Robotics

Maestro: developing the future of smart and adaptive robotics

Date:
Changed on 11/02/2025
A major transformation is underway in the field of robotics, driven by the need to design smarter and more adaptive systems capable of being used in real-life contexts. Whether it’s for driverless vehicles or humanoid robots, the demand for advanced control architecture - integrating traditional optimisation and modern artificial intelligence - has never been greater. Maestro, a consortium led by Inria, is seeking to meet this challenge head-on.

Bringing robots out of the factory: the challenge facing Maestro

Locomotion et perception - Image créée avec Midjourney

The goal is clear: to take robots out of strictly controlled environments such as factories and warehouses and put them alongside humans in complex and dynamic environments. 

In such settings, robots must have the capacity to adapt to their surroundings, and must also be able to carry out a wide variety of different tasks.

In response to these challenges, the Inria-led consortium Maestro is seeking to bring together research teams and industrial partners to work on the development of differentiable control architecture, using the software program Pinocchio as its cornerstone.

Imagining intelligent control systems for tomorrow's robotics

Robotic systems work very well in factories as these are highly regulated environments where robots are required to perform the same task over and over again. But given the difficulties these robots have in perceiving, interpreting and adapting to their surroundings, or executing a sequence of different tasks, it is not yet possible to get them interacting with humans or complex environments. 

This is down to the limitations of their programming, which is centred around a strict hierarchical decision-making process. Adapting these robots to variations in their surroundings is very difficult, thereby limiting their flexibility and versatility. Modern robots also lack the capacity to learn from past experiences and then use that information to adapt to new situations.

          Justin Carpentier, head of the research team Willow

Key areas for Maestro

Manipulation au contact - image générée par Midjourney.
  1. Next generation differentiable simulatorsMaestro will develop high-performance simulators incorporating optimisation and machine learning, ensuring real-time adaptability.
  2. AI-based optimisation tools - from model predictive control (MPC) to advanced solvers, will develop state of the art software for the deployment of end-to-end differentiable architecture.
  1. An open and collaborative ecosystem - Maestro will put an emphasis on collaboration and openness, ensuring the tools developed are accessible worldwide. 

Maestro is seeking to overcome these longstanding issues through the development of new differentiable control architecture, a concept that combines physical models with data-driven machine learning. The aim of this is to create robotic systems that are quicker, more accurate and capable of adapting to dynamic environments in real time.

An open software platform to drive research and innovation in robotics

By taking an open-source approach, Maestro is seeking to encourage collaboration between academia, startups and major industrial stakeholders. This spirit of openness is designed to ensure that the tools developed are accessible to as many people as possible, stimulating innovation in a wide range of sectors - from industry and logistics to agriculture and personal support - while creating new opportunities to improve people’s everyday lives.

As Justin Carpentier explains, “Open-source software programs developed within research laboratories are now used on a mass scale all over the world for applications in robotics. One of the programs developed by the Willow research team is Pinocchio, which is widely used with quadruped and humanoid robots. Scientific breakthroughs made as part of research projects such as Agimus or ARTIFACT are directly integrated into our software.

We also have access to a cutting-edge robotics platform, funded in part by the DIM AI4IdF[1] and the AI research institute PRAIRIE, enabling us to test our software on a wide range of robots, whether these are quadrupeds, dexterous robotic hands or humanoid robots.” 


[1] A ‘domaine d’intérêt majeur’ (English: ‘key field of interest’) or DIM is a research network given funding by the Ile-de-France region for a period of four years.

Four software programs for coordinating robot movement and intelligence

The progress made in robotics is centred around innovative software tools which enable robots to understand their surroundings, plan their movements and interact effectively with the world around them. Here are four key software tools from Maestro’s software stack:

More about Maestro

Maestro consortium website.

Contacts :

The official launch of the Maestro consortium in 2025 will mark the beginning of an ambitious project aimed at shaping the future of robotics. Maestro presents researchers, developers and industrial stakeholders with an opportunity to contribute to a revolutionary movement in the world of robotics and artificial intelligence.

          Pierre-Guillaume Raverdy, consortium creation coordinator.

 

See also

Truly autonomous robots are just around the corner

Tiago, un robot de dernière génération mis à disposition des partenaires du projet européen AGIMUS par l’entreprise barcelonaise PAL Robotics.

Pinocchio, the software that brings robots to life

Manipulation robot dans le cadre du projet européen euROBIN

euROBIN: a European network for excellence robotics

Robot Tiago

AGIMUS: a collaborative EU project aimed at simplifying the integration of robots

TIRREX: an accelerator for French robotics research

How research is pushing the boundaries of robotics