A Lo calization T echnique bots F r o M ulti-Agent Prateek - - PowerPoint PPT Presentation

a lo calization t echnique bots f r o m ulti agent
SMART_READER_LITE
LIVE PREVIEW

A Lo calization T echnique bots F r o M ulti-Agent Prateek - - PowerPoint PPT Presentation

A Lo calization T echnique bots F r o M ulti-Agent Prateek Humane and Neelay Trivedi R obot F ormations What Di Did We Do Do? Indoor Localization System Arduino-Based Sensor Module Polygonal Formation Algorithm HARD RDWARE RE


slide-1
SLIDE 1

Lo T F M R F A calization echnique r

  • ulti-Agent
  • bot
  • rmations

bots

Prateek Humane and Neelay Trivedi

slide-2
SLIDE 2

What Di Did We Do Do?

Indoor Localization System Arduino-Based Sensor Module Polygonal Formation Algorithm

slide-3
SLIDE 3
slide-4
SLIDE 4

HARD RDWARE RE

slide-5
SLIDE 5

Sen ensor M Module O e Over erview

Low-Cost

sensor module under $50

Model System

new sensors = easy to add

Flexible

cross-environment capable

Co Compass S ss Senso sor

Pololu LSM303 compass sensor for absolute robot heading calculation (includes accelerometer for data filtering)

Xbee ee M Module e

Transmits a radio signal for peer to peer and broadcast communication with the other robots

Ping ng S Sens nsor

  • r

Ultrasonic sensor with range

  • f 2cm to 3m. Every robot

uses the ultrasonic receiver except for the beacon robot that only uses the transmitters

Magnetic E Encoder ers

Used in conjunction with the compass sensor for increased precision while turning

slide-6
SLIDE 6

SOFTWARE

slide-7
SLIDE 7

System em Overview ew

Localization Formation Algorithm

slide-8
SLIDE 8

Distanc nce Calcul ulation

  • n
slide-9
SLIDE 9
slide-10
SLIDE 10
slide-11
SLIDE 11
slide-12
SLIDE 12
slide-13
SLIDE 13
slide-14
SLIDE 14
slide-15
SLIDE 15
slide-16
SLIDE 16
slide-17
SLIDE 17
slide-18
SLIDE 18
slide-19
SLIDE 19
slide-20
SLIDE 20
slide-21
SLIDE 21

Putti tting it t all to together

slide-22
SLIDE 22

D = T * R

(speed of sound) (time of flight) Distance between robots

slide-23
SLIDE 23

735 cm

slide-24
SLIDE 24

81 cm

slide-25
SLIDE 25

62 cm

slide-26
SLIDE 26

60 cm

slide-27
SLIDE 27

63 cm

slide-28
SLIDE 28

67 cm

slide-29
SLIDE 29

735 cm

slide-30
SLIDE 30

The he C Coor

  • ordina

nate S System

Master Workers

0° 30 ° 240 ° 40 cm

slide-31
SLIDE 31

Wha hat’s the he P Poi

  • int

nt of

  • f the

he L Loc

  • calization
  • n?

Accurate

2 cm margin of error

Works Indoors

useful for close-range tasks

Closed System

Works in any environment

slide-32
SLIDE 32
slide-33
SLIDE 33

How

  • w D

Doe

  • es I

It Wor

  • rk?
slide-34
SLIDE 34

How

  • w D

Doe

  • es I

It Wor

  • rk?
slide-35
SLIDE 35

How

  • w D

Doe

  • es I

It Wor

  • rk?
slide-36
SLIDE 36

How

  • w D

Doe

  • es I

It Wor

  • rk?
slide-37
SLIDE 37

How

  • w D

Doe

  • es I

It Wor

  • rk?
slide-38
SLIDE 38

Wha hat is a a C Conv

  • nvex Hull? (and

nd why hy i it m matters rs)

Reduces Complexity Motion Planning 3D Formations

slide-39
SLIDE 39

Wha hat’s the he P Poi

  • int

nt of

  • f the

he F For

  • rmation
  • n A

Algorithm hm?

Scalable

works for n robots

Adaptable

works with any n-sided regular polygon

Extendable

stepping stone for more complex systems

slide-40
SLIDE 40

Future Work

mesh networks more robots!

more fun! for cooler formations

better algorithms

relaying data quickly

robot mapping

for unknown environments ground-air coordination

add drones make it all faster

to improve response time

slide-41
SLIDE 41

Why hy D Doe

  • es T

Thi his Matter?

slide-42
SLIDE 42

Quest stions

slide-43
SLIDE 43

Appe pendi dix

slide-44
SLIDE 44

Pol

  • lar Coor
  • ordina

nate Math

slide-45
SLIDE 45

Mor

  • re on
  • n Conv
  • nvex Hulls
slide-46
SLIDE 46

XBee Radio Module

Series 1 model 1714 27mm x 24mm x 9mm 5v power module

Specifications Networ

  • rking P

g Prot

  • toc
  • col
  • ls

Broadcasting Peer to Peer Uses Serial to communicate with Arduino

Microcontroller interface Switching b bet etween een networ

  • rk pr

prot

  • toc
  • col
  • ls

Allows you to switch between the protocols in code using AT commands Easy to wire due to Serial communications. Simple to setup network logistics using XCTU (software provided by Digi)

Setup Network I ID

Can be identified by unique address under a specified operating channel

slide-47
SLIDE 47

Parallax Ping Sensor

Sender Receiver

  • Uses ultrasonic transmitter(on the right)

and a receiver (eye on the left) to measure distance

  • 2 cm to 3 m range
  • Led flickers to show when burst is sent
  • 3 pins (data, power and ground)

○ Uses data line that can be set to high or low to measure distance

  • r to send ultrasonic burst

Ultrasonic sensor sends a ping from the right and measures the time it takes until the left ping receives the ultrasonic burst.

slide-48
SLIDE 48

LSM303 Compass Sensor

I2C device Uses an accelerometer in addition to the magnetometer to compensate for sudden movements Mounted higher up on standoffs so that the electromagnetic field from the motors doesn’t interfere with the compass magnetometer Returns a heading in degrees relative to north

slide-49
SLIDE 49

Sensor Module Overview

Parallax Ping Sensor

Ultrasonic sensor with range of 2cm to 3m. Every robot uses the ultrasonic receiver except for the beacon robot that only uses the transmitters

Compass Sensor

Pololu LSM303 compass sensor for absolute robot heading calculation (includes accelerometer for data filtering)

Xbee Radio Module

Transmits a radio signal for peer to peer and broadcast communication with the other robots

Magnetic Encoders

Used in conjunction with the compass sensor for increased precision while turning

Cost-Efficient

Our prototyped sensor model was under $70. However, since we used Parallax Ping Sensors that we already owned and not generic 4-pin sensors, our true cost is actually under $50.

Closed System

Cross-environment functionality, no need to communicate with outside technologies like GPS, satellite, etc;

Indoor Accuracy

Margin of error under 2 cm indoors, unlike GPS and other methods

slide-50
SLIDE 50

Formation Algorithm Overview/Flowchart

After each robot’s relative polar coordinate has been calculated, each robot is assigned a new point in space to travel to and the robot formation is constructed.

  • Finding the convex hull of the shape formed by

the robots guarantees that the resulting shape will be non self-intersecting and that its centroid will lie inside the area bounded by its edges.

  • The centroid of the shape is not necessarily the

point assigned to the beacon robot: rather, point assignment is based upon ranking distances

  • sequentially. Therefore, minimum distances to

the target and maximized and maximum distances to the target are minimized, thus

  • ptimizing the distance each robot must travel to

reach its target.