Transcript PCM_Ideas4

Parametric Contact Model (PCM) Development Plan
Milestone
4.1.2.1 – Purchase DE
Date
Software Goal
Experimental Goal
10/1/03
(Dynamics Engine by Arachi)
(11/1/03)
Revised Estimate
4.1.2.3 – Identification of
initial PCM for
development
10/1/03
Have PCM Development Plan
Characterized stage 1 PCM
(10/17/03)
Revised Estimate
4.1.2.5 – PCM design
review
10/31/03
Stage 1 implemented in DE
Characterized stages 2-4
Have experimental goals and
plan in place for gecko, roach,
and robot feet.
(11/7/03)
Revised Estimate
4.1.2.7 – PCM prototype
v0.1 to be exercised by
users
12/2/03
Stage 2 implemented in DE
Test facilities in place and
operating.
4.1.2.10 – PCM v1.0
2/2/04
Stage 3 implemented in DE
First batch of experimental
results on various feet.
Begin matching to Stage 3
parameters.
Stage 1 - Simple Contact
Model Description:
Rigid foot when in contact, free when not
Contact is event driven
Release is time based
Leg has linear and rotational
spring/damper at the foot
Ry
Rx
Model Complexity:
Only 1 PCM parameter – Tr the time of release
Geometry of foot is a simple sphere with appropriate
springs/dampers at the ankle
Questions Model Can Answer:
Measure reaction forces to evaluate leg trajectories and foot compliance,
How much does leg squeezing reduce reaction forces?
Is 2.5kg excessive? How much do we gain/pay for changing mass?
What leg trajectories minimize adhesion forces?
How much adhesion will feet need to provide? And for how long?
Stage 2 - Simple Contact with time/random effects
Model Description:
Rigid foot when in contact, free when not
Contact is state driven with random element
Release is time based or load based
(including time-dependencies)
Model could be extended to handle foot slip,
with no motion until force limit is exceeded,
then planer sliding with simple friction rule
Ry < Limit
Rx < Limit
Friction
Model Complexity:
Additional PCM parameters: Slip force thresholds, time dependence,
% chance of finding/losing a foothold, sliding friction
Geometry of foot is a simple sphere with spring/dampers in leg
Questions Model Can Answer:
Evaluate gait strategies, foot-hold finding strategies, role of redundancy,
Determine if gait is too fast (can’t find a foot-hold) or too slow (begin to slip),
Evaluate how inhomogeneous surfaces affect getting a foot-hold
Stage 3 - Non-trivial Geometry
Model Description:
Foot with multiple toes (claws & sticky pads)
Toes with different contact properties
Compliance between toes
Claw
Pad
Model Complexity:
Additional PCM parameters: Pad friction model, claw adhesion model
Geometry is a set of simple shapes with spring/dampers between
Questions Model Can Answer:
Foot Design strategies: How many toes? What arraignment? How much
compliance between toes? How many claws/pads?
We can begin to match experimental data for claws, setae, prototype feet
Stage 4 - Non-trivial Geometry on Surfaces
Model Description:
Foot with multiple toes (claws & sticky pads)
Toes with different contact properties
Compliance between toes
Details of surface interaction including:
Viscoelastic/plastic impact
Time dependent friction
Statistical surface properties
Claw
Pad
Model Complexity:
Additional PCM parameters: Time dependent adhesion or friction,
Surface deformation properties, More complex pad and claw models,
Velocity dependent impact and friction
Geometry is a set of simple shapes with spring/dampers between
Questions Model Can Answer:
Foot & Behavior designs for finding holds on different surfaces
Feed-forward vs. feed-back foothold finding algorithms.
We can better match experimental data for claws, setae, prototype feet