TRANGO. TRIP AND GO

Background

TRANGO is a service developed by the Agilia Center R&D team under the umbrella of the ARTICONF project and supported by the EGI-ACE project. 

This service is based on a  multi-platform mobile application which gathers all these functionalities through different roles. Powered by blockchain technology, the system will evaluate the service agreement and will release the money according to a decentralised decision based on different sources of information.

Supported by a machine learning  algorithm, the system will suggest to you the best location for your vehicle according to the demand prediction. This allows you to increase the profitability of your vehicle.

Moreover, TRANGO tries to reward users who change the previously established route to reduce the idle time of the vehicles.

Challenge

The objective of the use case is to create an AI-driven mobile app that mixes car sharing and carpooling models, allowing cost sharing and safe travel through direct trustworthy transactions among passengers.

There are two different objectives: 

• Behind the use case, and within the ARTICONF project, we have developed a blockchain network to register user behaviour and automatically evaluate the service through a smart contract. Since the use case has a demand peak at a particular time a day, we need to prove that the blockchain network is able to manage the amount of interactions needed to this end, guaranteeing high availability rates

• Apart from that, to efficiently allocate a fleet in a city, we need to demonstrate that our machine learning model is able to predict the demand in advance with enough accuracy (greater than 85% in the most demanding zones)

Work Plan

For the blockchain network we need to test the availability of services along with the maximum throughput achieved in the network:

In total we would need 5 VMs

• 4 VMs with the following requirements:
• 8GB RAM and 4 cores
• 100GB SSD

For the scalability of the APIs and to train the IA model along with the dataset for this training:

• 1 VMs
• 4GB RAM
• 2 cores
• 100GB SSD

To validate the machine learning model, we will use a different dataset depending on the city. The demand prediction must be performed for every city and this will be shown. We will provide this dataset coming from open data source like public transport datasets