Signals, Instruments, and Systems – W10An Introduction to

Localization and Power Management in Embedded

Systems

1

Outline• Localization

– Motivation– Technique overview– A fundamental technique: odometry

• Power management– Power consumption– Power generation– Power management

2

Motivation for Localization

3

Motivation• Environmental data most often

useless without location information• Sophisticated localization equipment

widely available• → Ubiquity of localization

information

Picture: courtesy of Sensorscope 4

Motivation• Localization is required at different

scales (1000’s km → cm)• Different environments

→ different methodologies

Pictures: courtesy of Sensorscope, NASA, NIST5

Positioning Systems

66

Classification axes

• Indoor vs. outdoor techniques• Absolute vs. relative positioning systems• Line-of-sight vs. obstacle passing/surrounding• Underlying physical principle and channel• Positioning available on-board vs. off-board• Scalability in terms of number of nodes

7

Selected Indoor Positioning Systems

• Multi-camera systems• Infrared (IR) + RF technology• Impulse Radio Ultra Wide Band (IR-UWB)

88

Overhead (Multi-)Camera Systems• Tracking objects with one

(or more) overhead cameras

• Absolute positions, available outside the robot/sensor

• Active, passive,or no markers

• Open source software• Major issues:

light, calibration• E.g. open-source software

SwisTrack (developed at DISAL)

Accuracy ~ 1 cm (2D)

Update rate ~ 20 Hz

# agents ~ 100

Area ~ 10 m2

99

3D Multi-Camera System• 10-50 cameras• mm accuracy• 150-500 Hz update• 6D pose estimation• 4-5 passive IR reflective

markers per object needed• 200-1000 m^3 arena• Various manufacturers (e.g.,

Vicon, MotionAnalysis)ETHZ coordinated ball throwinghttp://www.youtube.com/watch?v=hyGJBV1xnJI

UPenn GRASP labhttp://www.youtube.com/watch?v=geqip_0Vjec

1010

Infrared + Radio Technology• Principle:

– belt of IR emitters (LED) and receivers (photodiode)

– IR LED used as antennas; modulated light (carrier 10.7 MHz), RF chip behind

– Range: measurement of the Received Signal Strength Intensity (RSSI)

– Bearing: signal correlation over multiple receivers

– Measure range & bearing can be coupled with standard RF channel (e.g. 802.11) for heading assessment

– Can also be used for 20 kbit/s IR com channel

– Robot ID communicated with the IR channel (ad hoc protocol)

[Pugh et al., IEEE Trans. on Mechatronics, 2009]

[Pugh et al., IEEE Trans. on Mechatronics, 2009] 1111

Infrared + Radio Technology

Performance summary:• Range: 3.5 m • Update frequency 25 Hz with 10 neighboring robots (or 250 Hz

with 2)• Accuracy range: <7% (MAX), generally decrease 1/d• Accuracy bearing: < 9º (RMS)• LOS method• Possible extension in 3D, larger range (but more power) and better

bearing accuracy with more photodiodes (e.g. Bergbreiter, PhD UCB 2008, dedicated ASIC, up to 15 m, 256 photodiodes, single emitter with conic lense), and faster if only sensing and no com (e.g., Roberts et al, Autonomous Robots, 2012).

1212

IR-UWB Systems• Impulse Radio Ultra-Wide Band• Based on time-of-flight (TDOA, Time

Difference of Arrival)• UWB tags (emitters, a few cm, low-power)

and multiple synchronized receivers• Emitters can be unsynchronized but then

dealing with interferences not trivial (e.g., Ubisense system synchronized)

• Absolute positions available on the receiving system

• Positioning information can be fed back to robots using a standard narrow-bad channel

• 6 - 8 GHz central frequency• Very large bandwidth (>0.5GHz)

→ high material penetrability• Fine time resolution

→ high theoretical ranging accuracy (order of cm)

1313

IR-UWB Systems

• Degraded accuracy performance if – Inter-emitter interferences– Non-Line-of-Sight (NLOS) bias– Multi-path

• Degradation can be compensated with sophisticated collaborative schemes and fingerprinting algorithms

Accuracy 15 cm (3D)

Update rate 34 Hz / tag

# agents ~ 10000

Area ~ 1000 m2

Ex. State-of-art system (e.g., Ubisense 7000 Series, Compact Tag)

1414

Selected Outdoor Positioning Techniques

• GPS• Differential GPS (dGPS)

1515

Global Positioning System

(image from Wikipedia)

1616

Global Positioning System• Initially 24 satellites (including three spares), 32 as of December 2012, orbiting the earth

every 12 hours at a height of 20.190 km.• Satellites synchronize their transmission (location + time stamp) so that signals are

broadcasted at the same time (ground stations updating + atomic clocks on satellites)• Location of any GPS receiver is determined through a time of flight measurement (ns

accuracy!)• Real time update of the exact location of the satellites:

- monitoring the satellites from a number of widely distributed ground stations- a master station analyses all the measurements and transmits the actual position to

each of the satellites• Exact measurement of the time of flight

– the receiver correlates a pseudocode with the same code coming from the satellite– The delay time for best correlation represents the time of flight.– quartz clock on the GPS receivers are not very precise– the range measurement with (at least) four satellites allows to identify the three values

(x, y, z) for the position and the clock correction ΔT• Recent commercial GPS receiver devices allows position accuracies down to a couple

meters with ground stations correction signals and good satellite visibility• 200-300 ms latency, so max 5 Hz GPS updates

17

dGPSPosition Position accuracy: typically from a few to a few tens of cm

1818

Odometry

19

“Using proprioceptive sensory data influenced by the movement of actuators to estimate change in pose over time”

• Start: initial position• Actuators:

– Legs– Wheels– Propeller

• Sensors (proprioceptive):– Wheel encoders (DC motors), step counters (stepper motors)– Inertial measurement units, accelerometers– Nervous systems, neural chains

• Idea: navigating a room with the light turned off

Definition

20

• Example: Cataglyphis desert ant• Excellent study by Prof. R. Wehner

(University of Zuerich, Emeritus)• Individual foraging strategy• Underlying mechanisms

– Internal compass (polarization of sun light) – Dead-reckoning (path integration on neural

chains for leg control; note: in robotics typically using also heading sensors)

– Local search (around 1-2 m from the nest)

• Extremely accurate navigation: averaged error of a few tens of cm over 500 m path!

Example

21

More examples• Human in the dark

• Very bad odometry sensors• dOdometry= O(1/m)

• (Nuclear) Submarine• Very good odometry sensors• dOdometry= O(1/103 km)

• Navigation system in tunneluses dead reckoning based on• Last velocity as measured by GPS• Car’s odometer, compass

Picture: Courtesy of US Navy

Picture: Courtesy of NavNGo22

1D Odometry using an Accellerometer

23

Accelerometer

24

Accelerometer

Microelectromechanical systems25

- Sampling frequency typically a function of the application and of the accelerometer characteristics

e-puck Accellerometer

26

Odometry in 1D

2727

1D Odometry: Error Modeling• Error happens!• Odometry error is cumulative.

→ grows without bound• We need to be aware of it.

→ We need to model odometry error.→ We need to model sensor error.

• Acceleration is random variable Adrawn from “mean-free” Gaussian (“Normal”) distribution.→ Position X is random variable with Gaussian distribution.

2828

1D Odometry with Gaussian Uncertainty

2929

2D Odometry using Wheel Encoders

30

Odometry formalized

1.) Initial position:

• GPS measurement

• Location on map

• Feature

• Measuring/Reading point

31

2.) Proprioceptive sensor (example):

• Measure position (or speed) of the wheels • Principle: pattern on a disc anchored to the motor shaft + optical

encoder• Integrate (count) wheel movements to get an estimate of the

position• Typical resolutions: 64 - 2048 increments per revolution.• For high resolution: interpolation

Odometry formalized

32

IR Ryx

R

R

ξθθ

ξ )(=

=

=

θξ I

I

yx

I

XI

YI

P

XR

YR

xI

yI

θ

From Introduction to Autonomous Mobile Robots, Siegwart R. and Nourbakhsh I. R.

−=

1000cossin0sincos

)( θθθθ

θR

xR

yR

Odometry formalized3.) Coordinate system: inertial vs. relative:

33

),,,,( 21 ϕϕθθ

ξ

rlfyx

I

I

I =

=

XI

YI v(t)

θω(t)

l r (wheel radius)

= wheel i speediϕ2

1

Odometry formalized4.) From wheel speed to forward speed:

34

Recap ME/PHY Fundamentals

v = tangential speedω = rotational speed r = rotation radiusφ = rotation angleC = rotation centerP = peripheral point

C

rrv ϕω ==

r

v

ω

φ

P

Rolling!P’v

r ω

P’= contact point at time t

WheelMotion

t-δt t

35

2221 ϕϕ rrv +=

YI

XI

vθω

l r1

XRYR

lr

lr

2221 ϕϕω −

+=

P

2

Linear velocity = averagewheel speed 1 and 2:

Rotational velocity = sum of rotation speeds (wheel 1 clockwise, wheel 2 counter-clockwise):

Odometry formalized5.) From wheel speed to forward speed:

Forward Kinematic Model

36

−+

+

−=

lr

lr

rr

I

22

022

1000cossin0sincos

21

21

ϕϕ

ϕϕ

θθθθ

ξ

RI R ξθξ )(1−=

0=Ry22

21 ϕϕ

rrvxR +==

lr

lr

R 2221 ϕϕωθ −

+==

YI

XI

vθω

l r1

XRYR

P

2

1.

2.

3.

4.

Odometry formalized6.) From wheel speed to forward speed:

37

• Given an initial position and orientation ξI0, after time T, the position and orientation of the vehicle will be ξI(T)

• ξI(T) computable with wheel speed 1, wheel speed 2, and parameters r and l

dtRdtTT T

RIIII ∫ ∫ −+=+=0 0

1 )()( 00 ξθξξξξ

Odometry formalized7.) From forward speed to position:

−+

+

−=

lr

lr

rr

I

22

022

1000cossin0sincos

21

21

ϕϕ

ϕϕ

θθθθ

ξ

38

dtRdtTT T

RIIII ∫ ∫ −+=+=0 0

1 )()( 00 ξθξξξξ

Odometry formalized8.) Error sources:

−+

+

−=

lr

lr

rr

I

22

022

1000cossin0sincos

21

21

ϕϕ

ϕϕ

θθθθ

ξ

• Wheel slip• Measurement error

1 2,ϕ ϕ∆ ∆

1 2, , , ,r lϕ ϕ θ∆ ∆ ∆ ∆ ∆

1 2, , , ,r lϕ ϕ θ∆ ∆ ∆ ∆ ∆ All error sources affect speed

1 2, , , ,r lϕ ϕ θ∆ ∆ ∆ ∆ ∆

Iξ∆

Iξ∆ The error in the speed causes an error in the position .Iξ∆

Iξ∆ →∞Because of the integral, even a constant error in speed causes an unbounded error in the position

39

Deterministic Error Sources• Limited encoder resolution• Wheel misalignment and small differences in wheel diameter Can be fixed by calibration

40

Non-Deterministic Error Sources• Variation of the contact point of the wheel• Unequal floor contact (e.g., wheel slip, nonplanar surface) Wheels cannot be assumed to perfect roll Measured encoder values do not perfectly reflect the actual motion Pose error is cumulative and incrementally increases

41

Example for a differential-wheel vehicle: Gaussian assumption; ellipses shows 3σ bound

[From Introduction to Autonomous Mobile Robots, Siegwart R. and Nourbakhsh I. R.]

Power Consumption

42

PowerP U I= ⋅

Examples:• MICAz:

2 * 1,5V battery, 25 mA power consumption → 2*1,5V*0,025A=80mW(standby: 80 μW)

• Campell Scientific:3D ultrasonic

anemometer:1,2W or 2,4W

• SHT1x temperature and humidity sensor:2μW – 3mW

43

EnergyE P t= ⋅

Examples:• Rechargeable battery (NiMH):

1,2V*2000mAh=2400mWh=2,4Wh

• Rechargeable battery (LiPo): 3,7V*1340mAh=4958mWh=4,958Wh

• Derived values:(remember 1 Ws=1 Joule)

1999 (Bluetooth

Technology)2004

(150nJ/bit) (5nJ/bit)1.5mW* 50uW

~ 190 MOPS(5pJ/OP)

Computation

Communication

44

Controlling Power ConsumptionConsumption vs. capabilities (example 1):

• Disdrometer 1 (“tipping bucket”) (≈ 0W)- no snow, sleet- no freezing- no drop statistics- resolution

• Disdrometer 2 (laser) (≈ 10W)+ snow, sleet+ freezing+ drop statistics- expensive- delicate

• Disdrometer 3 (hot plate) (≈ 100W)+ snow, sleet+ freezing+ simple- drop statistics

45

Controlling Power ConsumptionConsumption vs. capabilities (example 2):

• Anemometer 1 (cup) (≈ 0W)- 1D- no freezing- temporal resolution (0.3 Hz)- Minimal wind speed high

• Anemometer 2 (ultrasonic) (≈ 10W)+ 2D+ some snow, sleet+ some freezing+ temporal resolution

(up to 60Hz)- expensive

• Anemometer 3 (Anemometer 2+heater) (≈ 100W)+ 2D+ snow, sleet+ freezing+ temporal resolution- expensive 46

Controlling Power ConsumptionConsumption vs. processing speed:

• P ~ fclock• Energy/operation = const

Consumption vs. transmission power:

• P=f(PRF)• sometimes linear: P ~ PRF• often: “sweet spot”

Source: MSP430 data sheet

47

Sensor Node Energy Roadmap

2000 2002 2004

10,000

1,000

100

10

1

.1

Aver

age

Pow

er (m

W)

• Deployed (5W)

• PAC/C Baseline (.5W)

• (50 mW)

(1mW)

Rehosting to Low Power COTS(10x)

-System-On-Chip-Adv Power ManagementAlgorithms (50x)

Source: ISI & DARPA PAC/C Program

48

Communication/Computation Technology Projection

Assume: 10kbit/sec. Radio, 10 m range.

Large cost of communications relative to computation continues

1999 (Bluetooth

Technology)2004

(150nJ/bit) (5nJ/bit)1.5mW* 50uW

~ 190 MOPS(5pJ/OP)

Computation

Communication

Source: ISI & DARPA PAC/C Program

49

Power

• Increased power– higher throughput– higher range– mobile systems: shorter battery life– increased health risk (?)

• Regulation– CH: OFCOM– e.g. WLAN: 100 mW

50

Power

• Unit: W (Watt)– Often written in dBm (decibels to 1 mW)

• Gain / loss: factors– Often written in dB (decibels)

51

PdBm

• 1mW → 10 log(1mW/1mW) → 10 log(1) = 10*0 = 0dBm• 2mW → 10 log(2mW/1mW) → 10 log(2) ≈10*0.3=3dBm• 10mW → 10 log(10mW/1mW) → 10 log(10) = 10*1=10dBm• 100mW → 10 log(100mW/1mW) → 10 log(100) = 10*2=20dBm

=

mW1log10 W

dBmPP )log()log()*log( yxyx +=

52

Link Budget

TX powerTX lossesTX antenna gainFree space path lossRX antenna gainRX losses

RX power

RX sensitivity

100 mW*0.5*1.6*1.0106*10-8

*1.6*0.5

20 dBm-3 dB+2 dB-80 dB+2 dB-3 dB

-85 dBm

-62 dBm

Margin 23 dB200

Typical WLAN link budget (100 m, dipole antennas):

0.00000064 mW

0.000000003 mW

53

Free Space Path Loss (Friis Law)

• Signal power decay in air:

• Proportional to the square of the distance d• Proportional to the square of the frequency f

– high frequency = high loss– low frequency = low bandwidth

54

Ex.: Mica-Z vs. TinyNode

Mica-Z (Crossbow)•Microcontroller:

• ATmega128L• TinyOs

• Transceiver:• Chipcon CC2420• 2.4 GHz carrier• Throughput: up to 250k bps• Range: up to 75 m

TinyNode (Shockfish)• Microcontroller:

• TI MSP430• TinyOs

• Transceiver: • Semtech XE 1205• 868 and 915 MHz carriers• Throughput: up to 153k bps• Range: up to 2 km

Sensorscope data logger

55

Power Generation

56

Power generation methods• Solar• Wind

• Temperature difference(Seebeck Effect)

• Vibration• Hydro

57

Solar power generation

1000W/m2

1000W

Efficiency:10-20%

150W

Efficiency:90%

135W

95W

Efficiency:70%

Efficiency:90%

85W

100W/m2

8,5W

100W 15W 13,5W

9,5W

0,1m2 per W

1m2

58

Power Storage

59

Batteries• primary (non rechargeable, red)

“often a good idea” • secondary (rechargeable, blue)

• Other important parameters:• number of cycles (few 100 – few 1000) optimal conditions• cold temperature behaviour

• Interesting alternative: super capacitor• 30 Wh/kg• expensive• very high number of cycles (> 100’000)

60

Conclusion

61

Take Home Messages• Localization is crucial for associating a spatial anchor to a given

sensor value• Odometry is a well-spread, usually cheap localization

techniques based on proprioceptive sensors• Odometry alone leads to cumulative localization errors and

must be augmented for achieving long-term accurate results• Power is most often the key design constraint in embedded

systems• Efficient power management strategies can decrease the

consumption by several orders of magnitude• Power is comparably difficult to generate and store

62

Additional Literature – Week 10Books• Weston J. and Titterton D, “Strapdown Inertial Navigation”, IET, 2005• Siegwart R., Nourbakhsh I., and Scaramuzza D., “Introduction to

Autonomous Mobile Robots, second Edition”, MIT Press, 2011. • Borenstein J., Everett H. R., and Feng L. “Navigating Mobile Robots:

Systems and Techniques”, A. K. Peters, Ltd., 1996.• Shearer F., “Power Management in Mobile Devices”, Elsevier, 2008.

63