Millions of people worldwide suffer from reduced motor abilities resulting from strokes, brain tumour surgery or accidents. In most cases, prosthetic and assistive devices are the only treatment option. However, current neuroprosthetics are limited in function as they replace only single motor impairments. But this looks set to change with an exciting new initiative called the MAIA project. This newly funded initiative aims to bring about a paradigm shift in human-AI interaction by designing human-centric artificial intelligence (AI) to control prosthetic and assistive devices.
By using human-centric AI the researchers will endeavour to develop prosthetic and assistive devices, such as robotic arms, wheelchairs and exoskeletons, that are multifunctional, adaptive, interactive and trustworthy – in other words intelligent. In this way, the researchers want to improve the way AI controls these devices, where human intention takes centre stage.
As the users’ experience is key to this project, it is therefore focused on building trust with the device through natural interaction and mutual learning. For this reason, the end-users – patients and caregivers – will be directly involved in the research programme.
The MAIA (Multifunctional, Adaptive and Interactive AI system for Acting in multiple contexts) project, which was recently awarded €4 million by the European Commission’s Horizon 2020 funding initiative, proposes a paradigm shift where human-centric AI will control prosthetic and assistive devices.
By investigating and resolving the crucial aspects of this technological advance, the project will develop a novel ‘intention-foraction’ decoding algorithm that extracts information from multiple perceptual and motor processes revealed in recent discoveries in neuroscience and implement a new scheme for trustworthy human-AI interactions. Additionally, they will develop a new type of database for acquired information from multiple sources.
The technology proposed by MAIA is based on a flexible and interactive decoding of human intentions. To provide the best possible results, researchers will investigate the principles underlying natural, fast and effective communication between humans and AI and explore new forms of combinations of neural and behavioural signal processing beyond current state-of-the-art AI.
The researchers say MAIA technology will be interactive, encouraging the confrontation between AI decision, user intention and real needs. Natural and fast communication as well as new forms of neural and behavioural data combinations will be identified, moving beyond current data processing methods. Such principles will form the core idea around which the MAIA ecosystem will be developed, laying the foundations for biomedical human-centric AI.
Decoding human intentions
MAIA AI technology will decode human intentions and communicate the decoded targets to assistive devices and to the users, to ensure compliance and develop trust through natural interaction and mutual learning. The technological outcome, the researchers say, will be a multifunctional human-centric AI controller at Technology Readiness Level (TRL) 4 with embedded trustworthy characteristics, suitable to be integrated in robotic arms, wheelchairs and exoskeletons. TRL4 requires that the technology has been validated in a lab setting.
With the new funding the MAIA project got underway in January this year and will run until December 2024. It is a cross-disciplinary collaborative project, which is being co-ordinated by the University of Bologna with participating researchers in Germany, Spain and Italy.
Additionally, under the MAIA project a European innovation ecosystem beyond the research labs will be established to stimulate related innovative enterprises.