CONTACT US: Contact info
|
Citation |
Haufe S, Treder MS, Gugler MF, Sagebaum M, Curio G, Blankertz B. J. Neural. Eng. 2011; 8(5): 056001. |
|
Affiliation |
Machine Learning Group, Department of Computer Science, Berlin Institute of Technology, Franklinstraße 28/29, D-10587 Berlin, Germany. Bernstein Focus Neurotechnology, Berlin, Germany. |
|
Copyright |
(Copyright © 2011, Institute of Physics Publishing) |
|
DOI |
|
PMID |
21799241 |
|
Abstract |
Emergency braking assistance has the potential to prevent a large number of car crashes. State-of-the-art systems operate in two stages. Basic safety measures are adopted once external sensors indicate a potential upcoming crash. If further activity at the brake pedal is detected, the system automatically performs emergency braking. Here, we present the results of a driving simulator study indicating that the driver's intention to perform emergency braking can be detected based on muscle activation and cerebral activity prior to the behavioural response. Identical levels of predictive accuracy were attained using electroencephalography (EEG), which worked more quickly than electromyography (EMG), and using EMG, which worked more quickly than pedal dynamics. A simulated assistance system using EEG and EMG was found to detect emergency brakings 130 ms earlier than a system relying only on pedal responses. At 100 km h(-1) driving speed, this amounts to reducing the braking distance by 3.66 m. This result motivates a neuroergonomic approach to driving assistance. Our EEG analysis yielded a characteristic event-related potential signature that comprised components related to the sensory registration of a critical traffic situation, mental evaluation of the sensory percept and motor preparation. While all these components should occur often during normal driving, we conjecture that it is their characteristic spatio-temporal superposition in emergency braking situations that leads to the considerable prediction performance we observed. Language: en |