manipulation in human environments
Post on 12-Jan-2016
24 Views
Preview:
DESCRIPTION
TRANSCRIPT
Edsinger+Kemp NEMS 2006
Manipulation in Human Environments
Aaron Edsinger & Charlie KempHumanoid Robotics Group
MIT CSAIL
Edsinger+Kemp NEMS 2006
Domo•29 DOF
•6 DOF Series Elastic Actuator (SEA) arms
•4 DOF SEA hands
•2 DOF SEA neck
•Active vision head
•Stereo cameras
•Gyroscope
•Sense joint angle + torque
•15 node Linux cluster
Edsinger+Kemp NEMS 2006
Manipulation in Human Environments
● Work with everyday objects● Collaborate with people● Perform useful tasks
Human environments are designed to match our cognitive and physical abilities
Edsinger+Kemp NEMS 2006
Applications
● Aging in place
● Cooperative manufacturing
● Household chores
Edsinger+Kemp NEMS 2006
Three Themes● Let the body do the thinking
● Collaborative manipulation
● Task relevant features
Edsinger+Kemp NEMS 2006
QuickTime™ and a decompressor
are needed to see this picture.
Edsinger+Kemp NEMS 2006
Let the Body do the Thinking
● Design – Passive compliance– Force control– Human morphology
QuickTime™ and aMotion JPEG OpenDML decompressor
are needed to see this picture.
Edsinger+Kemp NEMS 2006
Let the Body do the Thinking
● Compensatory behaviors– Reduce uncertainty– Modulate arm stiffness– Aid perception (motion, visibility)– Test assumptions (explore)
Edsinger+Kemp NEMS 2006
Let the Body Do the Thinking
QuickTime™ and a decompressor
are needed to see this picture.
Edsinger+Kemp NEMS 2006
Collaborative Manipulation
•Complementary actions•Person can simplify perception and action for the robot•Robot can provide intuitive cues for the human
•Requires matching to our social interface
Edsinger+Kemp NEMS 2006
Collaborative ManipulationSocial amplification
Edsinger+Kemp NEMS 2006
Collaborative Manipulation● A third arm:
– Hold a flashlight– Fixture a part
● Extend our physical abilities: – Carry groceries– Open a jar
● Expand our workspace:– Place dishes in a cabinet– Hand a tool – Reach a shelf
Edsinger+Kemp NEMS 2006
Task Relevant Features
● What is important?● What is irrelevant?
*Distinct from object detection/recognition.
Edsinger+Kemp NEMS 2006
Structure In Human Environments
Donald NormanThe Design of Everyday Objects
Edsinger+Kemp NEMS 2006
Structure In Human Environments
● Sense from above● Flat surfaces● Objects for human hands● Objects for use by humans
Human environments are constrained to match our cognitive and physical abilities
Edsinger+Kemp NEMS 2006
Edsinger+Kemp NEMS 2006
Why are tool tips common?
● Single, localized interface to the world● Physical isolation helps avoid irrelevant
contact– Helps perception– Helps control
Edsinger+Kemp NEMS 2006
QuickTime™ and aYUV420 codec decompressor
are needed to see this picture.
Edsinger+Kemp NEMS 2006
Tool Tip Detection
● Visual + motor detection method● Kinematic Estimate● Visual Model
Edsinger+Kemp NEMS 2006
Edsinger+Kemp NEMS 2006
Mean Pixel Error for Automatic and Hand Labelled Tip Detection
Edsinger+Kemp NEMS 2006
Mean Pixel Error for Hand Labeled, Multi-Scale Detector, and Point Detector
Edsinger+Kemp NEMS 2006
Model-Free Insertion
QuickTime™ and a decompressor
are needed to see this picture.
•Active tip perception
•Arm stiffness modulation
•Human interaction
Edsinger+Kemp NEMS 2006
Other Examples
● Circular openings● Handles● Contact Surfaces● Gravity Alignment
Edsinger+Kemp NEMS 2006
Future:Generalize What You've Learned
● Across objects – Perceptually map tasks across objects– Key features map to key features
● Across manipulators – Motor equivalence– Manipulator details may be irrelevant
Edsinger+Kemp NEMS 2006
RSS 2006 WorkshopManipulation for Human Environments
Robotics: Science and Systems University of Pennsylvania , August 19th, 2006
manipulation.csail.mit.edu/rss06
Edsinger+Kemp NEMS 2006
Summary
● Importance of Task Relevant Features
● Example of the tool tip– Large set of hand tools– Robust detection (visual + motor)– Kinematic estimate– Visual model
Edsinger+Kemp NEMS 2006
In Progress
● Perform a variety of tasks– Insertion– Pouring– Brushing
Edsinger+Kemp NEMS 2006
Learning from Demonstration
Edsinger+Kemp NEMS 2006
The Detector Responds To
Fast Motion Convex
Edsinger+Kemp NEMS 2006
Multi-scale Histogram (Medial-Axis, Hough Transform for Circles)
Motion Weighted Edge MapVideo from Eye Camera
Local Maxima
Edsinger+Kemp NEMS 2006
Defining Characteristics● Geometric
– Isolated– Distal– Localized– Convex
● Cultural/Design– Far from natural grasp location– Long distance relative to hand size
Edsinger+Kemp NEMS 2006
Other Task Relevant Features?
Edsinger+Kemp NEMS 2006
Detecting the Tip
Edsinger+Kemp NEMS 2006
Include Scale and Convexity
Edsinger+Kemp NEMS 2006
Distinct Perceptual Problem
● Not object recognition● How should it be used● Distinct methods and features
Edsinger+Kemp NEMS 2006
Edsinger+Kemp NEMS 2006
Edsinger+Kemp NEMS 2006
Use The Hand's Frame
● Combine weak evidence● Rigidly grasped
Edsinger+Kemp NEMS 2006
Edsinger+Kemp NEMS 2006
Acquire a Visual Model
top related