Open Source Software Hub

The MobiSpaces project has continuously generated open-source software and data. This page brings together all the code repositories where the MobiSpaces partners have stored the code of the open-source software they have developed. With a commitment to transparency and accessibility, this page serves as a gateway, allowing stakeholders to explore, contribute to, and benefit from the resources cultivated by the MobiSpaces project.

Mobility XAI

A comprehensive eXplainable AI (XAI) system for maritime ship trajectory prediction, consisting of three microservices that provide end-to-end solutions for predicting and explaining ship movements.

Access the resource

Time Series XAI Docker Service

A comprehensive Dockerized Flask service that provides explainable AI (XAI) capabilities for time series analysis using multiple explanation techniques including GradCAM, LIME, and SAX-based text explanations.

Access the resource

Findability

The Findability Component supports both back-end and front-end operations for the Creation and Storage of mobility related metadata, implementing a semantic meta-model within the MobiSpaces Data Governance Platform. Metadata is stored in the Virtuoso database in RDF format to be easily retrieved using sparql queries in compliance with FAIR principles.

Access the resource

MobiML

MobiML is the first open source framework that provides modular components specifically for mobility AI development. These components significantly simplify the development of mobility AI solutions and have large potential to improve the efficiency and reliability of scientific software for mobility applications.

Access the resource

MobilityNebula

An open-source trajectory data streaming platform built on NebulaStream. It leverages MobilityDB temporal and spatio-temporal datatypes and functions via the MEOS middleware to enable real-time ingestion, processing, and analysis of moving-object data in a high-performance IoT stream processing system.

Access the resource

FedVRF

FedVRF is an efficient LSTM-based solution for Vessel Route Forecasting trained in both centralized and cross-silo Federated Learning (FL) paradigms. To address the limitations of edge devices, a cross-device FL variant of FedVRF was developed and deployed at the MarineTraffic edge devices. A Visual Analytics framework tracks real-time AIS activity and uses FedVRF to visualize forecasted routes.

Access the resource

TrajParquet

TrajParquet is a column-oriented format file prototype for storing and querying trajectory data. The format enables efficient accessing of trajectory data as they are stored indexed. For this purpose, it exploits the trajectory segmentation technique and groups groups spatio-temporal positions to generate compact data pages that increase the opportunity for data skipping.

Access the resource

PyMEOS

PyMEOS provides Python bindings for the MEOS C library, wrapping all of MEOS temporal and spatiotemporal data manipulation functionalities in Python classes. It allows developers to seamlessly integrate MobilityD trajectory processing capabilities into Python workflows and applications.

Access the resource

MobilityDB

MobilityDB is an open-source extension to PostgreSQL/PostGIS that adds support for temporal and spatiotemporal datatypes and functions for moving-object trajectories. It offers compact storage, a full SQL interface for mobility analytics, and compatibility with the PostgreSQL ecosystem for large-scale trajectory data management.

Access the resource

MobiSpaces Privacy Aware Visualization (PAVA)

Access the resource