– Ireland’s Centre For AI

Projects

Manolo Project

Project Status: Live Project

EU Programme: HORIZON.2.4 – Digital, Industry and Space

Start Date: 01/01/2024

Finish Date: 31/12/2026

Project Website: https://manolo-project.eu/

CeADAR is leading the Manolo Project, an €8.7m EU project, with 18 partners across 8 countries. The project’s overarching vision is to build a trustworthy and energy‑efficient AI stack, this means designing algorithms and tools that deliver high‑quality AI models while reducing energy use and ensuring compliance with upcoming EU AI regulations.

By combining hardware‑aware training optimisations, model compression and other approaches, MANOLO seeks to lower AI’s environmental footprint while increasing user trust. It also plans a handbook to help SMEs navigate the EU AI Act and ensure AI ethics compliance.

Objectives

Design a next-generation Hardware-aware in training optimisation for trustworthy efficient AI.

Deploy new guidelines for the implementation of trustworthy efficient AI systems.

Reduce environmental footprint, increasing trustworthiness & and edge autonomy.

Develop open-source and benchmarks for promoting excellence in ADRA communities.

Introduce new business models for cloud-edge continuum AI software and hardware.

Integrate with Horizon Europe projects, platforms (e.g., AI-on-Demand), and networks (e.g., ADRA-e).

Use Cases:

Industrial Robotics – In an agile manufacturing scenario, a mobile manipulator (PAL Robotics’ TIAGo robot) will use hybrid cloud–edge AI for human detection, object recognition and adaptive motion planning. The aim is to improve perception and safety while reducing hardware requirements.

Mobile Devices in Telecommunications – Computer‑vision models on mobile devices currently require uploading unencrypted images to centralised servers. MANOLO deploys a cloud–edge pipeline to process images locally, improving privacy and reducing latency

Companion Robots in Healthcare – A robot for assisted living will perform on‑board speech interaction, object recognition and adaptive motion; hybrid cloud–edge computing is expected to enhance detection accuracy and safety.

Wearables in Healthcare – Sleep‑monitoring devices will integrate MANOLO’s hybrid architecture to lower hardware costs by more than 50 % while achieving >90 % accuracy in detecting brain signals