RideCare

Monitor a car sharing fleet in real time

and watch over occurred anomalies

Get Started

Illustraton by Kemal Sanli

Manage your fleet

Receive real time notifications

Locate the anomalies

Confer past anomalies

Illustraton by Aleksandar Savic

Find out about inadequate use of your vehicles

It’s very challenging to keep track of the state of multiple vehicles from a fleet. Assurance of good conditions between usages is fundamental for any Car Sharing service’s reputation. With RideCare you can monitor a car-sharing fleet in real-time and watch over occurred anomalies. Inside the web application, you have access to a dashboard in which you can check all the past anomalies and the corresponding evolution of the sensorial data before, during, and after its occurrence. Once an anomaly occurs, you will be notified and every information about it at your disposal.

Normal

Even in the absence of anomalies, data is still collected and persisted to further improve the machine learning algorithms.

Smoke

Car sharing vehicles are not a place where you can smoke. Whenever RideCare detects smoke inside a vehicle, that information will be stored in our infrastructure.

Stink

RideCare is also capable of evaluating the presence bad smells inside the vehicle, mainly trash related.

Supervised and unsupervised Machine Learning Algorithms

RideCare takes advantage of real-time classification algorithms, based on classic Machine Learning methods to detect anomalous situations inside shared vehicles. From labelled and unlabelled data, several algorithms from Supervised and Unsupervised Learning paradigms were tested and insights were taken during that extensive exploratory work. Our best algorithm achieved more than 97% accuracy on all predictions made, for all scenarios considered*. Additionally, other supervised learning algorithms, which performed well during our research process, are hosted by a cloud application -AlertAI Cloud, allowing alternative predictions for the raw data collections.
*These results were obtained under controlled capture scenarios, taking into account real environments inside vehicles.

Illustraton by Chris Randalls

Dataset composed by data collected from real scenarios

In car system installed

Real time data crawling

Real time data classification

A data crawler script and the machine learning algorithm resides inside the vehicle and does the evaluation locally. Whenever our machine learning algorithms detect anomalies in the sensorial data collected by our sensors, that information will be sent to our servers.

No network coverage? We've got you covered!

A queue system was implemented in order to keep function even if the car is in an area with no network coverage.

Illustraton by Maryam Ebrahimi

Robust, fail tolerant
cloud-based infrastructure

Client Application demonstration video

Meet the team

André Coutinho

Gabriela Martins

João Costa

José Pinto

Leandro Costa

Luís Correia

Paulo Jorge

Pedro Barbosa

Rafaela Soares

Rui Santos

Tiago Fontes

Other resources

Repository


Technical Report


Machine Learning Report