MURI Kick-off Meeting 1998

1
MURI Kick-off Meeting 1998
Locomotor Performance in Unstructured Environments
Professor Robert J. Full
University of California at Berkeley Department
of Integrative Biology rjfull_at_socrates.berkeley.ed
u http//polypedal.berkeley.edu
2
Control Challenge
Gross
Precise
Control
Repetitive
Novel
Rapid
Slow
Dynamic
Mechanical
Static
(Preflexes)
Neural
Feedforward
Feedforward
Continuous
Feedback
Continuous
Feedback
(Reflexes)
(Reflexes)
3
Rough Terrain
Fractal Surface Variation -
3 times the height of the center of mass
4
EMG Rough Terrain
Flat
Rough
5
Neuro-mechanical Model
aero- , hydro, terra-dynamic
Higher
Sensors
Environment
Centers
Open-loop
Mechanical
Feedforward
Behavior
System
Controller
(CPG)
(Actuators, limbs)
Feedback
Closed-loop
Controller
Adaptive
Sensors
Controller
6
Road Map
1. System Compliance 2. Segment Compliance
3.
Joint Compliance 4. Role of muscle
7
Spring-mass Systems
Legged
SIX-
Legged
EIGHT-
Cockroach
Crab
B
o
d
y
V
e
r
t
i
c
a
l
TWO-
Legged
Legged
FOUR-
W
e
i
g
h
t
F
o
r
c
e
Fore-aft
F
o
r
c
e
T
i
m
e
Blickhan 1989
Human
Dog
8
Leg Stiffness
F
mg
TROTTERS

RUNNERS
HOPPERS
l
100
Blickhan and Full, 1993
Human
Quail
Dog
Cockroach
10
k
rel,leg
Hare
Kangaroo
Crab
1
0.01
0.001
0.1
1
10
100
Mass (kg)
9
Road Map
1. System Compliance 2. Segment Compliance
3.
Joint Compliance 4. Role of muscle
10
Compliant Segments
1. Survivability/robustness 2. Penetrate new
environments
3. Aid in control 4. Energy storage
11
Strain Measurement
12
Segment Loading
13
MURI Interactions
Rapid Prototyping
Stanfor
d
Muscles and
Motor Control
Learning
Locomotion UC Berkeley
Johns Hopkin
s
MURI
Actuators
Manipulation
Legs
Harvar
d
UC Berkele
y
Sensors / MEMS
Stanfor
d
14
Biomimetic Leg
15
Road Map
1. System Compliance 2. Segment Compliance
3.
Joint Compliance 4. Role of muscle
16
Cockroach
17
3D Dynamic Model
Rigid Segments
Raibert - Boston Dynamics Inc.
18
PolyPEDAL Control
T
o
r
q
u
e
s



Feedforward

M
o
t
o
r

p
r
o
g
r
a
m
?
?
Predictive
L
i
m
i
t
e
d
S
t
a
b
i
l
i
t
y
19
PolyPEDAL Control
T
o
r
q
u
e
s



Feedforward Preflexes


M
o
t
o
r

p
r
o
g
r
a
m
I
n
t
r
i
n
s
i
c

m
u
s
c
u
l
o
-
?
s
k
e
l
e
t
a
l

s
y
s
t
e
m
p
r
o
p
e
r
t
i
e
s
Predictive
Z
e
r
o

o
r
d
e
r
L
i
m
i
t
e
d
S
e
l
f
-
s
t
a
b
i
l
i
z
i
n
g
S
t
a
b
i
l
i
t
y
20
3D PolyPEDAL Control
T
o
r
q
u
e
s



Feedforward Preflexes





R
e
f
l
e
x
e
s
M
o
t
o
r

p
r
o
g
r
a
m
I
n
t
r
i
n
s
i
c

m
u
s
c
u
l
o
-
N
e
u
r
a
l
s
k
e
l
e
t
a
l

s
y
s
t
e
m
f
e
e
d
b
a
c
k
p
r
o
p
e
r
t
i
e
s
Predictive
Z
e
r
o

o
r
d
e
r
S
l
o
w

a
c
t
i
n
g
L
i
m
i
t
e
d
S
e
l
f
-
s
t
a
b
i
l
i
z
i
n
g
A
c
t
i
v
e
S
t
a
b
i
l
i
t
y
S
t
a
b
i
l
i
z
a
t
i
o
n
21
MURI Interactions
Rapid Prototyping
Stanfor
d
Muscles and
Motor Control
Learning
Locomotion UC Berkeley
Johns Hopkin
s
MURI
Actuators
Manipulation
Legs
Harvar
d
UC Berkele
y
Sensors / MEMS
Stanfor
d
22
Road Map
1. System Compliance 2. Segment Compliance
3.
Joint Compliance 4. Role of muscle
23
Muscle Function
1. Simulation 2. Direct measurements
3. Capacity
vs. Realized function 4. Perturbation
experiments
24
Insect Advantages
Human
Stimulation (EMG)
Muscle Force
Cockroach
25
Musculo-skeletal Model
Hill Model
N
o
r
m
a
l
i
z
e
d
1
M
u
s
c
l
e
F
o
r
c
e
1
N
o
r
m
a
l
i
z
e
d
F
i
b
e
r
V
e
l
o
c
i
t
y
N
o
r
m
a
l
i
z
e
d
1
M
u
s
c
l
e
F
o
r
c
e
1
N
o
r
m
a
l
i
z
e
d
M
u
s
c
l
e
-
F
i
b
e
r
L
e
n
g
t
h
26
Muscle Lever
Control
Stimulation
Stimulation
- pattern
- magnitude
- phase
Servo and
Strain
Force
- pattern
Transducer
- magnitude
Frequency
27
Workloop Technique
28
Muscle EMG
.
100
extension
flexion
Coxa- Femur Joint Angle
80
60
40
(degrees)
177c
20
EMG
179
0.0
0.05
0.2
0.15
0.1
Time (s)
29
Muscle Force
.
100
extension
flexion
80
Coxa- Femur Joint Angle
60
40
(degrees)
20
177c
179
Relative Muscle Force
0.0
0.05
0.2
0.15
0.1
Time (s)
30
Workloops
Force
(mN)
300
-
200

100
0
9.6
8.8
9.2
9.4
9.8
9.0
Muscle Fiber Length (mm)
31
Muscle Capacity
179
Powerspace
177c
Powerspace
Power
2 Muscle Action Potentials
3 Muscle Action Potentials
(W/kg)
100
0.0
80
60
Stimulation phase ()
-100.0

40

in vivo
in vivo
20
conditions
conditions
-200.0
0
20
4
6
8
10
12
14
5
10
15
Muscle Strain
32
MURI Interactions
Rapid Prototyping
Stanfor
d
Muscles and
Motor Control
Learning
Locomotion UC Berkeley
Johns Hopkin
s
MURI
Actuators
Manipulation
Legs
Harvar
d
UC Berkele
y
Sensors / MEMS
Stanfor
d
33
Conclusions
1. Legged animals behave as spring-mass
systems 2. Segments are compliant - Why?
34
Conclusions
3. Distributed control at joints -
Preflexes 4. Musculo-skeletal units are diverse
35
Leg Controller