Biomimetic Robots for Robust Operation in Unstructured Environments

1
Biomimetic Robots for Robust Operation in
Unstructured Environments

• M. Cutkosky and T. KennyStanford University
• R. Full and H. KazerooniU.C. Berkeley
• R. HoweHarvard University
• R. Shadmehr
• Johns Hopkins University

Site visit -- Stanford University, Aug. 9, 2000
Supported by the Office of Naval Research under
grant N00014-98-1-0669
http//cdr.stanford.edu/biomimetics
August 1, 2000
2
Project approach
Low-Level Control
Biomimetic Robots
MURI
Study locomotion in insects, adaptation in
humans.
Study mechanical properties and preflexes in
insects.
Create structures with tailored materials
properties and embedded components.
Shape DepositionManufacturing
2
3
Guiding questions
What passive properties are found in Nature?
Preflexes Muscle and Exoskeleton Impedance
Measurements (Berkeley Bio.)
Low-Level Control
What properties in mechanical design?
Biological implications for Robotics Basic
Compliant Mechanisms for Locomotion
(Stanford) Effects of compliance in joints
(Harvard, Stanford) Fast runner with biomimetic
trajectory (Berkeley ME)
Fabrication
How should properties be varied for changing
tasks, conditions ?
Matching impedance for unstructured dynamic tasks
(Harvard, Johns Hopkins)
4
Low level mapping from passive mechanical
properties of insects to biomimetic robot
structures
Study biological materials, components, and their
roles in locomotion.
Study Shape Deposition Manufacturing (SDM)
materials and components.
viscoelasticmaterial
Hysteresis loop _at_10Hz
stiff material
Models of material behavior and design rules for
creating SDM structures with desired properties
5
Guiding questions
How is Compliance used in Locomotion?
MURI
Low-Level Control
High-Level Control
Berkeley Stanford Measurements of Cockroach
Locomotion
What Compliance Strategies in Human-level Tasks?
Fabrication
Harvard Johns Hopkins Learning and Compliance
Strategies for Unstructured Environments
6
High level results of experiments on human
motion adaptation

• Long-latency feedback adapts to force field,
  through adaptation of the forward model.
• Primitive motions can be combined for complex
  behavior.
• The tool used, a parametric approximator, can
  also be used in model-based control of robots.
• Next step test the approach on a robot -- vary
  walking parameters.

7
Guiding questions
High-Level Control
Low-Level Control
How do we build robust biomimetic structures and
systems?
Shape deposition manufacturing of integrated
parts, with embedded actuators and sensors
(Stanford)
Fabrication
How do we build-in tailored compliance and
damping?
Structures with functionally graded material
properties (Stanford)
Effects of Compliance in simple running machine
(Stanford, Berkeley ME)
8
Fabrication process example creating a robot leg
9
SprawlitaRobust, Dynamic Locomotion with a
Hand-Sized Robot

• Cockroach inspired design
• 0.27Kg mass, 15 cm long
• Robust, dynamic locomotion
• Speed over 2.5 body/sec
• Hip height (3 cm) obstacle traversal
• Shape Deposition Manufactured

First SDM hexapod completed 1.25.2000
Body in mold, half way through fabrication process
Legs with flexures, half waythrough fabrication
process
10
915 - 1030 Low level biological mechanisms

• New results on measurements of muscles,
  exoskeleton, compliance, damping. Comparison with
  artificial muscles (Full et al. 40min)
• Gecko foot adhesion (Liang, Kenny 15 min)
• Discussion of low level mechanisms

1030 1215 High level biological control
and adaptation

• Cockroach locomotion results and implications
  (Full et al. 30 min.)
• New measurement techniques (Bartsch 10 min.)
• Adaptation and impedance matching strategies
  (Howe, Shadmehr 50 min)

11
1245 - 115 Low level robotic mechanisms

• Introduction to fabrication issues (Cutkosky 5
  min)
• SDM robot fabrication overview and results
  (Clark 20 min)

115 - 200 Lab tours and live demonstrations
12
Robot leg fabrication for low-level biomimetic
stabilization
Functional Biomimesis
Shape Deposition Manufactured Robot
Cockroach Geometry
flexure

• Passive Compliant Hip Joint
• Effective Thrusting Force

• Damped, Compliant Hip Flexure
• Embedded Air Piston

13
Sprawl Family History
last years site visit
today
MiniSprawl (power, compliance)
Flexures (bend _at_hip only)
14
Hill climbing adaptation is needed for best
results
Velocity versus slope for different stride
frequencies
60
24 deg.
Frequency 5 Hz
Sprawlita on 24 deg. slope
40
Frequency 11 Hz
20
0
-10
0
10
20
15
200 - 330 High level robotic control

• Introduction to control issues (Cutkosky 2
  min).
• Alternative robot locomotion results and
  implications (Motohide, Kazerooni 30 min)
• Robot locomotion modeling and implications for
  design, control, adaptation (Bailey, Cham 30
  min)
• SDM robot locomotion experiments and ongoing work
  (Cham, Froehlich 20 min.)

16
Are we doomed to succeed?
another Guiding Question
In other words, Is a springy, damped, hexapod
bound to locomote?

• In one sense, yes locomotion will almost
  certainly occur.
• And stable locomotion is not difficult to
  achieve, in practice or in simulation.
• But fast, efficient locomotion is another matter.
  It is quite sensitive to minor changes in
  environmental parameters (e.g. slope, terrain)
  and robot parameters (e.g. leg angles, stride
  frequency, compliance).

per Dan Koditscheks IJRR paper on the theory
behind the one legged hopper
17
Wrap up

• Status
• Programmatic issues
• Plans
• Feedback

18
Status (last site visit 9.2.99)

• Detailed characterization of passive (fixed) and
  active components (adjustable) of preflexes in
  cockroach.
• Gecko foot adhesion characterized using new
  micromachined sensors. New sensor for cockroach
  leg forces being designed.
• SDM environment used to create small robot limbs
  with embedded sensing and actuation and
  functionally graded material properties.
• SDM robot limbs and compliant non-SDM robot
  undergoing testing and comparison with results
  from insect legs.
• Compliant whole-arm-manipulator test-bed and
  minimum impedance control strategies
  demonstrated. Human impedance testing in
  progress.
• Model of human motor control learning tested and
  validated.
• Fast walker with biomimetic foot trajectory
  designed and tested SDM compliant limb retrofit
  underway.

Shape Deposition Manufacturing
19
Status (today 8.9.2000)

• Detailed characterization of passive (fixed)
  components in cockroach and correlations with
  SDM structures
• Detailed characterization of cockroach muscles
  under working conditions and correlations with
  artificial muscle
• Gecko foot adhesion characterized using new
  micromachined sensors. MEMS sensors for insect
  leg forces being tested.
• First small hexapedal robots created using SDM
  -- ahead of schedule
• run at over 2.5 body lengths/second (0.4
  m/second)
• climb belly-height obstacles (3 cm)
• climb slopes to 24 degrees
• exhibit robust, stable locomotion without complex
  feedback
• Models of human motor adaptation and impedance
  regulation tested and validated. Testing on robot
  nearly ready.

Shape Deposition Manufacturing
20
Gecko adhesion in the news...
The article in Nature evidently caught the public
eye

• Scientific American
• ABC today
• In local papers
• and many more...

21
Sprawl robots in the news...
MiniSprawl, from Robosapiens (MIT Press)
22
Plans for next year

• Focus on sensing and adaptation to variations in
  slope, terrain.
• Continue work on insect measurement with new
  sensors
• Continue development of alternative platforms,
  including un-tethered designs (eSprawl).
• Funds permitting design and fabricate a batch of
  Sprawlettes for distribution to members of this
  MURI and to others (e.g., Koditschek) for
  analysis, testing, comparison with animals, and
  validation of control adaptation algorithms.