kawato/Ppdf/Naist 2007.pdfMorimoto J, Endo G, Nakanishi J, Hyon SH, Cheng G: Modulation of simple...

15
Discovery Channel Schaal S, Sternad D, Osu R, Kawato M: Rhythmic arm movement is not discrete. Nature Neuroscience, 7, 1137-1144 (2004). Nakanishi J, Morimoto J, Endo G, Cheng G, Schaal S, Kawato M: Learning from demonstration and adaptation of biped locomotion, J. Robotics and Autonomous Systems, 47, 79-91 (2004). Biped

Transcript of kawato/Ppdf/Naist 2007.pdfMorimoto J, Endo G, Nakanishi J, Hyon SH, Cheng G: Modulation of simple...

Discovery Channel

Schaal S, Sternad D, Osu R, Kawato M: Rhythmic arm movement is not discrete. Nature Neuroscience, 7, 1137-1144 (2004).Nakanishi J, Morimoto J, Endo G, Cheng G, Schaal S, Kawato M: Learning from demonstration and adaptation of biped locomotion,J. Robotics and Autonomous Systems, 47, 79-91 (2004).

Biped

Morimoto J, Endo G, Nakanishi J, Hyon SH, Cheng G: Modulation ofsimple sinusoidal patterns by a coupled oscillator model for bipedwalking. 2006 IEEE International Conference on Robotics andAutomation, ICRA-06, submitted (2005)

By ATR and SONY IDL Collaborative Research

Shibata, T. and Vijayakumar, S. and Conradt, J. and Schaal, S.: Biomimetic OculomotorControl. Adaptive Behavior, 9, 189-208 (2001).

DB 26 Arts

An ATR/SARCOS development with ICORP/JST funding

CB_CPG_WalkMonkey locomotion

Parallel Fiber and Climbing Fiber Inputs to

Purkinje Cells induce

Simple Spikes and Complex Spikes

A

Purkinje cellClimbing

fiber (CF)

Parallel fiber (PF)

Complex spike (CS)

Error signal

Inputsignal

Output signal Error signal

B

Complex spike (CS)

Error signal

Simple spike (SS)

Internal model output

1

2

1

2

1

2

1 = (M2L12

+ 2M2L1S2cos 2 + I1 + I2) ˙̇ 1

+ (M2 L1S2cos2 + I2 ) ˙̇ 2

M2 L1S2(2 ˙1 + ˙

2) ˙2sin 2 + B1

˙1

2 = (M2L1S2cos 2 + I2 ) ˙̇1 + I2˙̇

2

+ M2L1S2˙12sin 2 + B2

˙2

1

2

˙1

˙2

˙̇1

˙̇2

1

2

1

2

˙1

˙2

1

2

1

2

˙1

˙2

Jun Nakanishi, and Stefan Schaal Feedback error learning and nonlinear adaptive control.Neural Networks, 17, 1453-1451 (2004)

desired

E =12( desired ff )T ( desired ff )

d dt = ( ff )T( desired ff )

d dt = ( ff )Tfb

( desired ff )fb

desiredfb

y n xi i

y = i xin

C Cspont

d i dt = xi(C Cspont)

d i dt = ( ff i) fb

= ( ( y) i) fb

= xi (C Cspont)

(1)

(2)

(3)

(4)

SS t = win

t xi t

dwi t dt = xi t {CS t CSspont} wi t

wi t ~ xi t {CS t CSspont}

SS t = xi t {CS t CSspont}n

xi t

~ {CS t CSspont}

Purkinje Cell Activity of Awake Monkeys

during Eye Movement Tasks (OFR)

Eye movements are recorded by

implanted search coil.

Kenji Kawano, Munetaka Shidara,

Hiroaki Gomi, Yasushi Kobayashi,

Aya Takemura, Yuka Inoue

Ocular Following Responses: Reflex eye movement induced by

movement of large visual field

Simple and Complex Spikes of Purkinje Cells

in Monkeys during Ocular Following Responses

Kawano, Shidara, Gomi, Kobayashi, Takemura

-+

+

-

-

+

f t( ) = M ˙ ̇ t +( ) + B ˙ t +( ) + K t +( ) + fbias

A

B

C

F

G

H

Spi

kes/

s

D

Simple spikes (purple) are maximally induced by

ipsiversive or downward stimulus movement.

Complex spikes (blue) are opposite.

Comparing simple spike and complex spike, temporal

pattern is opposite, and firing rate is very different

(100 spikes/sec vs. 1 spike/sec)

Firing rate is related to magnitude of eye movement.

• Temporal profile of simple spike firing and complex spike firing for eachneuron is very similar (just opposite in sign and firing rate is different).

– Complex spike firing may be a template for simple spike firing.

– Simple spike of individual cell is prescribed by complex spike.

��

-+

+

-

-

+�

1. IDM representation

2. Mirror between CS and SS

3. Population coding to rate coding

4. Two opposite axes for CS and SS

5. Full learning simulation

��

Shidara M, Kawano K, Gomi H, Kawato M: Inverse-dynamics model eye movement control by Purkinjecells in the cerebellum. Nature, 365, 50-52 (1993).

Kawato M: Internal models for motor control andtrajectory planning. Current Opinion inNeurobiology, 9, 718-727 (1999).

E 2

E 3

I1

I 2

I 3V1

A1

V2

V3

A2

A3

10000

p 2+300 p +10000

36000p 2

+120 p + 90000

p

p2

r t( )v t( )

p

E1

e0.039 p SLV t( )

e 0.012 pt( )

Tracking arm

movement task

of monkeys

(7cm/sec)

Monkeys moved their

arms under 10 force

fields; 5 viscous and 5

elastic.

Movement kinematics

(position, velocity, etc)

were similar under 10

fields.

Movement

dynamics

(muscle

activities and

force) were

largely

different.

EMG activities of flexor carpi

radialis at each hand position

Simple spike activities

at each hand position

Simple

spike

activities

were very

similar.

Pasalar et al. Nature Neurosci. 2006)

Simple spike activities in cerebellar arm related area (lobule V-

VI) do not encode movement dynamics in slow tracking arm

movements (Pasalar et al., Nature Neuroscience, 2006)

Fast Arm Reaching

under Two Force Fields

flexion and extension of

elbow (max 80cm/sec)

After training

Biceps

muscle

Similar

kinematics

Largely different

dynamics

Triceps

muscle

Yamamoto K, Kawato M, Kotoaska S, Kitazawa S: Encoding of movement dynamics

by Purkinje cell simple spike activity during fast arm movements under resistive and

assistive force fields. Journal of Neurophysiology, 97 1588-1599 (2007)

-> Simple spike activities in 94 of 96 Purkinje cells

were significantly different between two fields.

Temporal firing

frequency of

simple spike

Mutual

information

6 Purkinje cells

Simple spikes: Different

Simple Spikes encode

Movement Dynamics

Simple Spikes

instantaneously switched

Monkeys switched muscle activities at

the first trials when force fields were

switched -> Monkeys should have

acquired internal models for the fields.

Monkeys switched simple spike activities

from the 1st trial after switching of force

fields.

• OFR VPFL

• Winkelman

& Frens (2005)

• Lange E

Sarah-Jayne Blakemore,Daniel M.Wolpert and Christopher D. Frith Abnormalities in theawareness of action. Trends in Cognitive Sciences 6, 237-242, 2002

• •

1 = 6( )1 = 12( )1 = 12( )

11 1

2 2 2

-

-

-

--

-

1

2

3

4

1

2

3

4

ˆ 1

ˆ 2

ˆ 3

ˆ 4

A

mcA xAmcB xB

Q

B,mc

mcB = mcA +(1 )mcA*

0

,mc

• (VOR)

• (OFR)

VS

MST , ,

:MST 500ms

MST

p q

xy

p

q

x

y

120

=q

p

y

x

cossin

sincos

x

y

=

p

q

Imamizu H, Miyauchi S, Tamada T, Sasaki Y, Takino R, Puetz B, Yoshioka T, Kawato M: Human cerebellar activity reflecting an acquired internal model of a new tool. Nature 403 192-195(2000)

X (mm)

Y (mm)

Right

Left

Anterior

Posterior

Z (mm)

Superior

Inferior

Organization of 16 Daily Tools in

the Right Superior Cerebellum

16 tools and utensils:Chopsticks, saw, scissors,pencil, hammer, screw-driver,fork, spoon, tooth brush,brush, comb, cutting pliers,monkey wrench, wrench,knife and clip

MRI

Mirror

Medial lateral

Ant.

Pos.

•Averaged across 8 subjects with fixed effect

model p<0.05 corrected.

•The peaks are projected in a horizontal plane

Conditions

•Tool-use execution

•Hold the tool and look at the object

control

•Tool-use motor imagery

Higuchi et al. Cortex (2007)

Higuchi et al. Cortex (2007)

Language and Tool Internal Models in

Broca’s Area: Segregation and Overlap

• 28 healthy subjects ( 14males and 14 females)

• Conditions

– Tool-use execution – Hold the tool and look

at the object

control – Tool-use motor

imagery– Story listening– Reversed story

listening (control)

Typical subject

Random effect

analysis

Higuchi, Chaminade,

Imamizu, Kawato

(in preparation)

MT :

Newsome et al., (1988)

MST :

:

Kawano et al., (1994) Kawano et al., (1992)

MSTd:

A

B

C

FL

VN

FLVPFL

VN

MST

FLVPFL

VN

P

Vhead eye

P

head eye

vision

P

Vhead eye

P

Vhead eye

P

Vhead eye

vision

P

Vhead eye

vision MST

Vhead eye

P

Vhead eye

vision MSTMST

A B C D

E F G HH E H

G G

E

EEH H

G G

+ -

+

+

-

-

Figure 2

P

.

.

.

.

.

.

.

. .

.

.

.

Kt

Ft

Ht

It+

-

+

+

Pt (r, V)ˆ+

e

f

e

+

-50ms

MST

MST