preliminary test track experiments
DESCRIPTION
Preliminary Test Track Experiments. April 16, 2004 Motohide Hatanaka, Emily Ma Stanford University. Preliminary Test Track Experiments. Goal: Validate simulations and to generate insights to improve on first generation foot design. Focus on three phases of leg trajectory. - PowerPoint PPT PresentationTRANSCRIPT
1
Preliminary Test Track Experiments
April 16, 2004Motohide Hatanaka, Emily Ma
Stanford University
2
Preliminary Test Track Experiments
Goal: Validate simulations and to generate insights to improve on first generation foot design. Focus on three phases of leg trajectory.Attachment Generate strong contact between claws and surface.
• Timing of attachment• Impact velocity• Orientation angle (compensate for motion in thru-
stroke?)
Through-stroke Ankle compliance – how much and for which axes?Detachment Detachment with least resistance and without breaking
claws.
• Orientation angle• Timing of detachment
Preliminary tests focused on timing of attachment and detachment.
3
Test Track Coordinates
force plate
foot
4
Timing of attach/detachment
5
Observations[all cases]The robot is always pushing itself away from the wall. (-Z)
[early attachment]high lateral pull in (+X)robot pushes itself down at initial contact (-Y)robot pushes itself away from surface at initial contact (-Z)[late attachment]small lateral pull in (+X)[late detachment]lateral push out at detachment (-X)significant pull down or drag (-Y)large kick off away from surface (-Z)[early attachment, late detachment]largest pull down (drag) at detachment (-Y)
6
DetachmentEarly Nominal Late
Attach
men
tVe
ry E
arly
Early
Nom
inal
Late
Lateral Force: X-direction (+X = pull in)
High pull-in force (+X) after early attachment
Low pull-in force (+X) after late attachment
Lateral push out (-X) at late detachment
6
optimal
7
DetachmentEarly Nominal Late
Attach
men
tVe
ry E
arly
Early
Nom
inal
Late
Vertical Force: Y-direction (+Y = pull body up)
Robot pushes itself down (-Y) at early attachment
Significant pull down = drag (-Y) at late detachment
7
optimal
8
DetachmentEarly Nominal Late
Attach
men
tVe
ry E
arly
Early
Nom
inal
Late
Normal Force: Z-direction (+Z = pull body into wall)
Large kick off away from wall (-Z) at late detachment
Robot pushes itself away from wall (-Z) especially at early attachment
8
optimal
9
Summary of findingsAssuming we want lateral pull in, vertical self push/pull up, and pull into wall, (i.e. +X,+Y,+Z):
• Early attachment is to be avoided for -Y and -Z but useful for +X.
• Late detachment is to be avoided for all parameters.
10
Next Steps (1)Further analyses with current
setup• Measure or calculate forces and torques at contact point (current contact point is offset from center of the force plate)
• Study the torques• Find out the reasons for the results
obtained (e.g. by closer observation at slower operation rate)
• Look at work generated in a cycle and quantify performance of each setup
11
Next Steps (2)Experiment modification
• Modify the trajectory to have +Z values to measure claw attachment strength on wall
• Include amplitude as an additional parameter for optimization
• Use pull-up motor to emulate the five legs that are not there
• Identify control parameters to match ideal force & torque profiles derived from biology.