Skip to main content

Robotics and Autonomous Systems Lab
at Vanderbilt

EEG-based Affective Computing

Personnel: Jing Fan

Goals/Objectives:

The goal of EEG-based affective computing is to incorporate implicit channel of communication as feedback to adapt human-machine interaction.

Outline:

Many systems rely on task performance, human postures as feedback to adapt human-machine interaction (HMI). Affective states such as anxiety, frustration, engagement, or boredom are important implicit channel of communication. With the ability to detect affective states, the systems are more aware of and responsive to users, and make HMI more natural and efficient. EEG-based brain computer interface may provide a non-invasive way to estimate the affective states and workload of users in order to enrich HMI by individualized system adaptation.

Several EEG features were extracted and analyzed for affective states and mental workload recognition, including the EEG engagement index, statistical features, fraction dimension features, higher order crossings (HOC)-based features, and power features. The EEG engagement index (EEI) were used to estimate older adult’s engagement intention during human-robot interaction. EEI were computed for ten older adults. We summarized the EEI trace for each activity and each older adult. There was a strong correlation between the summarized EEI and older adults’ self-rating of activity preference. EEG data collected from a virtual reality based driving task were used to build group-level classification models to recognize affective states and mental workload of individuals with autism spectrum disorder. The classification results imply that models based on EEG activations are able to detect with high accuracy the states of low engagement, low enjoyment, high frustration, and high workload. The most discriminative features for affect and workload recognition were extracted from frontal electrodes.

Publications:

  1. Jing Fan, Joshua W. Wade, Alexandra P. Key, Zachary E. Warren, and Nilanjan Sarkar, “EEG-based Affect and Workload Recognition in a Virtual Driving Environment for ASD Intervention,” IEEE Transactions on Biomedical Engineering, 2017. [http://ieeexplore.ieee.org/document/7898495/]
  2. Jing Fan, Joshua W. Wade, Dayi Bian, Alexandra P. Key, Zachary E. Warren, Lorraine C. Mion, and Nilanjan Sarkar, “A step towards EEG-based Brain computer interface for autism intervention,” in Engineering in Medicine and Biology Society (EMBC), 2015 37th Annual International Conference of the IEEE, 2015, pp. 3767-3770: IEEE. [http://ieeexplore.ieee.org/abstract/document/7319213/]