Transcript Fleets of inexpensive robots

Teeny-Weeny Hardware Platforms
That Get Up and Walk Around:
Smart Dust, Microrobots, and Macrorobots
Michael Scott, Brett Warneke, Brian Leibowitz, Seth
Hollar, Anita Flynn, Sarah Bergbreiter, and Kris Pister
UC Berkeley
NEST Retreat, Spring 2002
Overview
• Smart Dust
– Concept
– Where we are now
• Microrobots
– Concept
– Where we are now
• Macrorobots
– Concept
– Where we are now
• Conclusions
NEST Retreat, Spring 2002
Overview
• Smart Dust
– Concept
– Where we are now
• Microrobots
– Concept
– Where we are now
• Macrorobots
– Concept
– Where we are now
• Conclusions
NEST Retreat, Spring 2002
Smart Dust - Concept
Interrogating
Laser Beam
Laser
Lens
Mirror
Mirrors
Passive Transmitter with
Corner-Cube Retroreflector
Active Transmitter
with Beam Steering
Sensors
Incoming Laser
Communication
Photodetector and Receiver
Analog I/O, DSP, Control
Power Capacitor
Solar Cell
Thick-Film Battery
1-2mm
NEST Retreat, Spring 2002
Smart Dust - Processes (CMOS)
330µm
TX Drivers
Power input
ambient light
sensor
Photodiode
ADC
Power
Oscillator
13 state
Sensor input
0-100kbps
CCR or diode
FSM
Optical Receiver
controller
1mm
What’s working – Oscillator, FSM, ADC, photosensor, TX drivers
What’s kind of working – Optical receiver (stability problems lead
to occasional false packets)
NEST Retreat, Spring 2002
Smart Dust - Processes (MEMS)
2.8mm
2.1mm
Solar Cells
CCR
Accelerometer
CMOS IC
NEST Retreat, Spring 2002
Smart Dust - Integration
Solar Cell Array
CCR
XL
CMOS
IC
16 mm3 total circumscribed volume
~4.8 mm3 total displaced volume
NEST Retreat, Spring 2002
Smart Dust - Microcontroller
•
•
•
•
•
•
•
8-bit datapath, 12-bit addressing
Dual program and data memories
Load-store RISC-style
32 registers, but most are for special
hardware interfacing
– Five general purpose
– Two autoincrementing
– 16-bit RTC
– Five timers
– Four config/status
Laser reprogrammable
System sleeps most of the time, but
woken up by various timers
Software support
– Assembler
– Cycle-accurate simulator
w/power estimation
– Compiler
•
Common sensor node tasks are
automated by the hardware
– Receiver decodes packets
• Stores new code or data in
memory
• Executes immediate
instructions
• Allows message packets to be
interpreted by the datapath
– Transmitter
• Synchronous or asynchronous
• Can stream a block of
memory
– ADC interface has several modes
ranging from automatically taking
a sample, thresholding it, and
storing it to memory to allowing
full program control
NEST Retreat, Spring 2002
Smart Dust - Instruction Set
•
•
•
•
•
•
•
•
•
•
•
•
•
Move, Move Immediate
Load
Store
Add, add w/carry
Sub, sub w/carry
AND, OR, XOR, complement
Compare, signed and
unsigned
Shift – one bit
Bit set/clear/xor/test
Branch, Always, zero, carry,
no carry
Call – direct and function
pointers
Software Interrupt
Halt
• 1 cpi for all, but no
pipelining
• Addressing modes
– Direct
– Register indexed
• Timers
– Transmit
– Receive – power up, check
for signal
– Sample sensor 1, 2
– Software wake-up
NEST Retreat, Spring 2002
Smart Dust - Peripheral Specs
• ADC
– Consumes 1.8 uW at 10 kS/s (180 pJ/sample 23 pJ/bit)
– 8-bits, “information-on-demand”
• Optical Receiver
– Consumes 26 uW at 375 kbps (69 pJ/bit)
– Receives 50 nW (-43 dBm) optical signals (visible through near IR)
– 1 mrad transmit beam @ 50m req. 10 mW optical transmit power
• CCR Passive Transmitter
– Consumes 350 pW at 175 bps (2 pJ/bit)
– Requires laser interrogation beam which acts as the downlink beam
as well
• Photosensor and 1-axis accelerometer integrated
NEST Retreat, Spring 2002
Overview
• Smart Dust
– Concept
– Where we are now
• Microrobots
– Concept
– Where we are now
• Macrorobots
– Concept
– Where we are now
• Conclusions
NEST Retreat, Spring 2002
Microrobots - Concept
Goal: Make silicon walk.
Motor and Linkages
Chip
Solar Cells,
High Voltage
Chip
NEST Retreat, Spring 2002
•Autonomous
•Articulated
•Size ~ 1-10 mm
•Speed ~ 1mm/s
CMOS
Chip
Microrobots - Processes
•
•
•
CMOS Process
• National’s 0.25 micron 5
metal layer process
High Voltage Electronics and
Solar Cell process
• Fabricated in-house
• Demonstrated Solar Cells
~ 30 Volts
Mechanical Linkages and
Actuators Process –Glass
Reflow Process
NEST Retreat, Spring 2002
Microrobots - Integration
Thinned Solar Cell/High Voltage Chip
Thinned CMOS Chip
Inchworm Motors
Leg
Solar Cells/
High Voltage
CMOS
Assembly
Legs and Motors
Substrate
Wire Bonds
NEST Retreat, Spring 2002
Microrobots - Test Results
NEST Retreat, Spring 2002
Overview
• Smart Dust
– Concept
– Where we are now
• Microrobots
– Concept
– Where we are now
• Macrorobots
– Concept
– Where we are now
• Conclusions
NEST Retreat, Spring 2002
Macrorobots - Concept
Objectives:
• Use off-the-shelf
components to build
inexpensive and modular
autonomous robots
• Take advantage of Rene
motes and TinyOS for
wireless networking and
modularity
Goals:
• Build 50 robots to test
various distributed
algorithms
NEST Retreat, Spring 2002
Macrorobots - Hardware
• Motor-Servo board
interfaces any combination
Motor-Servo Board
of two motors, servos, and
solenoids to a toy car
platform
• Sensor boards are
currently being prototyped,
including a whisker board
(Top)
for obstacle detection and
a digital accelerometer
Whisker-Accel
(ADXL202) board for crude
Board
odometry
• Low-level software
components written to
(Bottom)
abstract hardware
NEST Retreat, Spring 2002
Macrorobots - TOS Components
• Abstract the hardware
from the application
(Ping-Pong)
• Allows use of separate
platforms or sensors
without having to change
code (whisker v. accel,
tank v. car)
• Simple applications
written without worrying
about hardware
NEST Retreat, Spring 2002
Overview
• Smart Dust
– Concept
– Where we are now
• Microrobots
– Concept
– Where we are now
• Macrorobots
– Concept
– Where we are now
• Conclusions
NEST Retreat, Spring 2002
Conclusions
Projected milestones:
• Smart Dust
– Microcontroller (10-bit addressing) – late March 2002
– 12-bit addressing version to follow
• Microrobots
– Integrated with dust mote – December 2002
• Macrorobots
– Several mobile motes – March 2002
– 50 “intelligent” mobile motes – May 2002
NEST Retreat, Spring 2002
Aside - Optical Communications
• Large antenna gain (~1e6)
• Small radiator (mm scale)
• Spatial division multiple access
(SDMA)
• Received power ~1/d2
(vs. ~1/d27 for RF)
• No FCC regulations/right-of-way
constraints
• Rx and Tx can be the same beam
• Output efficiency
2km
– Optical
• Laser slope efficiency
• Poverhead = 1µW-100mW
Laser
CCR
RF
True
Efficiency
Slope
Efficiency
– RF
• GMSK slope efficiency ~50%
• Poverhead = 1-100mW
NEST Retreat, Spring 2002
100nJ
Pout
Poverhead
Pin
Aside - Energy Comparisons
• Bluetooth
– Transmit 1mW for 1ms - 1nJ/bit fundamental Tx cost.
– Actual Tx, Rx power drain ~100mW - 100nJ/bit, 10s
of meters?
• GSM
– Rx power drain= ~200mW  2uJ/bit
– Tx power drain= ~4W  40 uJ/bit, <10km
• Optical (laser)
– 10nJ/bit 1-10km
– 20pJ/bit 0-50m
NEST Retreat, Spring 2002