The MIT Leg Lab: From Robots to Rehab

1
(No Transcript)
2
(No Transcript)
3
(No Transcript)
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
The MIT Leg Lab From Robots to Rehab
10
State Of The Art
11
State of the Art Prosthetist defines knee
damping
Otto Bock C-Leg
12
The MIT Knee A Step Towards Autonomy
13
How The MIT Knee Works Mechanism
14
How The MIT Knee WorksSensors

• Knee Position
• Axial Force
• Bending Moment

• Measured Local to Knee Axis (no ankle or foot
  sensors)

Amputee can use vertical shock system
15
How the MIT Knee Works Stance Control
Goal Early Stance Flexion Extension
16
Stance Control Three States

• Stance Flexion Stance Extension
• A variable hydraulic damper
• Damping scales with axial load
• Late Stance
• Minimize damping

Toe-Loading to trigger late-stance zero damping
is automatically adjusted by system
17
Stance Flexion
18
How the MIT Knee Works Swing Control
Goal Control Peak Flexion Angle Terminal Impact
19
Swing Control Flexion
20
Swing Control Flexion
21
Swing Phase Extension

• Extension damping adaptation
• Stage one
• Map tc versus impact force
• Apply appropriate damping
• Stage two
• Control final angle while minimizing impact force

Foot Contact Time
22
The MIT Knee In Action
23
Human Knees Brake and Thrust
Power (W/Kg)
Percent Gait Cycle
24
Human Ankles are Smart Springs
Leg stiffness control in walking and running
humans
Variable stiffness foot-ankle systems
25
Human Ankles are Powered
26
Future of OP Leg Systems Intelligent
Application of Power

• Greater Distance Less Fatigue
• Natural Gait - Dynamic Cosmesis
• Enhanced Stability
• Increased Mobility

27
Human Rehab A Road Map to the Future
Better Power Systems and Actuators
28
Series-Elastic Actuators(Muscle-Tendon)
29
Controlling Force, not Position
Weight 2.5 lbs. Stroke 3 in. Max. Force 300
lbs. Force Bandwidth 30 Hz
30
Biomechatronics Group Hybrid Robots

• Nearly autonomous
• Controllable
• Swam 0.5 body length per second

31
Human Rehab A Road Map to the Future
Improved Walking Models
32
Low Stiffness Control Virtual Model Control
Language

• Passive walkers work using physical components
• Q Can active walker algorithms be expressed
  using physical metaphors?
• A Yes, and they perform surprisingly well

33
Virtual Assistive Devices for Legged Robots
34
Troody
35
Technology
Science
What are the biological models for human walking?
Virtual Model Control
Active OP Leg Systems
36
Human Rehab A Road Map to the Future
Distributed Sensing and Intelligence
37
(No Transcript)
38
Collaborators
Leg Laboratory Gill Pratt Biomechatronics
Group Robert Dennis (UM) Nadia Rosenthal
(MGH) Richard Marsh (NE) Spaulding Gait
Laboratory Casey Kerrigan Pat Riley
39
Sponsors

• Össur
• DARPA
• Schaeffer Foundation

40
Summary
Advances in the science of legged locomotion,
bioactuation, and sensing are necessary to step
towards the next generation of OP leg systems