Smart Sensors HW/SW Interface based on Brain-actuated Personal Care Robot for Ambient Assisted Living

• Main Authors: Mezzina, Giovanni, De Venuto, Daniela
• Format: Conference Proceeding
• Language: English
• Subjects: BCI; Cameras; Electroencephalography; HRI; Intelligent sensors; Navigation; Personal Care Robot; Real-time systems; Robots; Sensors
• ISSN: 2168-9229
• DOI: 10.1109/SENSORS47125.2020.9278808

In this paper, a smart sensors system aiming to realize a human-robot interface (HRI) for Ambient Assisted Living is proposed. Specifically, this work proposes a direct communication pathway between the human brain and an external mechatronic actuator, by interlacing - among each other - heterogeneous sensors, worn by the human user, embedded on personal care robot (PCR) and distributed all around the environment where the user is living. In the here described case study, the patient/user wears a wireless headset for electroencephalography (EEG). Then, the acquired brain signals are wirelessly sent to a PCR that analyzes them via a Brain-Computer Interface (BCI). Specifically, the BCI exploits an end-to-end binary processing technique known as symbolization to extract features from the EEG data, through a fast classifier to discriminate user intentions and needs. BCI outputs are translated in specific autonomous navigation and object manipulation routines, which allow the PCR to satisfy the user requests. Both routines are based on the joint analysis of data from RGB cameras, 3D sensors, sonars, and IR sensors. In-vivo tests demonstrated that the PCR embedded BCI algorithm is able to decode EEG signals and send a command to the actuator in ~883ms while keeping accuracy in classification of 84%. The results demonstrate that the 75% of the requests formalized by the users are successfully satisfied by the PCR.