Luca Tonin: Treating Brain-Computer Interfaces as a neuroprosthesis

Creating Brain-Computer Interfaces (BCI) that are reliable in everyday situations is a core part of developing a device that’s right for each user. Luca Tonin, Associate Professor and leader of Team WHi, tells us more about why user learning is so important for developing BCI technology, and how CYBATHLON’s tasks can help to understand those needs.

Hello, Luca! Can you tell us a bit about your research and background?                                              

I received my PhD in robotics from the École Polytechnique Fédérale de Lausanne (EPFL), Switzerland, in 2013. I am now an Associate Professor of Computer Engineering in the Department of Information Engineering at the Università degli Studi di Padova. My research focuses on exploring advanced techniques for Brain-Machine Interfaces (BMI) and their integration with intelligent robotic assistive devices.

What’s the ultimate goal for the WHi Team?

Our primary goal is to demonstrate that BMI technology is mature enough to be used outside the lab and in daily life conditions. In particular, my research revolves around the hypothesis that user learning is key to making BMI systems robust and reliable.

In my view, BMI and its users are two symbiotic entities, and investigating their interactions and understanding how they learn from each other are fundamental aspects of BMI research. This is why the WHi Team's approach to the CYBATHLON focuses on promoting the learning aspects rather than relying on advanced processing or machine learning methods to decode users' intentions.

You've been involved with CYBATHLON since 2016, can you tell us how you've seen the competition develop? What are the current challenges in BCI research and development?

Since 2016, the challenges in the BCI discipline have constantly increased.

We started with three discrete commands required to control a BCI game and progressed to more sophisticated control strategies based on the integration of continuous and discrete control, similar to using a joystick. The tasks have also become more reflective of real-life scenarios, such as controlling a wheelchair, a robotic arm, or a cursor on a screen.

However, despite the evolving nature of the tasks in CYBATHLON, the core challenge in BCI research remains the same: making it work in every condition and scenario.

Using a BCI in a laboratory to move a bar left to right is very different to using the same technology to accomplish a real task in daily life. BCIs need to be more robust and reliable across all circumstances, which might be achieved by enabling the user to learn how to use the technology and supporting them in this.

How can research in BCI technology benefit from the rapid progress we have seen in AI technology in the last two years? Are there any areas in particular where AI is more useful?

Artificial intelligence (AI) is an important and fascinating research area that can be applied in several fields.

However, I consider BCI as a neuroprosthesis that enables a new and alternative communication channel between the user and the world around them. For this reason, the most important goal is to understand how the user adapts their neural patterns to this new output channel to properly decode their intentions. While AI methods can support us in decoding, I believe they cannot directly help us understand these learning interactions.

From my point of view, the key research question is how to merge AI with human intelligence, especially when BCI is used to control intelligent robotic devices. This is an extremely interesting field that can help make the use of BCI-driven robotic devices—such as smart wheelchairs, telepresence robots, and upper- and lower-limb exoskeletons—more reliable in daily tasks.

One of your most cited papers is from 2015: “Towards independence: a BCI telepresence robot for people with severe motor disabilities”. Can you tell us about the progress made in this area by your team and the wider field? Where are we today?

In this paper, we demonstrate two important facts: first, integrating robotic intelligence, to take care of all the low level details of navigation, into BCI-driven robots helps increase the reliability of the entire system; second, this integration of user and robotic intelligence allows people with severe disabilities to mentally drive a telepresence robot from their home.

In recent years, the field has advanced in this direction, with more studies involving end-users and taking place outside the lab. I believe this is crucial for the field's advancement. Currently, we are investigating how to make BCI ready for extensive use outside the lab.

A few years ago, we demonstrated that 'learning' was the key to our past successes in the CYBATHLON events [Perdikis et al., 2018; Tortora et al., 2022]. Additionally, we showed that end-users can successfully drive a smart wheelchair in a clinical environment using a non-invasive BCI [Tonin et al., 2022] and that the integration of robotic and user intelligence enables walking with a lower-limb exoskeleton in the presence of obstacles on the ground [Trombin et al., 2024].

How do you think events like CYBATHLON help to promote research and development in key areas, such as BCI technology?

I believe that the CYBATHLON was—and continues to be—fundamental in advancing BCI technology research. The key contribution of the CYBATHLON is that it pushes teams to make their systems work in challenging scenarios like the arena in Kloten.

This is a radical departure from traditional BCI experiments, where systems are often evaluated in fully controlled environments. And the competitive nature of the event forces teams to consider both the technical reliability of their BCIs and the proper training of their pilots to overcome the obstacles in each task.

Together, these factors have highlighted important aspects that were often neglected in the past, such as the absence of technical failures, ease of software use, minimisation of hardware requirements, and, most importantly, user-centred design of BCI systems.

–

Do you want to see Team WHi and WHi Students in action? They will be competing in CYBATHLON 2024 at the Swiss Arena, Kloten from 25 October - 27 October.  Make sure to grab a ticket early so you don’t miss out! https://cybathlon.ethz.ch/en/cybathlon-2024/ticketing