Driving Intention Decoding from EMG Signals for Human-Vehicle Interaction

This paper put forward a decoding model based on electromyography (EMG) to classify intentions of emergency braking, normal driving, and soft braking. EMG signals are different in time domain and frequency domain for the three driving intentions. The potential amplitude of EMG signals in the time domain and power spectrum magnitude in the frequency domain are cascaded as features. Three binary classifiers based on regularized linear discrimination analysis (RLDA) are developed to decode the three driving intentions. Experimental results show that the proposed model performs well. This study has important reference value for the development of adaptive assistant driving systems in the future.