駕駛員閃避突發障礙物及不同駕駛風格之腦波反應研究

Abstract

本論文探討當駕駛員夜間開車，閃避前方突如其來的障礙物時，受到不預期刺激的腦波反應。我們使用虛擬實境結合腦波量測分析來研究夜間駕車之腦波特性，此為本論文之一大創新；實驗場景為夜間直線高速公路之駕車環境，並有不預期之障礙物突然出現在駕駛員前方車道，使其產生一不預期差異之認知狀態，紀錄並觀察腦波訊號，此外另設計有警示標誌出現來提醒駕駛員將出現障礙物，進而來觀察有警示與無警示下兩者之腦波差異。我們探討兩個主題，首先是觀察駕駛員看到障礙物時的認知狀態。透過獨立成分分析將原始腦波分離為數個獨立之成分，進而使用交叉相關性分析比較各個駕駛員之間獨立成分之關聯性，實驗結果發現與心理負荷相關之腦波獨立成分來源在顱頂中央(CPz)，且於6~7Hz頻帶有功率增強之情形。其次，我們藉由駕駛員閃避障礙物之行為模式，從其駕駛車輛之行進軌跡以及方向盤轉向之資訊，辨別出不同之駕駛風格，分為過度駕駛與適當駕駛以分析不同駕駛風格之腦波特性，經由獨立成分分析後，找到可用於駕駛風格鑑別之腦波特徵，其在顱頂(Cz)位置之10Hz,20Hz頻帶中具有能量上的差異，可藉此來估測駕駛員之駕駛特性。

In this thesis, we want to study the EEG relates of surprising status and driving style. Accidents usually caused by lack of alertness and awareness have a high fatality rate especially in night driving environments. It becomes extremely dangerous in some situations such as the appearance of an unexpected obstacle in the middle of the road. Combining the technology of virtual reality (VR), a realistic driving environment is developed to provide stimuli to subjects in our research. The VR scene designed in our experiment is driving a car on the freeway at nighttime. Independent Component Analysis (ICA) is used to decompose the sources in the EEG data. ICA combined with power spectrum analysis and correlation analysis is employed to investigate the EEG activity related to surprising level and driving style. According to our experimental results, the appearance of ERP at CPz is highly correlated to the surprising status. Furthermore, the level of surprising status can be evaluated with the amplitude of the ERP. An extension analysis of driving style has also been further studied in the experiments. It is observed that the magnitudes of ERP power spectrum at 10Hz and 20Hz are different respecting to different driving styles.