SlideShare une entreprise Scribd logo
1  sur  34
Télécharger pour lire hors ligne
Jerome Mies (Lecturer/ Researcher at Hogeschool van Amsterdam)
23 May, 2023
Biking on the edge
Monitoring the conditions
of streets in Amsterdam
with AWS IoT Greengrass
Introduction video –
http://vimeo.com/769075918/f2159baf0b
Overview
• What are we working on?
• What is edge computing?
• How do we use edge computing?
• How do we use AWS?
• How can we integrate AWS?
• Example models
Together improving the wellbeing in public spaces
Hogeschool van Amsterdam
Centre of expertise for applied
artificial intelligence
Smart Asset Management Lab
SURF
SURF is the collaborative
organisation for IT in Dutch
education and research
Municipalities
Velotech Solutions
Sustainable object detections
in the public space
Conditions of the assets in the city
impact our safety and wellbeing
Light posts
Traffic lights
Garbage
Traffic signs
Road
markings
Road surface
Trees
1
Our ambition is to cover the complete city maintenance
with LiDAR & video AI
Video image
recognition
LiDAR mapping
Light posts
Traffic lights
Garbage
Traffic signs
Road
markings
Road surface
Trees
1
Provide an accurate & complete overview of deficiencies
in the city through eco-friendly inspections
3
Detailed routes per area Image analysis using algorithms Registration in eform-app
Bicycles can go where cars can’t Validation of the inspections
Integration with maintenance
subcontractor systems
Key features of the inspections
Inspections (periodically) of designated areas by bicycle/vehicle
Number of identified objects, defects & exact GPS locations
Reporting & dashboard that is linked with the existing workflows
Purpose
Improve social & traffic safety
Minimize liability risks
Preventing peak workload for
maintenance parties in winter
Route
planning
Schedule
riders
Data gathering
& mobile
mapping
Before & after
inspections
Data
analysis
Reporting
municipality
Reporting
maintenance
contractor
Green and fast methods to get insights in the public space
by using bikes & technology
AI
2
Our approach:
Overview
• What are we working on?
• What is edge computing?
• How do we use edge computing?
• How do we use AWS?
• How can we integrate AWS?
• Example models
Edge computing is real-time computing at the end devices
What are edge devices?
A cloud computing provider can help integrate
applications with edge devices
• Cloud providers
• Amazon AWS
• Microsoft Azure
• Edge Impulse
• IoT (Edge) devices
• Jetson Nano
• Raspberry Pi
• Arduino Nano
Edge computing versus Cloud computing
Advantages of edge computing
• Real-time computing
• Limited sending of
information
• Privacy
Overview
• What are we working on?
• What is edge computing?
• How do we use edge computing?
• How do we use AWS?
• How can we integrate AWS?
• Example models
Light post detection at the edge
4
Inspections on a bicycle
Real-time inference on
images to detect defects
Bicycle-mounted edge
devices operate
autonomously.
From light post detection to tilted street lights
The angle of a light post
detected in an image is
determined in multiple
steps.
The angle of the light post is compared
to the horizon.
5
Light posts in a training set are labelled
using the Roboflow application and
models are trained.
Modular edge
device
• NVIDIA Jetson Nano or
NVIDIA Xavier NX
• 32 GB SD card
• 4G and WiFi connectivity
• Battery pack
• Camera
• GPS sensor and
gyroscope
Modular edge
device
Multi-vehicle use
Overview
• What are we working on?
• What is edge computing?
• How do we use edge computing?
• How do we use AWS?
• How can we integrate AWS?
• Example models
• Deploy software to all remote devices at
once from a central cloud location
• Software versioning
• CI/CD, Infrastructure as Code
• Modular Greengrass components make it
possible to design flexible architectures
adapting to multiple use cases
• Docker support
• Dealing with unstable network connections
reliably
Managing remote devices
with AWS Greengrass
• Every Greengrass component acts as a
microservice in a data pipeline, e.g.
taking pictures with a camera, doing the AI
inference, removing sensitive data,
uploading to the cloud, etc.
• Orchestrate event-driven workflows at the
device level, with components talking over
inter-process communication
• Trigger or schedule workflows remotely over
MQTT from AWS IoT Core
• Address individual edge devices using a
dedicated MQTT topic structure
Greengrass components
and event-driven workflow
• Using publicly available models
• Dedicated models to detect objects of interest, e.g. models from
municipalities for garbage detection
• Pre-trained models, such as YOLOv5 trained on a COCO dataset, to
detect faces for further removal
• Training custom models with transfer learning
• Training data is collected from the bicycles and labelled manually.
• Train models with Amazon SageMaker Studio
• Track training experiments and model performance with MLflow
Models and training
• Model optimization is crucial for battery-powered edge
applications.
• Limited to GPU capacity of the device (NVIDIA Jetson Nano vs NVIDIA
Xavier NX)
• Faster models (and more powerful devices) will allow us to move
from image processing to video processing to LIDAR processing at
the edge.
Model optimization and limitations of the edge
• With remote device management, scaling towards multiple edge
devices is easy.
• Remote connectivity enables real-time updates and event-driven
data processing.
• Modular design allows us to accommodate multiple use cases.
• Let’s make cities a safer place through cloud and edge technology
and AI.
• We have developed a blueprint for smart edge-computing
applications for use in public sector and research.
Cloud is the key to scalability at the edge
We can use AWS to integrate with our dashboard for simpler use
Integration AWS with a dashboard
Overview
• What are we working on?
• What is edge computing?
• How do we use edge computing?
• How do we use AWS?
• How can we integrate AWS?
• Example models
Examples of use cases for VeloTech
• Road deterioration
• Broken street lights
• Tilted lanterns
• Street signs pollution
• Lidar scans
Overview
• What are we working on?
• What is edge computing?
• How do we use edge computing?
• How do we use AWS?
• How can we integrate AWS?
• Example models
Road deterioration example
- text in Dutch
Broken street lights
example
Tilted street lights
example
Street signs pollution
example
Lidar scans – some ideas
Tilted street lights
Falling trees
Collapsing quays (‘kades’) and
bridges
Mapping the city
Questions?

Contenu connexe

Similaire à Biking on the edge - Jerome Mies - SRD23

cloud computing
cloud computingcloud computing
cloud computing
Hyderabad
 
Are you ready to be edgy? Bringing applications to the edge of the network
Are you ready to be edgy? Bringing applications to the edge of the networkAre you ready to be edgy? Bringing applications to the edge of the network
Are you ready to be edgy? Bringing applications to the edge of the network
Megan O'Keefe
 

Similaire à Biking on the edge - Jerome Mies - SRD23 (20)

Iit 1782 designing for the internet of things (io t) v4 gb
Iit 1782 designing for the internet of things (io t) v4 gbIit 1782 designing for the internet of things (io t) v4 gb
Iit 1782 designing for the internet of things (io t) v4 gb
 
AWS re:Invent 2016: Internet of Things (IoT) Edge and Device Services (IOT202)
AWS re:Invent 2016: Internet of Things (IoT) Edge and Device Services (IOT202)AWS re:Invent 2016: Internet of Things (IoT) Edge and Device Services (IOT202)
AWS re:Invent 2016: Internet of Things (IoT) Edge and Device Services (IOT202)
 
cloud computing
cloud computingcloud computing
cloud computing
 
Soldatos io t-academy-cosmote-231117-v-final
Soldatos io t-academy-cosmote-231117-v-finalSoldatos io t-academy-cosmote-231117-v-final
Soldatos io t-academy-cosmote-231117-v-final
 
Edge Computing and 5G, a powerful digital mix for IoT - AIT
Edge Computing and 5G, a powerful digital mix for IoT - AITEdge Computing and 5G, a powerful digital mix for IoT - AIT
Edge Computing and 5G, a powerful digital mix for IoT - AIT
 
Cloud Computing
Cloud ComputingCloud Computing
Cloud Computing
 
cloud computing
cloud computingcloud computing
cloud computing
 
Azure iot edge and AI enabling the intelligent edge
Azure iot edge and AI  enabling the intelligent edgeAzure iot edge and AI  enabling the intelligent edge
Azure iot edge and AI enabling the intelligent edge
 
System Support for Internet of Things
System Support for Internet of ThingsSystem Support for Internet of Things
System Support for Internet of Things
 
Exploring IoT Edge
Exploring IoT EdgeExploring IoT Edge
Exploring IoT Edge
 
Presentation on Cloud computing
Presentation on Cloud computingPresentation on Cloud computing
Presentation on Cloud computing
 
Are you ready to be edgy? Bringing applications to the edge of the network
Are you ready to be edgy? Bringing applications to the edge of the networkAre you ready to be edgy? Bringing applications to the edge of the network
Are you ready to be edgy? Bringing applications to the edge of the network
 
DataPalooza: ML & IoT Workshop
DataPalooza: ML & IoT WorkshopDataPalooza: ML & IoT Workshop
DataPalooza: ML & IoT Workshop
 
d.pptx
d.pptxd.pptx
d.pptx
 
11-Module-4 Opportunities and Challenges, Architectures for convergence,Data ...
11-Module-4 Opportunities and Challenges, Architectures for convergence,Data ...11-Module-4 Opportunities and Challenges, Architectures for convergence,Data ...
11-Module-4 Opportunities and Challenges, Architectures for convergence,Data ...
 
AWS Serverless Community Day Keynote and Vendia Launch 6-26-2020
AWS Serverless Community Day Keynote and Vendia Launch 6-26-2020AWS Serverless Community Day Keynote and Vendia Launch 6-26-2020
AWS Serverless Community Day Keynote and Vendia Launch 6-26-2020
 
Cloud computing
Cloud computingCloud computing
Cloud computing
 
"Keeping Brick and Mortar Relevant, A Look Inside Retail Analytics," A Presen...
"Keeping Brick and Mortar Relevant, A Look Inside Retail Analytics," A Presen..."Keeping Brick and Mortar Relevant, A Look Inside Retail Analytics," A Presen...
"Keeping Brick and Mortar Relevant, A Look Inside Retail Analytics," A Presen...
 
Machine Learning at the Edge
Machine Learning at the EdgeMachine Learning at the Edge
Machine Learning at the Edge
 
IoT at the Edge - Greengrass and More - AWS PS Summit Canberra 2017
IoT at the Edge - Greengrass and More - AWS PS Summit Canberra 2017IoT at the Edge - Greengrass and More - AWS PS Summit Canberra 2017
IoT at the Edge - Greengrass and More - AWS PS Summit Canberra 2017
 

Plus de SURF Events

Hoe door een multidisciplinaire aanpak de ondersteuning van het leren en werk...
Hoe door een multidisciplinaire aanpak de ondersteuning van het leren en werk...Hoe door een multidisciplinaire aanpak de ondersteuning van het leren en werk...
Hoe door een multidisciplinaire aanpak de ondersteuning van het leren en werk...
SURF Events
 
Betekenisvol leren in de digitalisering van het onderwijs - Michiel de Jong (...
Betekenisvol leren in de digitalisering van het onderwijs - Michiel de Jong (...Betekenisvol leren in de digitalisering van het onderwijs - Michiel de Jong (...
Betekenisvol leren in de digitalisering van het onderwijs - Michiel de Jong (...
SURF Events
 

Plus de SURF Events (20)

Onderwijsinnovatie Management CardGame - Ysbrand Hoetjes - OWD22
Onderwijsinnovatie Management CardGame - Ysbrand Hoetjes - OWD22Onderwijsinnovatie Management CardGame - Ysbrand Hoetjes - OWD22
Onderwijsinnovatie Management CardGame - Ysbrand Hoetjes - OWD22
 
Hoe door een multidisciplinaire aanpak de ondersteuning van het leren en werk...
Hoe door een multidisciplinaire aanpak de ondersteuning van het leren en werk...Hoe door een multidisciplinaire aanpak de ondersteuning van het leren en werk...
Hoe door een multidisciplinaire aanpak de ondersteuning van het leren en werk...
 
Gebruikers aan het woord, effectief doorontwikkelen - Jeffrey Braak & Larissa...
Gebruikers aan het woord, effectief doorontwikkelen - Jeffrey Braak & Larissa...Gebruikers aan het woord, effectief doorontwikkelen - Jeffrey Braak & Larissa...
Gebruikers aan het woord, effectief doorontwikkelen - Jeffrey Braak & Larissa...
 
Hoe geven de HU en Fontys vorm aan studentgestuurd onderwijs met hulp van een...
Hoe geven de HU en Fontys vorm aan studentgestuurd onderwijs met hulp van een...Hoe geven de HU en Fontys vorm aan studentgestuurd onderwijs met hulp van een...
Hoe geven de HU en Fontys vorm aan studentgestuurd onderwijs met hulp van een...
 
Blended education en de organisatie van onderwijs - Judith Vennix& Ralf Hille...
Blended education en de organisatie van onderwijs - Judith Vennix& Ralf Hille...Blended education en de organisatie van onderwijs - Judith Vennix& Ralf Hille...
Blended education en de organisatie van onderwijs - Judith Vennix& Ralf Hille...
 
Kickstart de blend- modellen en benaderingen - Barend Last (UM) - OWD22
Kickstart de blend- modellen en benaderingen - Barend Last (UM) - OWD22Kickstart de blend- modellen en benaderingen - Barend Last (UM) - OWD22
Kickstart de blend- modellen en benaderingen - Barend Last (UM) - OWD22
 
Erasmus without paper - een mijlpaal in de ondersteuning van studentmobilitei...
Erasmus without paper - een mijlpaal in de ondersteuning van studentmobilitei...Erasmus without paper - een mijlpaal in de ondersteuning van studentmobilitei...
Erasmus without paper - een mijlpaal in de ondersteuning van studentmobilitei...
 
Accelereren op leven lang ontwikkelen met (micro)credentials, badges en actio...
Accelereren op leven lang ontwikkelen met (micro)credentials, badges en actio...Accelereren op leven lang ontwikkelen met (micro)credentials, badges en actio...
Accelereren op leven lang ontwikkelen met (micro)credentials, badges en actio...
 
Green Award; trigger voor het verduurzamen van ict onderwijs - Claudia Tempel...
Green Award; trigger voor het verduurzamen van ict onderwijs - Claudia Tempel...Green Award; trigger voor het verduurzamen van ict onderwijs - Claudia Tempel...
Green Award; trigger voor het verduurzamen van ict onderwijs - Claudia Tempel...
 
De Le-Do-Loop- een nieuwe methode om hoogcognitieve leerdoelen te bereiken - ...
De Le-Do-Loop- een nieuwe methode om hoogcognitieve leerdoelen te bereiken - ...De Le-Do-Loop- een nieuwe methode om hoogcognitieve leerdoelen te bereiken - ...
De Le-Do-Loop- een nieuwe methode om hoogcognitieve leerdoelen te bereiken - ...
 
Een open boek ontwikkelen- het proces - Loes van Hove & Martijn Ouwehand (TU/...
Een open boek ontwikkelen- het proces - Loes van Hove & Martijn Ouwehand (TU/...Een open boek ontwikkelen- het proces - Loes van Hove & Martijn Ouwehand (TU/...
Een open boek ontwikkelen- het proces - Loes van Hove & Martijn Ouwehand (TU/...
 
Betekenisvol leren in de digitalisering van het onderwijs - Michiel de Jong (...
Betekenisvol leren in de digitalisering van het onderwijs - Michiel de Jong (...Betekenisvol leren in de digitalisering van het onderwijs - Michiel de Jong (...
Betekenisvol leren in de digitalisering van het onderwijs - Michiel de Jong (...
 
The impact of technology on society - how to become a responsible designer of...
The impact of technology on society - how to become a responsible designer of...The impact of technology on society - how to become a responsible designer of...
The impact of technology on society - how to become a responsible designer of...
 
Uncovering technologies #2- Cloud, Edge & XR - Tim Kok, David Salek, Esther v...
Uncovering technologies #2- Cloud, Edge & XR - Tim Kok, David Salek, Esther v...Uncovering technologies #2- Cloud, Edge & XR - Tim Kok, David Salek, Esther v...
Uncovering technologies #2- Cloud, Edge & XR - Tim Kok, David Salek, Esther v...
 
Trainen met impact in onderwijslogistiek - Iwan Brouns & Lisanne van Brouwers...
Trainen met impact in onderwijslogistiek - Iwan Brouns & Lisanne van Brouwers...Trainen met impact in onderwijslogistiek - Iwan Brouns & Lisanne van Brouwers...
Trainen met impact in onderwijslogistiek - Iwan Brouns & Lisanne van Brouwers...
 
From Emergency Remote Teaching to Post Covid Teaching & Learning - Chris Rouw...
From Emergency Remote Teaching to Post Covid Teaching & Learning - Chris Rouw...From Emergency Remote Teaching to Post Covid Teaching & Learning - Chris Rouw...
From Emergency Remote Teaching to Post Covid Teaching & Learning - Chris Rouw...
 
Digitale en open leermaterialen in NL- samenwerking centraal - Lieke Rensink ...
Digitale en open leermaterialen in NL- samenwerking centraal - Lieke Rensink ...Digitale en open leermaterialen in NL- samenwerking centraal - Lieke Rensink ...
Digitale en open leermaterialen in NL- samenwerking centraal - Lieke Rensink ...
 
Kan jij wat. Over de groei en reflectie tool voor de onderwijskundig ict prof...
Kan jij wat. Over de groei en reflectie tool voor de onderwijskundig ict prof...Kan jij wat. Over de groei en reflectie tool voor de onderwijskundig ict prof...
Kan jij wat. Over de groei en reflectie tool voor de onderwijskundig ict prof...
 
Hybride onderwijsvormen - Michel Jansen (SURF), Jet Bierman & Natasa Brouwer ...
Hybride onderwijsvormen - Michel Jansen (SURF), Jet Bierman & Natasa Brouwer ...Hybride onderwijsvormen - Michel Jansen (SURF), Jet Bierman & Natasa Brouwer ...
Hybride onderwijsvormen - Michel Jansen (SURF), Jet Bierman & Natasa Brouwer ...
 
Your VUture - het studieloopbaanspel - Nadia van der Pool & Bo Janssen(VU) - ...
Your VUture - het studieloopbaanspel - Nadia van der Pool & Bo Janssen(VU) - ...Your VUture - het studieloopbaanspel - Nadia van der Pool & Bo Janssen(VU) - ...
Your VUture - het studieloopbaanspel - Nadia van der Pool & Bo Janssen(VU) - ...
 

Dernier

Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Victor Rentea
 
Architecting Cloud Native Applications
Architecting Cloud Native ApplicationsArchitecting Cloud Native Applications
Architecting Cloud Native Applications
WSO2
 

Dernier (20)

"I see eyes in my soup": How Delivery Hero implemented the safety system for ...
"I see eyes in my soup": How Delivery Hero implemented the safety system for ..."I see eyes in my soup": How Delivery Hero implemented the safety system for ...
"I see eyes in my soup": How Delivery Hero implemented the safety system for ...
 
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot Model
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot ModelMcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot Model
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot Model
 
Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024
 
Corporate and higher education May webinar.pptx
Corporate and higher education May webinar.pptxCorporate and higher education May webinar.pptx
Corporate and higher education May webinar.pptx
 
Exploring Multimodal Embeddings with Milvus
Exploring Multimodal Embeddings with MilvusExploring Multimodal Embeddings with Milvus
Exploring Multimodal Embeddings with Milvus
 
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...
 
Architecting Cloud Native Applications
Architecting Cloud Native ApplicationsArchitecting Cloud Native Applications
Architecting Cloud Native Applications
 
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
 
Six Myths about Ontologies: The Basics of Formal Ontology
Six Myths about Ontologies: The Basics of Formal OntologySix Myths about Ontologies: The Basics of Formal Ontology
Six Myths about Ontologies: The Basics of Formal Ontology
 
AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of Terraform
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdf
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
Apidays New York 2024 - APIs in 2030: The Risk of Technological Sleepwalk by ...
Apidays New York 2024 - APIs in 2030: The Risk of Technological Sleepwalk by ...Apidays New York 2024 - APIs in 2030: The Risk of Technological Sleepwalk by ...
Apidays New York 2024 - APIs in 2030: The Risk of Technological Sleepwalk by ...
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024
 
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data DiscoveryTrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
 
Platformless Horizons for Digital Adaptability
Platformless Horizons for Digital AdaptabilityPlatformless Horizons for Digital Adaptability
Platformless Horizons for Digital Adaptability
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
 
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
 
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...
 

Biking on the edge - Jerome Mies - SRD23

  • 1. Jerome Mies (Lecturer/ Researcher at Hogeschool van Amsterdam) 23 May, 2023 Biking on the edge Monitoring the conditions of streets in Amsterdam with AWS IoT Greengrass
  • 3. Overview • What are we working on? • What is edge computing? • How do we use edge computing? • How do we use AWS? • How can we integrate AWS? • Example models
  • 4. Together improving the wellbeing in public spaces Hogeschool van Amsterdam Centre of expertise for applied artificial intelligence Smart Asset Management Lab SURF SURF is the collaborative organisation for IT in Dutch education and research Municipalities Velotech Solutions Sustainable object detections in the public space
  • 5. Conditions of the assets in the city impact our safety and wellbeing Light posts Traffic lights Garbage Traffic signs Road markings Road surface Trees 1
  • 6. Our ambition is to cover the complete city maintenance with LiDAR & video AI Video image recognition LiDAR mapping Light posts Traffic lights Garbage Traffic signs Road markings Road surface Trees 1
  • 7. Provide an accurate & complete overview of deficiencies in the city through eco-friendly inspections 3 Detailed routes per area Image analysis using algorithms Registration in eform-app Bicycles can go where cars can’t Validation of the inspections Integration with maintenance subcontractor systems Key features of the inspections Inspections (periodically) of designated areas by bicycle/vehicle Number of identified objects, defects & exact GPS locations Reporting & dashboard that is linked with the existing workflows Purpose Improve social & traffic safety Minimize liability risks Preventing peak workload for maintenance parties in winter
  • 8. Route planning Schedule riders Data gathering & mobile mapping Before & after inspections Data analysis Reporting municipality Reporting maintenance contractor Green and fast methods to get insights in the public space by using bikes & technology AI 2 Our approach:
  • 9. Overview • What are we working on? • What is edge computing? • How do we use edge computing? • How do we use AWS? • How can we integrate AWS? • Example models
  • 10. Edge computing is real-time computing at the end devices
  • 11. What are edge devices?
  • 12. A cloud computing provider can help integrate applications with edge devices • Cloud providers • Amazon AWS • Microsoft Azure • Edge Impulse • IoT (Edge) devices • Jetson Nano • Raspberry Pi • Arduino Nano
  • 13. Edge computing versus Cloud computing Advantages of edge computing • Real-time computing • Limited sending of information • Privacy
  • 14. Overview • What are we working on? • What is edge computing? • How do we use edge computing? • How do we use AWS? • How can we integrate AWS? • Example models
  • 15. Light post detection at the edge 4 Inspections on a bicycle Real-time inference on images to detect defects Bicycle-mounted edge devices operate autonomously.
  • 16. From light post detection to tilted street lights The angle of a light post detected in an image is determined in multiple steps. The angle of the light post is compared to the horizon. 5 Light posts in a training set are labelled using the Roboflow application and models are trained.
  • 17. Modular edge device • NVIDIA Jetson Nano or NVIDIA Xavier NX • 32 GB SD card • 4G and WiFi connectivity • Battery pack • Camera • GPS sensor and gyroscope
  • 19. Overview • What are we working on? • What is edge computing? • How do we use edge computing? • How do we use AWS? • How can we integrate AWS? • Example models
  • 20. • Deploy software to all remote devices at once from a central cloud location • Software versioning • CI/CD, Infrastructure as Code • Modular Greengrass components make it possible to design flexible architectures adapting to multiple use cases • Docker support • Dealing with unstable network connections reliably Managing remote devices with AWS Greengrass
  • 21. • Every Greengrass component acts as a microservice in a data pipeline, e.g. taking pictures with a camera, doing the AI inference, removing sensitive data, uploading to the cloud, etc. • Orchestrate event-driven workflows at the device level, with components talking over inter-process communication • Trigger or schedule workflows remotely over MQTT from AWS IoT Core • Address individual edge devices using a dedicated MQTT topic structure Greengrass components and event-driven workflow
  • 22. • Using publicly available models • Dedicated models to detect objects of interest, e.g. models from municipalities for garbage detection • Pre-trained models, such as YOLOv5 trained on a COCO dataset, to detect faces for further removal • Training custom models with transfer learning • Training data is collected from the bicycles and labelled manually. • Train models with Amazon SageMaker Studio • Track training experiments and model performance with MLflow Models and training
  • 23. • Model optimization is crucial for battery-powered edge applications. • Limited to GPU capacity of the device (NVIDIA Jetson Nano vs NVIDIA Xavier NX) • Faster models (and more powerful devices) will allow us to move from image processing to video processing to LIDAR processing at the edge. Model optimization and limitations of the edge
  • 24. • With remote device management, scaling towards multiple edge devices is easy. • Remote connectivity enables real-time updates and event-driven data processing. • Modular design allows us to accommodate multiple use cases. • Let’s make cities a safer place through cloud and edge technology and AI. • We have developed a blueprint for smart edge-computing applications for use in public sector and research. Cloud is the key to scalability at the edge
  • 25. We can use AWS to integrate with our dashboard for simpler use Integration AWS with a dashboard
  • 26. Overview • What are we working on? • What is edge computing? • How do we use edge computing? • How do we use AWS? • How can we integrate AWS? • Example models
  • 27. Examples of use cases for VeloTech • Road deterioration • Broken street lights • Tilted lanterns • Street signs pollution • Lidar scans
  • 28. Overview • What are we working on? • What is edge computing? • How do we use edge computing? • How do we use AWS? • How can we integrate AWS? • Example models
  • 33. Lidar scans – some ideas Tilted street lights Falling trees Collapsing quays (‘kades’) and bridges Mapping the city