Truly autonomous robots are just around the corner

The ERC Starting Grant that you have just been awarded will fund the “Artificial Motion Factory” (ARTIFACT) project, designed to push back the current boundaries of robotics. What are the current limits?

Today, robotic systems work very well in factories because they are highly controlled environments in which robots are required to perform the same task over and over again. However, getting these robots to interact with humans or complex environments remains impossible, as they have difficulty in perceiving, interpreting and adapting to their environment, or in performing a sequence of different tasks. 

This obstacle is due to the limitations of their programming, which is based on an extremely hierarchical decision-making process. The information sent by the visual module to the planning and control modules is often limited to a certain context in order to carry out a certain task. It is very difficult to adapt it to variations in the environment, which limits the flexibility and agility of robots. Today's robots are also unable to learn from past experiences and to use this information to adapt to new situations. 

How will your project overcome these obstacles?

To make robots truly autonomous, dexterous and agile, we will use a combination of optimal control theory, which we will adapt to the specific context of robotics, and deep learning, exploiting existing technological building blocks to enable robots to fully perceive and understand their environment. 

Our aim is to develop software that can be adapted to any robot, combining enhanced environmental perception, rapid decision-making and machine learning. In other words, we want to lay the algorithmic foundations for a genuine artificial intelligence of motion. 

What assets can help you meet this challenge?

The ERC Starting Grant of €1.5 million will enable us to recruit up to ten PhD researchers, post-doctoral researchers and engineers over the five years of the project, which is due to start in 2025. We will also be drawing on the recognised expertise of the Willow joint project team, whose open source tools such as Pinocchio are already being used to model numerous robots around the world. To further increase our capacity to meet such an ambitious challenge, we are currently creating an open source software consortium in order to promote our scientific and technical advances directly to the different academic and industrial players.

We also have a robotics platform that will enable us to experiment with our software and test it on a wide range of robots (quadrupeds, dexterous hands, humanoids, etc.). Finally, we will be capitalising on the issues emerging from the European AGIMUS project, in which I am also involved, and which aims to improve the versatility of robots in industrial environments.

What impact could this future software have on society?

The software we are developing at ARTIFACT will be open source, so that it can be used by all robot manufacturers, opening up opportunities in the medical, economic and industrial sectors. This will make it possible to envisage autonomous robots operating in hospitals, at home... anywhere, in fact! Robots will finally emerge from controlled environments to interact fully with humans.