A Mobile Sensor Droplet for Mapping Hidden Pipeline

69
PipePro be A Mobile Sensor Droplet for Mapping Hidden Pipeline Tsung-te (Ted) Lai Yu-han (Tiffany) Chen Polly Huang Hao-hua Chu National Taiwan University

description

Pipe Probe. Tsung-te (Ted) Lai Yu- han (Tiffany) Chen Polly Huang Hao-hua Chu National Taiwan University. A Mobile Sensor Droplet for Mapping Hidden Pipeline. Outline. Motivation Layout mapping algorithm Design iterations Testbed and evaluation Limitations Related work - PowerPoint PPT Presentation

Transcript of A Mobile Sensor Droplet for Mapping Hidden Pipeline

Page 1: A Mobile Sensor Droplet for Mapping Hidden Pipeline

PipeProbeA Mobile Sensor Droplet for Mapping Hidden Pipeline

Tsung-te (Ted) Lai Yu-han (Tiffany) Chen

Polly Huang Hao-hua Chu

National Taiwan University

Page 2: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Motivation2. Layout mapping algorithm3. Design iterations4. Testbed and evaluation5. Limitations6. Related work7. Future work

Outline

Page 3: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Water scarcity

Page 4: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Series10.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%Toilet

Clothes Washer

ShowerFaucet

Leakage

Other Domestic Bath

Dish Washer

Source: Residential End Uses of Water, AWWA Research Foundation

35 liters/person/day

Residential water usage

Page 5: A Mobile Sensor Droplet for Mapping Hidden Pipeline

ShowerFaucet

x18

Residential water usage

Page 6: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Motivation

hidden pipes

Pipes are often hidden behind walls or underneath floors

Page 7: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Motivationleaking

leaking

Leakage often occurs at the joints of tubes

Page 8: A Mobile Sensor Droplet for Mapping Hidden Pipeline

PipeProbe system‧Map 3D spatial topology of water pipelines‧Mobile sensing approach‧Leverage natural water flow for mobility

Page 9: A Mobile Sensor Droplet for Mapping Hidden Pipeline

ECo wireless sensor mote (Pai Chou, UC Irvine) ‧ Low-power ‧ 13mm(L) x 11mm(W) x 7mm(H), 3 grams ‧ Radio ‧ 3-axis accelerometer

Page 10: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Pressure sensor‧0 – 14 bars, resolution: mbar ‧< 5uA operating current

Page 11: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Gyroscope ‧yaw (z) axis rotation angle ‧ ±300 deg/second

Page 12: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Drop PipeProbe into the main water inlet2. Open a water outlet3. Collect sensor readings from the pressure and gyro sensors4. Analyze the pressure and rotation angle readings

Mapped topology

Page 13: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Open another water outlet to map out the fork path

Page 14: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Gyroscope graph Pressure graph

Page 15: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Motivation2. Layout mapping algorithm3. Design iterations4. Testbed and evaluation5. Limitations6. Related work7. Future work

Outline

Page 16: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Problem formulation

Horizontal layer

Vertical tube

Page 17: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Starting position

Ending position

Problem formulation

Page 18: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Starting position

Ending position

Problem formulation

Page 19: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Assumptions

1. Diameter of pipes is uniform2. Turns are 90-degree

Page 20: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Assumptions

1. Diameter of pipes are uniform2. Turns are 90-degree

Page 21: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Layout mapping algorithm

(2) Conquer

(1) DivideVertical tube

(3)MergeHorizontal

layer

(3)Merge

Page 22: A Mobile Sensor Droplet for Mapping Hidden Pipeline

(2) Conquer

Layout mapping algorithm

(1) DivideVertical tube

(3)Merge Horizontal

layer

(3)Merge

Page 23: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Time

Divide phasepartition pipes into vertical tubes and horizontal layers of tubesuse pressure graph to detect vertical-to-horizontal or horizontal-to-vertical turns.

Page 24: A Mobile Sensor Droplet for Mapping Hidden Pipeline

(2) Conquer

Layout mapping algorithm

(1) DivideVertical tube

(3)MergeHorizontal

layer

(3)Merge

Page 25: A Mobile Sensor Droplet for Mapping Hidden Pipeline

∆height

940

960

980

1000

1020

1040

1060

1080

1100

1120

Time

Pre

ssur

e(m

bar)

∆P =

Conquer phase Estimate vertical tube length Based on pressure principle to estimate vertical tube length

Page 26: A Mobile Sensor Droplet for Mapping Hidden Pipeline

(2) Conquer

(1) DivideVertical tube

(3)MergeHorizontal

layer

(3)Merge

Layout mapping algorithm

Page 27: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Time

Conquer phase Map horizontal pipe layout (1) Detect horizontal turns linking horizontal pipes

based on a change in rotation angles (2) Estimate horizontal tube length

Page 28: A Mobile Sensor Droplet for Mapping Hidden Pipeline

∆ length = ∆t * v

Time

t2t1∆t = t2 –t1

Conquer phase Map horizontal pipe layout (1) Detect horizontal turns linking horizontal pipes

based on a change in rotation angles (2) Estimate horizontal tube length - ∆ length = time * water flow velocity - water flow velocity (constant)

= volume of water outflow / pipe cross-section area~ capsule moving velocity

Page 29: A Mobile Sensor Droplet for Mapping Hidden Pipeline

(2) Conquer

Layout mapping algorithm

(1) DivideVertical tube

(3)MergeHorizontal

layer

(3)Merge

Page 30: A Mobile Sensor Droplet for Mapping Hidden Pipeline

θ

360 degrees of freedom

Merge phase Link vertical pipes to start/end points of each horizontal pipe layout Problem: Vertical-to-horizontal turn angle (θ) is non-deterministic

Page 31: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Θ

Starting position

Ending position

Merge phase How to determine θ?

Page 32: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Motivation2. Layout mapping algorithm3. Design iterations4. Testbed and evaluation5. Limitations6. Related work7. Future work

Outline

Page 33: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1Prototype PressureSensor

Mote

Design: pressure sensor + Eco mote in a round and flat capsule Problem: unstable flow velocity

Page 34: A Mobile Sensor Droplet for Mapping Hidden Pipeline

2Prototype

Design: spherical capsulecapsule flow velocity ≈ water velocity added weight such that PipeProbe’s density ≈ water density

Problem: arbitrary rotation caused unreliable sensor reading

Page 35: A Mobile Sensor Droplet for Mapping Hidden Pipeline

3 PressureSensor

Gyro

Bottom

Prototype

Design: heavy bottom halfpressure sensor on the top, gyro sensor flat on bottom

Problem: arbitrary horizontal spinning caused high noisy gyro reading

Page 36: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Design: tail-like fin aligns capsule’s heading to the water flow direction

4 Final Prototype

Tail-like Fin

Page 37: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Pressure sensor on top and gyro sensor vertical to ground2. Flow velocity ≈ water velocity 3. Flow straight

Gyro graph Pressure graph

Page 38: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Motivation2. Layout mapping algorithm3. Design iterations4. Testbed and evaluation5. Limitations6. Related work7. Future work

Outline

Page 39: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Evaluation metric #1: length error

Actual length: L1

Estimated length:L2

Length error = actual pipe length – estimated pipe length = L1 – L2

Page 40: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Evaluation metric #2: positional error

estimated position (x2, y2, z2) error

actual position(x1, y1, z1)

Positional error (of the pipe turning point) = Euclidean distance between the actual and estimated positions

Page 41: A Mobile Sensor Droplet for Mapping Hidden Pipeline

41Experimental testbed

Page 42: A Mobile Sensor Droplet for Mapping Hidden Pipeline

42Water pipeline testbed

Page 43: A Mobile Sensor Droplet for Mapping Hidden Pipeline

43Control valves to produce different flow paths

Page 44: A Mobile Sensor Droplet for Mapping Hidden Pipeline

44Create a flow path

Page 45: A Mobile Sensor Droplet for Mapping Hidden Pipeline

45

inlet

outlet outlet

Testbed spatial layout (unit: cm)

Page 46: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1 2 3 4 5

6 7 8 9 10

11 12

flow path

Experimental Procedure (12 test scenarios)

Page 47: A Mobile Sensor Droplet for Mapping Hidden Pipeline

pipe probe (2010)Test 11 (flow path in red)

Page 48: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Test 11 (actual flow path)

flow path

Page 49: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Test 11 (1st mapping trip)

flow pathestimates

Page 50: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Test 11 (2nd mapping trip)

flow pathestimates

Page 51: A Mobile Sensor Droplet for Mapping Hidden Pipeline

flow pathestimates

Test 11 (3rd mapping trip)

Page 52: A Mobile Sensor Droplet for Mapping Hidden Pipeline

flow pathestimates

Test 11 (4th mapping trip)

Page 53: A Mobile Sensor Droplet for Mapping Hidden Pipeline

flow pathestimates

Test 11 (5th mapping trip)

Page 54: A Mobile Sensor Droplet for Mapping Hidden Pipeline

flow pathestimates

Test 11 (6th mapping trip)

Page 55: A Mobile Sensor Droplet for Mapping Hidden Pipeline

55

6 mapping trips per test scenarioDataset stats: - 516 length estimates - 588 positional estimates - avg pipe length: 76cm - avg flow distance: 335cm

Mapping trip results (12 test scenarios)

Page 56: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Horizontal length error > vertical length error - Estimation method is different

Overall median error < 2cm; 90 percentile error < 7cm- Precise enough to locate hidden pipes

56

CDF of pipe length errors

Page 57: A Mobile Sensor Droplet for Mapping Hidden Pipeline

57

CDF of positional errorsMedian error < 7cm; 90 percentile error < 16cm

Page 58: A Mobile Sensor Droplet for Mapping Hidden Pipeline

58

accumulation effect

Positional errors vs. flow distance

Page 59: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Introduction2. Layout mapping algorithm3. Design iterations4. Testbed and evaluation5. Limitations6. Related work7. Future work

Outline

Page 60: A Mobile Sensor Droplet for Mapping Hidden Pipeline

60Limitation: uniform pipe diameter

Page 61: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Limitation: capsule size

Page 62: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Motivation2. Layout mapping algorithm3. Design iterations4. Testbed and evaluation5. Limitations6. Related work7. Future work

Outline

Page 63: A Mobile Sensor Droplet for Mapping Hidden Pipeline

NAWMS (SenSys’08)

Page 64: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Detect and localize leakage by pressure and ultrasonic sensors

PipeNet (IPSN’07)

Page 65: A Mobile Sensor Droplet for Mapping Hidden Pipeline

toilet

kitchen sink

shower

HydroSense (Ubicomp’09)Single-point pressure-based sensor of water usage

Page 66: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Multi-pointSensing

Single -pointSensing

Mobile Sensing

NAWMS HydroSense PipeProbe

PipeNet

Comparison to relate work

Page 67: A Mobile Sensor Droplet for Mapping Hidden Pipeline

1. Motivation2. Layout mapping algorithm3. Design iterations4. Testbed and evaluation5. Limitations6. Related work7. Future work

Outline

Page 68: A Mobile Sensor Droplet for Mapping Hidden Pipeline

68

Petrochemical plant

Page 69: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Thank reviewers & shepherd for valuable comments

Questions & Answers

PipeProbe: A Mobile Sensor Droplet for Mapping Hidden Pipeline

Tsung-te (Ted) Lai, Yu-han (Tiffany) ChenPolly Huang, Hao-hua Chu

Ubicomp labhttp://mll.csie.ntu.edu.tw

National Taiwan University