Reaching and Grasping

1
Reaching and Grasping
Reaching control synthetic human arm to reach for
object or position in space while possibly
avoiding obstacles
Grasping Control synthetic hand and fingers to
simulate the supporting of an object
Transporting Move object in space, move hand
with object, and use IK to move arm along
2
Reaching human-like
What is human-like?
Experiments Planning happens in Cartesian
space Not joint space Not muslce
space Bell-shaped velocity curves Obstacle
avoidance low curved sections Connected by highly
curved sections
3
Reaching inverse kinematics
Inverse kinematics End-effector hand No sense
of body Hard to enforce joint limits Motions
not necessarily human-like
Most often used to enforce constraints once
end-effector almost in correct position, e.g.,
Foot on ground Hand at object
4
Human reach
IKAN http//hms.upenn.edu/software/ik/ik.html
Use analytic IK whenever possible
Human motion uses pre-defined planar
configurations Hand orientation determines
plane Use wrist to position hand - independent
from arm
5
Human reach
Given body and goal
6
Obstacle Avoidance
Path planning End effector Intermediate links
Goal
Obstacles
7
Obstacle Avoidance
1. Look for global optimal solution
2. Reason from starting position Greedy
algorithm w/backtracking
3. Use precompiled known solutions
8
Obstacle Avoidance
Genetic algorithm Search space of possible
paths Evaluate path Collisions, joint limits,
comfort Minimize end-effect path
length maximum acceleration jerk
9
Obstacle Avoidance
10
Obstacle Avoidance
11
Tool Manipulation
Gravity
Minimize torque?
Orient tool for maximum comfort?
12
Strength Guided Motion
P. Lee, S. Wei, J. Zhao, N. Badler, SIGGRAPH 90
Moving a load to a specified position in space
Strongly influenced by Strength comfort
Objective to find trajectories, both joint and
end-effector, that a human-like linkage would
traverse to complete a task.
13
Strength Guided Motion
P. Lee, S. Wei, J. Zhao, N. Badler, SIGGRAPH 90
Comfort level maximum torque ratio summed over
entire body
Perceived exertion depends on Amount of strength
required (perceived) Amount of strength available
14
Strength Guided Motion
P. Lee, S. Wei, J. Zhao, N. Badler, SIGGRAPH 90
Each DoF has two muscle groups extension and
flexion
Each muscle group strength is modeled as a
function of Body position Anthropometry Gender Han
dedness fatigue Etc.
15
Strength Guided Motion
P. Lee, S. Wei, J. Zhao, N. Badler, SIGGRAPH 90
3 Components Condition monitor - current
state Path Planning Scheme - IK plus
headroom Rate Control Process - determines joint
rates
Motion Strategies Available torque Reducing
moment Pull back Added joint, and jerk
16
Grasping
Hand-object interaction Hand ability for various
configurations Object lends itself to certain
uses (affordances)
Regrasping
Multiple hand manipulation
17
Task Planning
Simulate vision, memory to identify object in
environment
Plan path of figure through environment
Position figure relative to workspace
Use DoFs of figure including arm to reach object
Configure hand relative to object to grasp it
Use strength of figure to manipulate object