B-RELIABLE: Boosting reliability and interaction on brain-machine interface systems integrating automatic error-detection
Project name: B-RELIABLE: Boosting reliability and interaction on brain-machine interface systems integrating automatic error-detection
Description: The occurrence of errors during human-machine interaction is inherent to all interaction systems and may be due to low system reliability or user’s error (e.g, due to attention lapses, misinterpretation, control imprecision). Automatic error detection may not only contribute to a significant increase in the reliability of HMI systems, but also be used to establish a primary communication channel, or as a mechanism to automate neurofeedback for rehabilitation. The project B-RELIABLE aims to deepen the investigation of the use of an electroencephalographic (EEG) signal called Error Related Potential (ErrP), which occurs naturally in the brain as a response to unexpected errors. Complementarily, we propose to search for other EEG error correlators based on neural connectivity, and also to predict emotional states propitiating the occurrence of error, detected from changes in the autonomic nervous system, (e.g., looking at galvanic skin response (GSR) and electrocardiography (ECG) bio-signals). The first research line comes in the follow-up of results already achieved, seeking to contribute to the improvement of mobility of people with severe motor disorders (with reduced or no muscle activity). We are now concerned to further improve the reliability/usability of BMIs for communication and control, with particular focus on steering an intelligent wheelchair in real-world daily tasks. BMI provides information that is sparse in time and with low reliability. Combining automatic error detection with collaborative control, one can provide increased reliability to achieve safe and effective navigation. The second research line of the project B-RELIABLE aims to investigate how ErrPs can be used in applications beyond motor disorders, namely: 1) car driving scenarios, for validation of advanced driving assistance systems, seeking how error detection can boost interaction or improve driving safety (collaboration with ALTRAN-PTcompany); and 2) neurofeedback paradigms, to improve cognitive function, researching how error-detection can be embedded in frameworks to improve error monitoring and goal-oriented behavior to boost neurorehabilitation through operant learning (e.g., children with neurodevelopment disorders such as autism). Project B-RELIABLE comprises 4 TASKS: TASK1 – SIGNAL PROCESSING AND CLASSIFICATION ALGORITHMS TO IMPROVE RELIABILITY IN BCI Focused on the development of signal processing and machine learning algorithms to improve the detection of event related potentials (e.g., Error Potentials) at a single trial level, improve BCI reliability and transfer rates and reduce or eliminate BCI calibration. TASK2 – A BRAIN COMPUTER INTERFACE TO TRAIN ATTENTION AND INHIBITORY CONTROL DURING ERROR MONITORING USING OPERANT LEARNING STRATEGIES Focused on the design and validation of BCI interfaces to test and train error monitoring capabilities in patients with Autism and ADHD. TASK3 – HUMAN-ROBOT COLLABORATIVE NAVIGATION Focused on the research of new human-robot collaborative navigation approaches to increase mobility of people with severe motor disabilities requiring a brain-actuated wheelchair to move in home settings. TASK4 – HUMAN-MACHINE INTERACTION IN CAR DRIVING Focused of on the use automatic error detection based on ErrPs to provide the necessary input for the validation system to evaluate the response of advanced driving assistance systems (ADAS) decisions. Task in collaboration ALTRAN-Portugal.
PI: Gabriel Pires
Budget: FEDER/OE through operational programs CENTRO2020 and FCT
Partners: Institute for Systems and Robotics-Coimbra (ISR-UC), University of Coimbra-ICNAS (UC- ICNAS) and Polytechnic Institute of Tomar-VITA (IPT- VITA)
Duration of the action: 2018-06-20 - 2021-06-19