Inspiration

Detailed information about infected people’s movements is crucial for tracking and controlling the epidemic as proven by South Korean experience. South Korean authorities launched a massive contact-tracing and testing regime to identify and then isolate infected people. https://coronamap.site/

Although the Korean model a few weeks ago raised the issue of privacy and anonymity (i.e. social exclusion), we believe that the reached global severity level and the related concern may put such issue in second place to the need to quickly zero the new confirmed cases.

The actual proposed systems to trace people in the COVID-19 setting, rely on the use of GPS data possibly integrated with other information (i.e. payments transactions, CC-TV camera, social app, ..) giving the possibility of mapping high risk “areas”. To ensure a targeted action there is the urgent need for high precise people localization (i.e. high risk buildings/stores/small rooms/..).

The proposed system offers a precise mapping enabling a reliable visualization of “people carrying COVID-19" around you anytime. If the app is adopted by the entire population (or most of) it will raise targeted alarm to avoid accidental contagions.

What it does

The Argus service aims at speeding up the virus confinement through citizens participation.

The proposed service aims at overcoming the current limitation of the existing COVID-19 fighting systems based on people tracing; by integrating instantaneous Bluetooth localization, Argus may reach a social distance level accuracy thus decreasing on one hand the false alarm probability of detecting and preventing high risk people encounterings, on the other hand, decreasing the related missing alarm probability thus activating a safe and fast mechanism to stop the unconscious virus spread.

The main objective will be reach by the following sub-objectives:

-by providing a social distancing scale monitoring of people location

-by relying on both real time and backtraced data to assign to everybody a calculated risk level

The service, if integrated with artificial intelligence algorithms (i.e. machine vision and machine learning) could also impact the following actions:

  • Police surveillance control automation: fast check if the user is on a previoulsy authorized route (i.e. children visiting, medical visits,..) or deduced through learning routes (i.e. path to work and back, local stores, dog walking,..)

  • Encouraging the use of Personal protective equipment (PPE) when outside or close to other people: the app, as soon as the user is outside, requests a selfie to perform a machine learning real-time PPE compliance monitoring

How we could build it

The proposed system enables a social distancing scale monitoring of people location by:

  • Locating global users position using GPS

  • Caching data inside the smartphone, uploading it when connection is available

  • Calculating proximity with other users using bluetooth

  • Automatic generation of graph to materialize users interactions

Technical strategies

*Database *

  • Location data is stored in cloud distributed database

  • Database is shared using small regions coordinates

  • Interactions with other users are stored in a graph database

  • Every «region» has a risk factor calculated on data provided by civil protection (number of confirmed case / total population in the zone / density)

Machine vision/Machine learning

When a user goes out from her/his home to buy food or whatever, the app asks her/him to make a selfie and with artificial intelligence it can determine if the user has PPE

Users that don’t agree to make selfies are classified as less collaborative and more at risk

Real time risk monitoring

The app can inform users in real time if they are close to people with a lot of interactions (multiplied by the risk factor)

Confirmed case

  • The app can track back (for example, up to 2 incubation cycles, 28 days) interactions the user has had

  • Warning can be managed by authorithy and/or users themselves

Certified moving

  • App learns routes, for example going to work or going to buy foods, walking the dog

  • Police can use the app in checkpoints and manage people in a matter of seconds

Share this project:

Updates