Transcript Computer off the Desk

Computers off the Desk:
Bugs, Hallway Bots, and Teddy Bears
Steven Bathiche and Andy Wilson
Mobile Telepresence, Teddy Bears, and Surface
Computers
Computers off the desk
Dichotomies
• Even our best portable computers are not really portable
• Mobil robots are anemic else not all that mobile
• Design Tradeoffs
– Highly interactive || ultra compact
– Low power || high performance
– Low cost || Powerful
Vision and other sensors
CPU
Sugar
How can we
use this?
MothMobile – Hybrid Robot
Cheating in power consumption and computation for an obstacle
avoidance mobile robot
Moth
Demo
Important Trends
LED flux
CPU Efficiency
Calculations per second
Haitz’s Law of LED Flux
•
LED Flux per package has doubled every 12-24 months for 30+ years.
– (Paul Martin, LumiLeds)
•
Today LED efficiency 60 lm/watt
Calculations per second, Kurzweil, 2001
Meyerson, IBM, 2004
Some Examples: Mobile Telepresence, Teddy
Bears, Surface Computers
• Three emerging trends point to:
– Interactive, big, and rich interfaces yet compact
– low power performance computing
– Decreasing Cost to Performance curve
• Enabling application specific form factors
using Personal Computer technology
PC Based Hallway Bots:
Telepresence, mobile telephones, and meeting recorders
(Bathiche, Ladas 2002)
• A computer that has:
– two drive motors
– Pan tilt camera
– array Microphone
• Served up as controls to a
remote client.
• Army of bots in each
building
• Tele-commuter
knowledge-worker tool
• Attend meetings and go to
people’s offices
• Walk the hallways
(autonomous navigation)
• Demo
Using the .NET Framework
(Bathiche, Ladas 2002)
• .NET framework to manage connections
• XML to communicate interfaces
• SOAP wrapper to expose functions different platforms
Layers
Transmission
Medium
Wireless Ad hoc
Network
Wireless Ad hoc
Network
Corporate Lan
Hardware
Infrastructure
Corporate Lan
IP
Protocol
IP
XML
Data
A/V
Software
Infrastructure
.Net Framework Server (C#)
Connection Manager (C#, VB .Net, C++)
API Robot Layer
API Camera Layer
Robot Drivers
Camera Drivers
Video Drivers
Robot Actuators
Camera Actuators
Video
Application
XML
A/V
.Net Framework Client (C#)
Connection Manager (C#, VB .NET, C++)
Application
Interface
Direct X
Audio Drivers
Driver
Controller Driver
Audio Drivers
Video Drivers
Array Microphone/
Speakers
Physical
Controller Pad
Mic/Speakers
Video
Localization using Sensor Fusion of 802.11 and Odometery
(Krumm and Bathiche, 2004)
3178
3012
3014
3176
3005
AP
3181
3016
3174
3019
3021
3169
3165
3161
3018
3172
3024
3033
3035
3037
3041 3041
3163
3159
3156
AP
3026
3028
3041
3032
3034
3141
3039
3138
Position Uncertainty vs. Time
• Odometer Resets when:
Manual
• Manual Docks
at a Charging Station
Docks at a Charging Station
The robot asks someone
• The robot asks someone
WiFi Localization
Position Uncertainty
•Odometer
Doorways
and
Wifi (geometry)
Resets
when:
Doorways and Wifi (geometry)
Other sensors
vision Other
sensorse.g.
e.g. vision
3154
3141
Odometer Reset
Odometer
Time
Demo
Surface Computers
Blending of Physical and Virtual Interaction
PlayAnywhere
• Lunchbox interactive vision system
• Self-contained, portable, no calibration beyond
factory
PlayAnywhere
• Short-throw projector,
very wide angle lens on
the camera
• Lens distortion &
projective transform
correction for cameraprojector alignment
• Off-axis IR LED illuminant
• Very few assumptions
about the appearance of
the surface
• All calibration is done at
the “factory”
Fast visual codes
• 12-bit visual code based on edge orientation at
center, then blind read of bit values
Paper tracking
• Edge orientation & Hough transform based
detection of rectangular objects
Shadow-based touch
Trackerless manipulation of objects
• Summarize optical flow field as simultaneous
translation, rotation, scaling
Computer off the Desk:
Hallway Bots, Teddy Bears, and Surface Computers
Computer off the Desk:
Bugs, Hallway Bots, and Teddy Bears
Steven Bathiche and Andy Wilson
appendix
MothMobile
• The car uses the insect: the electronics sense
the consequence of the moth’s decisions.
• The insect replaces a computer, vision and
obstacle avoidance algorithms, and camera.
• Hybrid Electronics:
– adapting organic neural network communicating with
a programmable artificial system.
• Insect adapts to the car:
– invert world
• The Neural Chip and Neuromancer
PC Based Hallway Bots:
Telepresence, mobile telephones, and meeting recorders
Scenarios
In a corporate setting each building has a
number of these bots in which any worker
can log on to have a presence in that area.
Tele-commuter knowledge-worker tool
Attend meetings and go to people’s offices
Walk the hallways (autonomous
navigation)
The robot can autonomously walk the
building to advertise your presence to
facilitate hallway discussion. If not
available people can leave messages
on the robot.
Allows you to be in two places at once
The Robots would self charge
Robie
Robie is a robot that has locomotive
actuators, motorized pan and tilt
camera, and an array Microphone
that are served up as controls to a
remote client.
The teleoperator logs on to the robot
server to initiate an audio and video
session, and sequester control of the
robot and camera actuators.