JUUL Control signal Prosthetic Arm System ' " ' \ \ " \ \ ' \ ' \ \ ' \ \ High level (cognitive) and may require fewer recordings to read out intentions. Visual and may show less degeneration or reorganization after spinal cord lesion. Plasticity, making it easier to adapt to the implant Spatially tuned local field potentials (LFP), which are easier to record than single cells.

16 , 450 Biobots: Roborat Electrodes in reward area (medial forebrain and somatosensory cortex) Trained to move forward or turn when medial forebrain is stimulated

17 450 Future activities Non-invasive technologies Sensory feedback Proprioception Integrated and multidisciplinary approach to improved prosthetic devices for amputees