Trust Recognition in Human-Robot Cooperation Using EEG

TL;DR

The study proposes an EEG Vision Transformer model coupled with a 3-D spatial representation to capture the spatial information of EEG, taking into account the topological relationship among electrodes, which outperforms baseline models in both recognition accuracy and generalization.

Abstract

Collaboration between humans and robots is becoming increasingly crucial in our daily life. In order to accomplish efficient cooperation, trust recognition is vital, empowering robots to predict human behaviors and make trust-aware decisions. Consequently, there is an urgent need for a generalized approach to recognize human-robot trust. This study addresses this need by introducing an EEG-based method for trust recognition during human-robot cooperation. A human-robot cooperation game scenario is used to stimulate various human trust levels when working with robots. To enhance recognition performance, the study proposes an EEG Vision Transformer model coupled with a 3-D spatial representation to capture the spatial information of EEG, taking into account the topological relationship among electrodes. To validate this approach, a public EEG-based human trust dataset called EEGTrust is constructed. Experimental results indicate the effectiveness of the proposed approach, achieving an accuracy of 74.99% in slice-wise cross-validation and 62.00% in trial-wise cross-validation. This outperforms baseline models in both recognition accuracy and generalization. Furthermore, an ablation study demonstrates a significant improvement in trust recognition performance of the spatial representation. The source code and EEGTrust dataset are available at https://github.com/CaiyueXu/EEGTrust.

Figures (5)

• Figure 1: Experimental tasks, setup and protocol for EEGTrust. (a) Overcooked-AI game paradigm. (b) Experimental setup. (c) Experimental protocol.
• Figure 2: The architecture of our proposed EEG Vision Transformer for trust recognition: (a) The 3-D spatial representation of EEG data. (b) The EEG Vision Transfomer model architecture.
• Figure 3: The distribution of participants' trust ratings in robots for different task difficulties and robot abilities. The rectangles are the $25\% \sim 75\%$ distribution of the self-assessment. The horizontal line at both ends is the range within 1.5IQR. The intermediate lines are the median line. The hollow points are the mean value. The solid points are the outliers.
• Figure 4: Mean accuracy and F1 score of each subject for trust recognition using our proposed model.
• Figure 5: ROC curve and AUC of the EEG trust recognition models with or without spatial representation. (a) Slice-wise cross-validation. (b) Trial-wise cross-validation.