Use Cases

Smart city
cross­ing safety

Ped­es­tri­ans are nowadays more eas­ily dis­trac­ted, giv­en the increase of inform­a­tion com­ing at them from dif­fer­ent sources. The best example is of smart­phone activ­it­ies while walk­ing. These kinds of dis­trac­tions are com­mon, as ana­lysed by the Euro­stat.

Improv­ing the citizen’s safety in cross­roads is of par­tic­u­lar import­ance for vari­ous muni­cip­al­it­ies, and also a main con­cern in the EU.

The object­ive of this use case is to increase the ped­es­tri­an cross­ing safety by lever­aging the IoT and edge infra­struc­tures of a smart city.

This approach focuses on equip­ping a num­ber of ped­es­tri­an cross­ings with devices exist­ing on the mar­ket that enable mon­it­or­ing of ped­es­tri­ans intend­ing to cross the road.

The object­ive of the use case is to spot any poten­tial dangers that might be nearby put­ting their safety at risk and provide means for noti­fic­a­tion alerts.

pic­ture of the cross­ing where we will set up our pilot (before)

Fea­tures
from DECENTER’s archi­tec­ture

  • Resource orches­tra­tion
    (Ver­tic­al)
  • Pri­vacy-pre­serving AI

  • Hier­arch­ic­al /distributed AI

  • Digit­al twin

  • AI Mod­el repos­it­ory
  • Multi-tier fog com­put­ing plat­form

Smart city cross­ing design

In sum what we would like to achieve with this use case is to cre­ate a solu­tion to help people cross the road safely.

And also we aim to help DECENTER to test dif­fer­ent func­tion­al­ity on our pilot. (resource orches­tra­tion, pri­vacy pre­serving AI, hier­arch­ic­al /distributed AI, and digit­al twin)

pic­ture of the cross­ing where we will set up our pilot (after)

DECENTER bene­fits
before & after

BEFORE DECENTER
AFTER DECENTER
DECENTER BENFITS
Video/audio streams are sent to the cloudVideo/audio streams are sent to the cloud
The video/audio streams can be elaborated in the edge, no sensible data are sent to the cloud
Privacy preserving: pre-elaborate video/audio on the edge reduces the amount of data sent to the cloud, that are anonymized allowing to sent on the web only of protect data
Difficult to manage edge resources
Resources can be managed from cloud to the edge within the DECENTER fog platform
Multi-tier fog platform: easy way to access and manage the resources from cloud to edge
A new object cannot be recognized due to not up-to-date AI model
A new pre-trained AI model can be easily retrieved by each service from the DECENTER model repository
AI Model Repository: allow the system to have update and trained AIs from well-known sources
Pedestrians are alerted with too much delay
The services are moved from cloud to edge reducing application latency
Vertical resource orchestration: allow the system to react to eventually latency problem choosing the most efficient way to deploy the services
Diffucult to know what’s happening at the pedestrian crossing
Digital Twin is integrated in DECENTER and allows have a representation of the pedestrian crossing
Digital Twin: have a constantly update replica of the pedestrian crossing available to remote

Robot­ic
Logist­ics

Great num­bers of companies/organizations accom­mod­ated in small build­ings involve a vast amount of mater­i­al trans­port through hall­ways, on elev­at­ors, in base­ments and to customer/patient units.

Logist­ic trans­port­ing robots are used on big hos­pit­als, malls and indus­tri­al areas, but there is not any cost-effect­ive autonom­ous logist­ic robot­ic sys­tem really adap­ted to small res­id­ences, ware­houses or medi­um sized indus­tri­al facil­it­ies.

The object­ive of this use case is to test a new, cost-effect­ive, robot­ic indoor trans­port solu­tion that will be spe­cially suited for ware­houses and will auto­mate the trans­port pro­cess and free work­force for tasks that entail high­er added value. To this end, the use case will per­mit the incor­por­a­tion of the swarm robot sys­tem from Robot­nik into the cloud/edge sys­tem ser­vices, allow­ing enhan­cing the func­tion­al­ity of the robots by the use of Edge Com­put­ing and a cent­ral­ized Cloud.

Fea­tures
from DECENTER’s archi­tec­ture

  • Resource orches­tra­tion
    (Ver­tic­al)

  • Pri­vacy-pre­serving AI

  • Hier­arch­ic­al /distributed AI

  • Digit­al twin

This use case envis­ages demon­strat­ing the applic­ab­il­ity of DECENTER plat­form, Edge and Cloud-to-Things Con­tinuum devel­op­ments, to the field of robot­ics as a mech­an­ism that allows rich­er inform­a­tion shar­ing and com­pu­ta­tion­al sup­port

DECENTER bene­fits
before & after

BEFORE DECENTER
AFTER DECENTER
DECENTER BENFITS
Robot software on bare-metal
Robot software in containers
More homogeneous deployments and the ability of rollback and easy remote updates
Robot only runs its processes
The robot processes can be places over the robot fleet or the edge device
CPU load decreases and the battery life improved
AI should be developed and trained by the robot provider
AI packages and models available from a repository
Use AI models from well-known sources
Use AI models from well-known sources
The AI recognize the objects of the images
The fleet manager system makes a smart decision in each case

Smart and safe
con­struc­tion site

Con­struc­tion is a very dynam­ic pro­cess. Each build­ing pro­ject is unique and usu­ally requires the col­lab­or­a­tion of sev­er­al com­pan­ies and act­ors.

Due to its very dynam­ic nature, it is a chal­len­ging engin­eer­ing work to organ­ise, mon­it­or, and imple­ment a con­struc­tion pro­ject includ­ing the vari­ous safety, secur­ity, logist­ics, inspec­tion and oth­er aspects, which require spe­cif­ic inform­a­tion sup­port.

The goal of the “Smart and safe con­struc­tion” use case is to explore mech­an­isms for inform­a­tion gath­er­ing, fusion and enrich­ment, which can provide intel­li­gence dur­ing the con­struc­tion pro­cess and help improve vari­ous aspects of the work.
Col­lect­ing rel­ev­ant inform­a­tion related to the con­struc­tion pro­cess, can be used for both time-crit­ic­al oper­a­tions and longer-term logist­ic and oth­er oper­a­tions.

design study of a pilot

Fea­tures
from DECENTER’s archi­tec­ture

  • Resource orches­tra­tion
    (Ver­tic­al)

  • Algorithms for QoS assur­ances, rank­ing and veri­fic­a­tion of Cloud deploy­ment options 
  • Pri­vacy-pre­serving AI

  • Hier­arch­ic­al /distributed AI

  • Digit­al twin

Use Case Pro­cess View

UC related lec­tures from Vlado Stankovski (UL) at World Con­struc­tion For­um 2019, Build­ings and Infra­struc­ture Resi­li­ence, April 8–11 2019, Ljubljana, Slov­e­nia

THEME 2
Con­struc­tion 4.0 – Advanced Con­struc­tion Engin­eer­ing:


Build­ing Smart And Safe Con­struc­tion Sites With Depend­able Decent­ral­ised Arti­fi­cial Intel­li­gence Applic­a­tions

Work­ers at a con­struc­tion site without DECENTER.
Work­ers at a con­struc­tion site with safety hel­met and without pro­tec­tion vests detec­ted.

DECENTER bene­fits
before & after

BEFORE DECENTER
AFTER DECENTER
DECENTER BENFITS
AI applications had to be manually deployed or setup on computing resources
AI applications are automatically deployed on computing resources that satisfy high QoS requirements
QoS-aware orchestration of resources
Data from construction sites was stored and processed in remote Cloud infrastructures that resulted with high computing latency and bandwidth limitations
Decentralised edge/fog computing placed between the sources of data and the Cloud
Effective and timely computation with improved network and computing performance
AI models had to be developed and trained by the construction site owner
A variety of AI models are available from a repository
Select AI models that are suitable for specific cases from an AI model repository
No known use of blockchain for data management at construction sites
Use blockchain’s Smart Contracts to support privacy preservation that will manage access to AI models
Smart Contracts facilitate trustful and secure access to the AI models and support privacy preservation regulations and certifications

Ambi­ent
Intel­li­gence

The focus of IoT-based ser­vices is mainly lim­ited to remotely mon­it­or­ing the cur­rent situ­ation using devices such as mobile phones, and these ser­vices are typ­ic­ally in the cloud. Round-trip delay caused by data trans­fer to the cloud may not be suit­able for real-time ser­vices. In addi­tion, there may be pri­vacy issues when upload­ing video streams to the pub­lic cloud.

In this use case, we test the mem­ber veri­fic­a­tion ser­vice at the edge using AI mod­els without send­ing any per­son­al inform­a­tion to the cloud.
This use case will show the main fea­tures of DECENTER based on an AI applic­a­tion for ambi­ent intel­li­gence.

This applic­a­tion checks the face of users vis­it­ing a cer­tain space and veri­fies wheth­er the per­son is author­ised to con­sume cer­tain con­tent in that space or not. For this use-case, the edge will use two veri­fi­ers veri­fy­ing each group mem­bers respect­ively.: A group veri­fi­er, and B group veri­fi­er. We assume that only these two groups are tar­get­ing to see spe­cif­ic con­tent. Thus the pro­cesses at the edge can veri­fy wheth­er the vis­it­or of a cer­tain space can con­sume cer­tain con­tent or not in that space, without shar­ing per­son­al inform­a­tion with the cloud.

design study of a pilot

Fea­tures
from DECENTER’s archi­tec­ture

  • Resource orches­tra­tion
    (Ver­tic­al)

  • Pri­vacy-pre­serving AI
  • Hier­arch­ic­al /distributed AI

  • Digit­al twin

In sum what we would like to achieve with this use case is to cre­ate a solu­tion to veri­fy mem­ber­ship at any edges without addi­tion­al per­son­al inform­a­tion.

After each edge has cre­ated an AI mod­el that iden­ti­fies spe­cif­ic group mem­bers and registered it on the cloud plat­form, it is easy to reuse this mod­el at oth­er edges without private inform­a­tion.

DECENTER bene­fits
before & after

BEFORE DECENTER
AFTER DECENTER
DECENTER BENFITS
AI software on bare-metal
AI software in containers
More homogeneous deployments and the ability to roll back and easy remote updates
Stand-alone AI application
Distributed AI with Microservice architecture
Can provide a more flexible and complex service
No AI model file management
AI Model Repository provides a structured way to manage AI model file
Maintenance of AI model file
AI model serving application
End-to-end application architecture with a combination of multiple AI methods
Increase the re-usability of AI method