SlideShare une entreprise Scribd logo
1  sur  27
Accurate GPS-free Positioning of Utility Vehicles for Specialty Agriculture Jacqueline Libby Robotics Institute Carnegie Mellon University George Kantor Robotics Institute Carnegie Mellon University
USDA: CASC (Comprehensive Automation for Specialty Crops) Carnegie Mellon University Plant Science Automation Robotics Localization Outreach GIS Ag Economics
Why do we need positioning?
Why isn’t GPS enough? Cost Sub-meter accuracy prohibitively expensive Performance Line of sight to satellites occluded by trees New fruit wall structures Orientation complement GPS
Experimental Platform Drive-by-wire electric vehicle Brake and steering motors Internal sensors Wheel encoders External Sensors 2 Sick LMS 291 laser range scanners Ruggedized Dell laptop Receive data via Ethernet, USB Ground Truth Applanix POS 220 LV high-accuracy positioning system
High Level Schematic
Wheel Encoders Already used for vehicle control We use to measure relative motion
Laser Range Scanners Already used for safety, obstacle detection We use to measure range and bearing to landmarks
Map Known locations of landmarks Created offline with Applanix ground truth
Localization Filter Estimate vehicle pose Extended Kalman Filter Bayesian estimation technique
Let’s first look at the map…..
Creating the map
Creating the map
Wheel Encoders -> Prediction Step
Prediction Step: Dead Reckoning Given previous pose Steering and wheel encoder values Predict: current pose Assumptions: Bicycle model  Error bad sensor data Wheel slip imperfect assumptions
Laser Range Scanners -> correction step
Correction Step
Correction Step
Correction Step
Correction Step
Correction Step
Correction Step
Extended Kalman Filter (EKF)
Tests and Results
Tests and Results
Tests and Results Error (m)
Conclusions and Future Work Very close to sub-meter accuracy goal Weakness: landmark spacing too dense Ongoing work:  Improve prediction step: laser scan matching Improve measurement step: natural features Turning at the end of the row Future work Remove mapping step (SLAM) Other low-cost sensors (IMU, low-cost GPS)

Contenu connexe

Tendances

GOOGLE'S AUTONOMUS CAR
GOOGLE'S AUTONOMUS CARGOOGLE'S AUTONOMUS CAR
GOOGLE'S AUTONOMUS CARjolsnaj
 
Google Driverless Car
Google Driverless CarGoogle Driverless Car
Google Driverless CarRunam Sharma
 
Google driverless car
Google driverless carGoogle driverless car
Google driverless carpiyupatel99
 
FME Before You Dig: The Sunesys One Call Automated Response System
FME Before You Dig: The Sunesys One Call Automated Response SystemFME Before You Dig: The Sunesys One Call Automated Response System
FME Before You Dig: The Sunesys One Call Automated Response SystemSafe Software
 

Tendances (6)

GOOGLE'S AUTONOMUS CAR
GOOGLE'S AUTONOMUS CARGOOGLE'S AUTONOMUS CAR
GOOGLE'S AUTONOMUS CAR
 
Google Driverless Car
Google Driverless CarGoogle Driverless Car
Google Driverless Car
 
Google driverless car
Google driverless carGoogle driverless car
Google driverless car
 
FME Before You Dig: The Sunesys One Call Automated Response System
FME Before You Dig: The Sunesys One Call Automated Response SystemFME Before You Dig: The Sunesys One Call Automated Response System
FME Before You Dig: The Sunesys One Call Automated Response System
 
Waymo Driverless car
Waymo Driverless carWaymo Driverless car
Waymo Driverless car
 
12
1212
12
 

En vedette

PRECISE AGRICULTURE USING GPS
PRECISE AGRICULTURE USING GPSPRECISE AGRICULTURE USING GPS
PRECISE AGRICULTURE USING GPSAbhiram Kanigolla
 
GPS IN PRECISION AGRICULTURE
GPS IN PRECISION AGRICULTUREGPS IN PRECISION AGRICULTURE
GPS IN PRECISION AGRICULTUREAmbuja Mohanty
 
Application of remote sensing in agriculture
Application of remote sensing in agricultureApplication of remote sensing in agriculture
Application of remote sensing in agriculturevajinder kalra
 
State of the Word 2011
State of the Word 2011State of the Word 2011
State of the Word 2011photomatt
 

En vedette (6)

Gps Navigation System
Gps Navigation SystemGps Navigation System
Gps Navigation System
 
Gps navigation
Gps navigationGps navigation
Gps navigation
 
PRECISE AGRICULTURE USING GPS
PRECISE AGRICULTURE USING GPSPRECISE AGRICULTURE USING GPS
PRECISE AGRICULTURE USING GPS
 
GPS IN PRECISION AGRICULTURE
GPS IN PRECISION AGRICULTUREGPS IN PRECISION AGRICULTURE
GPS IN PRECISION AGRICULTURE
 
Application of remote sensing in agriculture
Application of remote sensing in agricultureApplication of remote sensing in agriculture
Application of remote sensing in agriculture
 
State of the Word 2011
State of the Word 2011State of the Word 2011
State of the Word 2011
 

Similaire à GPS-free Positioning for Utility Vehicles in Specialty Agriculture

Presentaion On Driverless Car
Presentaion On Driverless Car Presentaion On Driverless Car
Presentaion On Driverless Car Neha Deshpande
 
Rohan Divekar-Resume
Rohan Divekar-ResumeRohan Divekar-Resume
Rohan Divekar-ResumeRohan Divekar
 
mobile mapping survey.pptx
mobile mapping survey.pptxmobile mapping survey.pptx
mobile mapping survey.pptxsooryaK18
 
tatacara kalibrasi Kendaraan.ppt
tatacara kalibrasi Kendaraan.ppttatacara kalibrasi Kendaraan.ppt
tatacara kalibrasi Kendaraan.pptdarmadi ir,mm
 
vehicle calibration.ppt
vehicle calibration.pptvehicle calibration.ppt
vehicle calibration.pptdarmadi ir,mm
 
Towards Rapid Implementation of Adaptive Robotic Systems
Towards Rapid Implementation of Adaptive Robotic SystemsTowards Rapid Implementation of Adaptive Robotic Systems
Towards Rapid Implementation of Adaptive Robotic SystemsMeshDynamics
 
The Internet Of (very big) Things
The Internet Of (very big) ThingsThe Internet Of (very big) Things
The Internet Of (very big) ThingsGeoff Ballinger
 
Location Intelligence from Imagery
Location Intelligence from ImageryLocation Intelligence from Imagery
Location Intelligence from ImageryUjavalGandhi
 
Cloud Based Autonomous Vehicle Navigation
Cloud Based Autonomous Vehicle NavigationCloud Based Autonomous Vehicle Navigation
Cloud Based Autonomous Vehicle NavigationWilliam Smith
 
FULLY AUTONOMOUS DRIVERLESS CARS : GOOGLE CAR
FULLY AUTONOMOUS DRIVERLESS CARS : GOOGLE CARFULLY AUTONOMOUS DRIVERLESS CARS : GOOGLE CAR
FULLY AUTONOMOUS DRIVERLESS CARS : GOOGLE CARGokul Gopi
 
Autonomous Driving
Autonomous DrivingAutonomous Driving
Autonomous DrivingUsman Hashmi
 
Online Camer Calibration
Online Camer CalibrationOnline Camer Calibration
Online Camer CalibrationFei-Fei Zheng
 
Webinar1 darpa07
Webinar1 darpa07Webinar1 darpa07
Webinar1 darpa07MKGJUICE
 
Role of localization and environment perception in autonomous driving
Role of localization and environment perception in autonomous drivingRole of localization and environment perception in autonomous driving
Role of localization and environment perception in autonomous drivingQualcomm Research
 
SPARSH (Solar Powered Automated Route Sensing Hexapod)
SPARSH (Solar Powered Automated Route Sensing Hexapod)SPARSH (Solar Powered Automated Route Sensing Hexapod)
SPARSH (Solar Powered Automated Route Sensing Hexapod)Shivam Chaubey
 
Hand free
Hand freeHand free
Hand freeNag Ra
 
A Deep Learning algorithm for automatic detection of unexpected accidents und...
A Deep Learning algorithm for automatic detection of unexpected accidents und...A Deep Learning algorithm for automatic detection of unexpected accidents und...
A Deep Learning algorithm for automatic detection of unexpected accidents und...19520SaiSree
 

Similaire à GPS-free Positioning for Utility Vehicles in Specialty Agriculture (20)

Presentaion On Driverless Car
Presentaion On Driverless Car Presentaion On Driverless Car
Presentaion On Driverless Car
 
Rohan Divekar-Resume
Rohan Divekar-ResumeRohan Divekar-Resume
Rohan Divekar-Resume
 
mobile mapping survey.pptx
mobile mapping survey.pptxmobile mapping survey.pptx
mobile mapping survey.pptx
 
tatacara kalibrasi Kendaraan.ppt
tatacara kalibrasi Kendaraan.ppttatacara kalibrasi Kendaraan.ppt
tatacara kalibrasi Kendaraan.ppt
 
vehicle calibration.ppt
vehicle calibration.pptvehicle calibration.ppt
vehicle calibration.ppt
 
Towards Rapid Implementation of Adaptive Robotic Systems
Towards Rapid Implementation of Adaptive Robotic SystemsTowards Rapid Implementation of Adaptive Robotic Systems
Towards Rapid Implementation of Adaptive Robotic Systems
 
The Internet Of (very big) Things
The Internet Of (very big) ThingsThe Internet Of (very big) Things
The Internet Of (very big) Things
 
Gps
GpsGps
Gps
 
Location Intelligence from Imagery
Location Intelligence from ImageryLocation Intelligence from Imagery
Location Intelligence from Imagery
 
Cloud Based Autonomous Vehicle Navigation
Cloud Based Autonomous Vehicle NavigationCloud Based Autonomous Vehicle Navigation
Cloud Based Autonomous Vehicle Navigation
 
FULLY AUTONOMOUS DRIVERLESS CARS : GOOGLE CAR
FULLY AUTONOMOUS DRIVERLESS CARS : GOOGLE CARFULLY AUTONOMOUS DRIVERLESS CARS : GOOGLE CAR
FULLY AUTONOMOUS DRIVERLESS CARS : GOOGLE CAR
 
Autonomous Driving
Autonomous DrivingAutonomous Driving
Autonomous Driving
 
Online Camer Calibration
Online Camer CalibrationOnline Camer Calibration
Online Camer Calibration
 
Auto pilot Mode ppt
Auto pilot Mode pptAuto pilot Mode ppt
Auto pilot Mode ppt
 
Webinar1 darpa07
Webinar1 darpa07Webinar1 darpa07
Webinar1 darpa07
 
Role of localization and environment perception in autonomous driving
Role of localization and environment perception in autonomous drivingRole of localization and environment perception in autonomous driving
Role of localization and environment perception in autonomous driving
 
SPARSH (Solar Powered Automated Route Sensing Hexapod)
SPARSH (Solar Powered Automated Route Sensing Hexapod)SPARSH (Solar Powered Automated Route Sensing Hexapod)
SPARSH (Solar Powered Automated Route Sensing Hexapod)
 
Hand free
Hand freeHand free
Hand free
 
A Deep Learning algorithm for automatic detection of unexpected accidents und...
A Deep Learning algorithm for automatic detection of unexpected accidents und...A Deep Learning algorithm for automatic detection of unexpected accidents und...
A Deep Learning algorithm for automatic detection of unexpected accidents und...
 
Sazz
SazzSazz
Sazz
 

Plus de Comprehensive Automation for Specialty Crops

Plus de Comprehensive Automation for Specialty Crops (20)

On-the-go Caliper and Counter Device for Shade and Fruit Tree Nursery Invento...
On-the-go Caliper and Counter Device for Shade and Fruit Tree Nursery Invento...On-the-go Caliper and Counter Device for Shade and Fruit Tree Nursery Invento...
On-the-go Caliper and Counter Device for Shade and Fruit Tree Nursery Invento...
 
Update: Automation for Bare Root Ornamental and Fruit Tree Inventory
Update: Automation for Bare Root Ornamental and Fruit Tree InventoryUpdate: Automation for Bare Root Ornamental and Fruit Tree Inventory
Update: Automation for Bare Root Ornamental and Fruit Tree Inventory
 
Automatic Monitoring of Insect Populations
Automatic Monitoring of Insect PopulationsAutomatic Monitoring of Insect Populations
Automatic Monitoring of Insect Populations
 
Vacuum Apple Harvester
Vacuum Apple HarvesterVacuum Apple Harvester
Vacuum Apple Harvester
 
WTFRC Italy Trip
WTFRC Italy TripWTFRC Italy Trip
WTFRC Italy Trip
 
Specialty Crop Market Opportunity
Specialty Crop Market OpportunitySpecialty Crop Market Opportunity
Specialty Crop Market Opportunity
 
A Redesigned Electronic Insect Trap for Automated Monitoring of Lepidoptera i...
A Redesigned Electronic Insect Trap for Automated Monitoring of Lepidoptera i...A Redesigned Electronic Insect Trap for Automated Monitoring of Lepidoptera i...
A Redesigned Electronic Insect Trap for Automated Monitoring of Lepidoptera i...
 
Distributed Sensing in Horticultural Environments
Distributed Sensing in Horticultural EnvironmentsDistributed Sensing in Horticultural Environments
Distributed Sensing in Horticultural Environments
 
Developments in Technology and Automation for Tree Fruit
Developments in Technology and Automation for Tree FruitDevelopments in Technology and Automation for Tree Fruit
Developments in Technology and Automation for Tree Fruit
 
Development of an Autonomous Sensing and Positioning System for Use With Frui...
Development of an Autonomous Sensing and Positioning System for Use With Frui...Development of an Autonomous Sensing and Positioning System for Use With Frui...
Development of an Autonomous Sensing and Positioning System for Use With Frui...
 
CASC Team Showcase: Sociological Implications
CASC Team Showcase: Sociological ImplicationsCASC Team Showcase: Sociological Implications
CASC Team Showcase: Sociological Implications
 
Novel Approaches to Bin Filling
Novel Approaches to Bin FillingNovel Approaches to Bin Filling
Novel Approaches to Bin Filling
 
CASC Showcase 2010 05 20
CASC Showcase 2010 05 20CASC Showcase 2010 05 20
CASC Showcase 2010 05 20
 
On-the-fly Tree Counting and Caliper Measure
On-the-fly Tree Counting and Caliper MeasureOn-the-fly Tree Counting and Caliper Measure
On-the-fly Tree Counting and Caliper Measure
 
Using Surveys to Overcome Obstacles to Specialty Crop Industry Adoption of Au...
Using Surveys to Overcome Obstacles to Specialty Crop Industry Adoption of Au...Using Surveys to Overcome Obstacles to Specialty Crop Industry Adoption of Au...
Using Surveys to Overcome Obstacles to Specialty Crop Industry Adoption of Au...
 
Improving Orchard Efficiency with Autonomous Utility Vehicles
Improving Orchard Efficiency with Autonomous Utility VehiclesImproving Orchard Efficiency with Autonomous Utility Vehicles
Improving Orchard Efficiency with Autonomous Utility Vehicles
 
Innovative ThinKing
Innovative ThinKingInnovative ThinKing
Innovative ThinKing
 
The Technology Collaborative Project:Sensor Networks for Disease Management i...
The Technology Collaborative Project:Sensor Networks for Disease Management i...The Technology Collaborative Project:Sensor Networks for Disease Management i...
The Technology Collaborative Project:Sensor Networks for Disease Management i...
 
Preliminary Trial: The WeedSeeker
Preliminary Trial: The WeedSeekerPreliminary Trial: The WeedSeeker
Preliminary Trial: The WeedSeeker
 
Towards Automated Detection of Stress in Tree Fruit Production
Towards Automated Detection of Stress in Tree Fruit ProductionTowards Automated Detection of Stress in Tree Fruit Production
Towards Automated Detection of Stress in Tree Fruit Production
 

Dernier

Guide Complete Set of Residential Architectural Drawings PDF
Guide Complete Set of Residential Architectural Drawings PDFGuide Complete Set of Residential Architectural Drawings PDF
Guide Complete Set of Residential Architectural Drawings PDFChandresh Chudasama
 
trending-flavors-and-ingredients-in-salty-snacks-us-2024_Redacted-V2.pdf
trending-flavors-and-ingredients-in-salty-snacks-us-2024_Redacted-V2.pdftrending-flavors-and-ingredients-in-salty-snacks-us-2024_Redacted-V2.pdf
trending-flavors-and-ingredients-in-salty-snacks-us-2024_Redacted-V2.pdfMintel Group
 
20220816-EthicsGrade_Scorecard-JP_Morgan_Chase-Q2-63_57.pdf
20220816-EthicsGrade_Scorecard-JP_Morgan_Chase-Q2-63_57.pdf20220816-EthicsGrade_Scorecard-JP_Morgan_Chase-Q2-63_57.pdf
20220816-EthicsGrade_Scorecard-JP_Morgan_Chase-Q2-63_57.pdfChris Skinner
 
Unveiling the Soundscape Music for Psychedelic Experiences
Unveiling the Soundscape Music for Psychedelic ExperiencesUnveiling the Soundscape Music for Psychedelic Experiences
Unveiling the Soundscape Music for Psychedelic ExperiencesDoe Paoro
 
Effective Strategies for Maximizing Your Profit When Selling Gold Jewelry
Effective Strategies for Maximizing Your Profit When Selling Gold JewelryEffective Strategies for Maximizing Your Profit When Selling Gold Jewelry
Effective Strategies for Maximizing Your Profit When Selling Gold JewelryWhittensFineJewelry1
 
Entrepreneurship lessons in Philippines
Entrepreneurship lessons in  PhilippinesEntrepreneurship lessons in  Philippines
Entrepreneurship lessons in PhilippinesDavidSamuel525586
 
Jewish Resources in the Family Resource Centre
Jewish Resources in the Family Resource CentreJewish Resources in the Family Resource Centre
Jewish Resources in the Family Resource CentreNZSG
 
20200128 Ethical by Design - Whitepaper.pdf
20200128 Ethical by Design - Whitepaper.pdf20200128 Ethical by Design - Whitepaper.pdf
20200128 Ethical by Design - Whitepaper.pdfChris Skinner
 
1911 Gold Corporate Presentation Apr 2024.pdf
1911 Gold Corporate Presentation Apr 2024.pdf1911 Gold Corporate Presentation Apr 2024.pdf
1911 Gold Corporate Presentation Apr 2024.pdfShaun Heinrichs
 
Pitch Deck Teardown: Xpanceo's $40M Seed deck
Pitch Deck Teardown: Xpanceo's $40M Seed deckPitch Deck Teardown: Xpanceo's $40M Seed deck
Pitch Deck Teardown: Xpanceo's $40M Seed deckHajeJanKamps
 
Healthcare Feb. & Mar. Healthcare Newsletter
Healthcare Feb. & Mar. Healthcare NewsletterHealthcare Feb. & Mar. Healthcare Newsletter
Healthcare Feb. & Mar. Healthcare NewsletterJamesConcepcion7
 
Memorándum de Entendimiento (MoU) entre Codelco y SQM
Memorándum de Entendimiento (MoU) entre Codelco y SQMMemorándum de Entendimiento (MoU) entre Codelco y SQM
Memorándum de Entendimiento (MoU) entre Codelco y SQMVoces Mineras
 
digital marketing , introduction of digital marketing
digital marketing , introduction of digital marketingdigital marketing , introduction of digital marketing
digital marketing , introduction of digital marketingrajputmeenakshi733
 
Cyber Security Training in Office Environment
Cyber Security Training in Office EnvironmentCyber Security Training in Office Environment
Cyber Security Training in Office Environmentelijahj01012
 
Church Building Grants To Assist With New Construction, Additions, And Restor...
Church Building Grants To Assist With New Construction, Additions, And Restor...Church Building Grants To Assist With New Construction, Additions, And Restor...
Church Building Grants To Assist With New Construction, Additions, And Restor...Americas Got Grants
 
business environment micro environment macro environment.pptx
business environment micro environment macro environment.pptxbusiness environment micro environment macro environment.pptx
business environment micro environment macro environment.pptxShruti Mittal
 
Cybersecurity Awareness Training Presentation v2024.03
Cybersecurity Awareness Training Presentation v2024.03Cybersecurity Awareness Training Presentation v2024.03
Cybersecurity Awareness Training Presentation v2024.03DallasHaselhorst
 
Excvation Safety for safety officers reference
Excvation Safety for safety officers referenceExcvation Safety for safety officers reference
Excvation Safety for safety officers referencessuser2c065e
 
Fordham -How effective decision-making is within the IT department - Analysis...
Fordham -How effective decision-making is within the IT department - Analysis...Fordham -How effective decision-making is within the IT department - Analysis...
Fordham -How effective decision-making is within the IT department - Analysis...Peter Ward
 
The McKinsey 7S Framework: A Holistic Approach to Harmonizing All Parts of th...
The McKinsey 7S Framework: A Holistic Approach to Harmonizing All Parts of th...The McKinsey 7S Framework: A Holistic Approach to Harmonizing All Parts of th...
The McKinsey 7S Framework: A Holistic Approach to Harmonizing All Parts of th...Operational Excellence Consulting
 

Dernier (20)

Guide Complete Set of Residential Architectural Drawings PDF
Guide Complete Set of Residential Architectural Drawings PDFGuide Complete Set of Residential Architectural Drawings PDF
Guide Complete Set of Residential Architectural Drawings PDF
 
trending-flavors-and-ingredients-in-salty-snacks-us-2024_Redacted-V2.pdf
trending-flavors-and-ingredients-in-salty-snacks-us-2024_Redacted-V2.pdftrending-flavors-and-ingredients-in-salty-snacks-us-2024_Redacted-V2.pdf
trending-flavors-and-ingredients-in-salty-snacks-us-2024_Redacted-V2.pdf
 
20220816-EthicsGrade_Scorecard-JP_Morgan_Chase-Q2-63_57.pdf
20220816-EthicsGrade_Scorecard-JP_Morgan_Chase-Q2-63_57.pdf20220816-EthicsGrade_Scorecard-JP_Morgan_Chase-Q2-63_57.pdf
20220816-EthicsGrade_Scorecard-JP_Morgan_Chase-Q2-63_57.pdf
 
Unveiling the Soundscape Music for Psychedelic Experiences
Unveiling the Soundscape Music for Psychedelic ExperiencesUnveiling the Soundscape Music for Psychedelic Experiences
Unveiling the Soundscape Music for Psychedelic Experiences
 
Effective Strategies for Maximizing Your Profit When Selling Gold Jewelry
Effective Strategies for Maximizing Your Profit When Selling Gold JewelryEffective Strategies for Maximizing Your Profit When Selling Gold Jewelry
Effective Strategies for Maximizing Your Profit When Selling Gold Jewelry
 
Entrepreneurship lessons in Philippines
Entrepreneurship lessons in  PhilippinesEntrepreneurship lessons in  Philippines
Entrepreneurship lessons in Philippines
 
Jewish Resources in the Family Resource Centre
Jewish Resources in the Family Resource CentreJewish Resources in the Family Resource Centre
Jewish Resources in the Family Resource Centre
 
20200128 Ethical by Design - Whitepaper.pdf
20200128 Ethical by Design - Whitepaper.pdf20200128 Ethical by Design - Whitepaper.pdf
20200128 Ethical by Design - Whitepaper.pdf
 
1911 Gold Corporate Presentation Apr 2024.pdf
1911 Gold Corporate Presentation Apr 2024.pdf1911 Gold Corporate Presentation Apr 2024.pdf
1911 Gold Corporate Presentation Apr 2024.pdf
 
Pitch Deck Teardown: Xpanceo's $40M Seed deck
Pitch Deck Teardown: Xpanceo's $40M Seed deckPitch Deck Teardown: Xpanceo's $40M Seed deck
Pitch Deck Teardown: Xpanceo's $40M Seed deck
 
Healthcare Feb. & Mar. Healthcare Newsletter
Healthcare Feb. & Mar. Healthcare NewsletterHealthcare Feb. & Mar. Healthcare Newsletter
Healthcare Feb. & Mar. Healthcare Newsletter
 
Memorándum de Entendimiento (MoU) entre Codelco y SQM
Memorándum de Entendimiento (MoU) entre Codelco y SQMMemorándum de Entendimiento (MoU) entre Codelco y SQM
Memorándum de Entendimiento (MoU) entre Codelco y SQM
 
digital marketing , introduction of digital marketing
digital marketing , introduction of digital marketingdigital marketing , introduction of digital marketing
digital marketing , introduction of digital marketing
 
Cyber Security Training in Office Environment
Cyber Security Training in Office EnvironmentCyber Security Training in Office Environment
Cyber Security Training in Office Environment
 
Church Building Grants To Assist With New Construction, Additions, And Restor...
Church Building Grants To Assist With New Construction, Additions, And Restor...Church Building Grants To Assist With New Construction, Additions, And Restor...
Church Building Grants To Assist With New Construction, Additions, And Restor...
 
business environment micro environment macro environment.pptx
business environment micro environment macro environment.pptxbusiness environment micro environment macro environment.pptx
business environment micro environment macro environment.pptx
 
Cybersecurity Awareness Training Presentation v2024.03
Cybersecurity Awareness Training Presentation v2024.03Cybersecurity Awareness Training Presentation v2024.03
Cybersecurity Awareness Training Presentation v2024.03
 
Excvation Safety for safety officers reference
Excvation Safety for safety officers referenceExcvation Safety for safety officers reference
Excvation Safety for safety officers reference
 
Fordham -How effective decision-making is within the IT department - Analysis...
Fordham -How effective decision-making is within the IT department - Analysis...Fordham -How effective decision-making is within the IT department - Analysis...
Fordham -How effective decision-making is within the IT department - Analysis...
 
The McKinsey 7S Framework: A Holistic Approach to Harmonizing All Parts of th...
The McKinsey 7S Framework: A Holistic Approach to Harmonizing All Parts of th...The McKinsey 7S Framework: A Holistic Approach to Harmonizing All Parts of th...
The McKinsey 7S Framework: A Holistic Approach to Harmonizing All Parts of th...
 

GPS-free Positioning for Utility Vehicles in Specialty Agriculture

Notes de l'éditeur

  1. Here is a picture of the robotic vehicle we use in our work.You can see here that it is driving by itself, based on the navigation and control algorithms developed by our group.Goal: determine position and orientation of this vehicle to sub meter accuracy without GPSMotivation:Reliable and affordable technology for specialty agriculture.High accuracy GPS cost-prohibitive
  2. Pos info so fundamental to these types of systemsthat we often forget why it is significant.for example, allows data collected by on-board sensors to be geo-registered into maps this plot shows 3d point clouds from our laser range scanners other sensors, such as temperature and humidity, could be registered in a similar mannerAll this information can be collated into a GIS databaseThese photographs show examples of devices that could be towed on such a vehicle.This is a mower, and this is a weedseeker.This weedseeker both detects weeks and then sprays them.You could imagine that the information on where a week was detected, and where spraying occurred could be registered into a GIS data base. This could be used not only for record management, But for decision making purposes For example, a manager could decide on targeted areas for where further weedseeking should be performed. By targeting these areas, valuable chemicals and resources can be conservedOptional Automated vehicles Follow paths to targeted areas Safety Reduce human exposure to chemicals Reduce worker fatigue
  3. Line of sight to satellites is occluded by treesNot a problem in broad-acre crops, where GPS has been used successfully for many yearsEven in tall tree canopies, if tractors are tall enough, then signal interference not an issuebut for smaller utility vehicles, such as the one used in this work, vehicle must operate well below tree lineApple orchards, orange groves, almond groves: all examples of specialty agriculture where tall tree canopies would cause an issue for small vehiclesOther applications: environmental/biologicalForests, tree farmsi.e. monitoring carbon sequestrationNew fruit wall structures:Engineered to maximize light interception by the canopyRule out possibility of mounting antennas above the tree lineCost:GPS systems that provide sub-meter accuracy are prohibitively expensiveGPS gives position, but not orientation, of the vehicleAutomation: Needed for controlling turnsCollecting data: needed for determining the position of an object in the environment wrt the vehicleComplementary:In some applications, perhaps a cheap GPS can still be usedOut algorithms can complement this GPSProvide corrections when the GPS fails -> robustnessProvide orientation on top of positionMore accuracy
  4. Just talk about internal/external********************************drive-by-wire electric vehicleBrake and steering motors can be controlled by eitherHuman operatorAutonomous commands from an on-board computerInternal sensors: measure properties internal to vehicleEncoders: measure distance traveled and steering angleExternal sensors: measure properties of the environment2 Sick LMS 291 laser scannersSend out horizontal fan of beams, 180 degrees, 1 deg intervalsMeasure range and bearing to surrounding objectsRigidly attached, angled 30 degRuggedized Dell LaptopReceiving data via Ethernet and USBApplanix POS 220 LV high-accuracy positioning systemProvides ground truthPosition estimate: a few centimentersOrientation: 0.05 degrees
  5. Use sensors already on vehicle for other purposes:Wheel encoders provide measurements of vehicle motionAlready being used for controlWe use for prediction of relative position
  6. Laser range scannersAlready being used for safetyWe use to measure range and bearing to landmarks at known locationsCorrect our estimate of the vehicle’s pose
  7. Knowing the locations of the landmarks requires having a map We create the map offline with the help of ground truth form the applanixThis is done offline
  8. The main part of the algorithm is this feedback loop here.The wheel encoders provide a prediction of the vehicle’s position, which provides the dead reckoning step, in this boxThe lasers detect features in the map, which are used to correct this prediction which provides the correction step hereThe filter is then an iterative process that loops through this cycle as the vehicle moves through the world.
  9. List of (x,y) positions of landmarks in the environmentReflective tape, lasers read as high-reflectivity returnsDrive through the orchard, passing by landmarks multiple times from different anglesRecording data from both lasers and applanix1) We start with the a range and bearing measurement to a landmark2) With then re-write this as the x,y position wrt the laser frame3) The “H’s” shown in the diagram are what we call homogeneous transformation matricesAllow us to transform (x,y) coordinates from one frame to anotherLaser frame -> vehicle frameVehicle frame -> world frame4) Laser -> vehicle: fixed position and orientation – we calibrate for this5) vehicle -> world: applanixAs we drive up and down the orchard rows, we record the world coordinates of all the high-refl returns,As you can see here in this map with the pink dotsWe then use a clustering technique to take each local point cloud region and turn it into a single point feature for the mapShown by the x’s
  10. List of (x,y) positions of landmarks in the environmentReflective tape, lasers read as high-reflectivity returnsDrive through the orchard, passing by landmarks multiple times from different anglesRecording data from both lasers and applanix1) We start with the a range and bearing measurement to a landmark2) With then re-write this as the x,y position wrt the laser frame3) The “H’s” shown in the diagram are what we call homogeneous transformation matricesAllow us to transform (x,y) coordinates from one frame to anotherLaser frame -> vehicle frameVehicle frame -> world frame4) Laser -> vehicle: fixed position and orientation – we calibrate for this5) vehicle -> world: applanixAs we drive up and down the orchard rows, we record the world coordinates of all the high-refl returns,As you can see here in this map with the pink dotsWe then use a clustering technique to take each local point cloud region and turn it into a single point feature for the mapShown by the x’s
  11. given:Vehicle pose at time tSteering and wheel encoder values at time t+TFind:Vehicle pose at time t+TAssumption:Ackerman steeringDifferential drive 4-wheel vehicleSimplify to bicycle modelEuler approximation: curved -> point and shootEncoder on motor gives estimate of the forward velocityEncoder on the steering wheels gives us an estimate of direction the vehicle is traveling in -> angular velocity
  12. ReasonDead reckoning builds up error over timeUse measurements to landmarks to correct for this errorGiven:Actual measurement: range and bearing readings from laserModel of the measurement: Estimated vehicle (and laser) poseLandmark location, in mapLook at difference between actual measurement and measurement modelUse this to correct from our estimated laser pose closer to the actual laser pose
  13. ReasonDead reckoning builds up error over timeUse measurements to landmarks to correct for this errorGiven:Actual measurement: range and bearing readings from laserModel of the measurement: Estimated vehicle (and laser) poseLandmark location, in mapLook at difference between actual measurement and measurement modelUse this to correct from our estimated laser pose closer to the actual laser pose
  14. ReasonDead reckoning builds up error over timeUse measurements to landmarks to correct for this errorGiven:Actual measurement: range and bearing readings from laserModel of the measurement: Estimated vehicle (and laser) poseLandmark location, in mapLook at difference between actual measurement and measurement modelUse this to correct from our estimated laser pose closer to the actual laser pose
  15. ReasonDead reckoning builds up error over timeUse measurements to landmarks to correct for this errorGiven:Actual measurement: range and bearing readings from laserModel of the measurement: Estimated vehicle (and laser) poseLandmark location, in mapLook at difference between actual measurement and measurement modelUse this to correct from our estimated laser pose closer to the actual laser pose
  16. ReasonDead reckoning builds up error over timeUse measurements to landmarks to correct for this errorGiven:Actual measurement: range and bearing readings from laserModel of the measurement: Estimated vehicle (and laser) poseLandmark location, in mapLook at difference between actual measurement and measurement modelUse this to correct from our estimated laser pose closer to the actual laser pose
  17. ReasonDead reckoning builds up error over timeUse measurements to landmarks to correct for this errorGiven:Actual measurement: range and bearing readings from laserModel of the measurement: Estimated vehicle (and laser) poseLandmark location, in mapLook at difference between actual measurement and measurement modelUse this to correct from our estimated laser pose closer to the actual laser pose
  18. Iterative processPrediction and update steps coming in at different timesDependent on the frequencies of the various sensorsNot going into details here because not enough time….
  19. This localization algorithm was tested successfully in a variety of settingsSite description:The plot here shows results from experiments in apple orchard blocks at the Penn State Fruit Research and Extension CenterDuring the summer of 2009Total area of block: 60 square m6 50 m long rowsOrchard block was relatively new planting (~3yr old), trained in a vertical trellis systemExperimental SetupTraffic cones were placed throughout the orchardA total of 39 landmarks, shows as black dotsSpaced about 20 m apart in pairsFirst collect a data set to create the map, which was constructed offlineSubsequent experiments run to test the localization algorithm onlineGenerated real-time position estimates as the vehicle drove through the blockAfter the tests, results were analyzed to generate statistics on performancePlotResults of a typical runExplain colorsVeering slightly offWhen the vehicle goes for a fair amount of time without receiving measurements to landmarksWhen the measurement is finally received, the vehicle corrects itselfHistogramPrimary metric we use to quantify error is the euclidean distance between the estimate and ground truth at each timestepMean: 20cm, max: 1.2 mThe cumulative error distribution is plotted in red, with the scale on the rightThis shows us, for instance, that 90% of the time, the error is under 35 cm
  20. This localization algorithm was tested successfully in a variety of settingsSite description:The plot here shows results from experiments in apple orchard blocks at the Penn State Fruit Research and Extension CenterDuring the summer of 2009Total area of block: 60 square m6 50 m long rowsOrchard block was relatively new planting (~3yr old), trained in a vertical trellis systemExperimental SetupTraffic cones were placed throughout the orchardA total of 39 landmarks, shows as black dotsSpaced about 20 m apart in pairsFirst collect a data set to create the map, which was constructed offlineSubsequent experiments run to test the localization algorithm onlineGenerated real-time position estimates as the vehicle drove through the blockAfter the tests, results were analyzed to generate statistics on performancePlotResults of a typical runExplain colorsVeering slightly offWhen the vehicle goes for a fair amount of time without receiving measurements to landmarksWhen the measurement is finally received, the vehicle corrects itselfHistogramPrimary metric we use to quantify error is the euclidean distance between the estimate and ground truth at each timestepMean: 20cm, max: 1.2 mThe cumulative error distribution is plotted in red, with the scale on the rightThis shows us, for instance, that 90% of the time, the error is under 35 cm
  21. This localization algorithm was tested successfully in a variety of settingsSite description:The plot here shows results from experiments in apple orchard blocks at the Penn State Fruit Research and Extension CenterDuring the summer of 2009Total area of block: 60 square m6 50 m long rowsOrchard block was relatively new planting (~3yr old), trained in a vertical trellis systemExperimental SetupTraffic cones were placed throughout the orchardA total of 39 landmarks, shows as black dotsSpaced about 20 m apart in pairsFirst collect a data set to create the map, which was constructed offlineSubsequent experiments run to test the localization algorithm onlineGenerated real-time position estimates as the vehicle drove through the blockAfter the tests, results were analyzed to generate statistics on performancePlotResults of a typical runExplain colorsVeering slightly offWhen the vehicle goes for a fair amount of time without receiving measurements to landmarksWhen the measurement is finally received, the vehicle corrects itselfHistogramPrimary metric we use to quantify error is the euclidean distance between the estimate and ground truth at each timestepMean: 20cm, max: 1.2 mThe cumulative error distribution is plotted in red, with the scale on the rightThis shows us, for instance, that 90% of the time, the error is under 35 cm
  22. Scan to scan matching***********We have demonstrated sub-meter accuracy with current experimentsPrimary weakness: landmark spacing denser than is practicalWhen landmarks placed > 20m apart, dead reckoning becomes too largeAlgorithm will associate measurements to the wrong landmarkOngoing workImproving prediction stepLaser scan matching, more accurate dead reckoning, reducing driftImproving measurement stepNatural features:Entire tree rows: line featuresTree trunks: point featuresHard problem: turning at the end of rowWheel slipLaser odometry and better feature extractionFuture Work:SLAM (Simultaneous Localization and Mapping)No longer need expensive ground truthing for mapping stepOther low-cost sensors IMU’sPartial GPS EKF is a general sensor fusion techniquesLends itself to fusing data from multiple sensorsBy using multiple cheap sensors, we can provide affortable and reliable systems