Before you entrust a decision to an algorithm, you need to know how it will behave. With ARENA, the platform developed by our researchers within the EU project TUPLES, we have made this assessment objective and repeatable, starting from a real planning challenge set by Airbus. The whitepaper covers the method and the results.
Research as the foundation of trustworthy AI
An algorithm creates value only when it is transparent and verifiable. In complex organisations, introducing a new one into key processes without first measuring its stability is a risk no decision-maker can afford.
That is why at Optit we see research and development as the engine of Decision Intelligence, not an academic exercise. It is also why we joined TUPLES, the EU project dedicated to Trustworthy AI. The goal is to define methods for testing an algorithm’s performance before it reaches the field, so decision-makers stay fully in control of their operations.
From theory to practice: ARENA put to the test with Airbus
Our research always becomes an operational tool. ARENA is a proprietary platform, designed by our researchers for those who develop mathematical models and need to validate them rigorously.
Within TUPLES, ARENA served as the core infrastructure for a competition built on a real Airbus use case. The challenge focused on managing and storing the aircraft components flown by the Beluga XL: a complex planning problem where efficiency, scalability and explainability matter equally. On ARENA, different approaches, from operations research to machine learning, were compared against shared, measurable criteria.
The result is an objective, standardised validation method. This is how we turn scientific complexity into industrial solutions.
A method to validate models and master complexity
The competition shows that our approach works on real industrial problems. With ARENA, we support companies, partners and research centres that want to run internal algorithmic competitions or assess new optimisation solutions more quickly, from machine learning to operations research.
The principle never changes: method first, implementation second. Because a reliable decision starts with a model whose behaviour you know in advance.
The whitepaper, available via this link, covers the method and the results of the competition.
Want to know how your decision models will behave before you adopt them? Get in touch: together, we will build the right validation path for your business.
