動態虛擬駕車平台研發及其在駕駛員心理負擔之腦波估測與應用

Abstract

許多災害的發生都是因為缺乏警覺性，注意力不集中，或者不當的決策時機所造成的。而在單調的工作場合中，人類的警覺性跟注意力更難一直維持在清醒的狀態下，容易因失神引起意外事件。我們使用模擬駕車之平台為基礎，以非侵入式的方法，量測腦波EEG (electroencephalograph)在心理工作負荷變化的情形，並分析與估測駕駛者開車的狀況。而虛擬實境與真實世界差異，在於有成本低、費時少、外界干擾小及安全性較高之優點。而在安全的實驗環境下，可設計更多真實駕駛條件以進行駕駛行為實驗與分析研究。 我們設計一系列駕駛車輛之場景，以量測與分析駕駛者在開車時之警覺性。並探討虛擬實境場景之開發，與車輛動態的模擬研究，以虛擬實境的環境進行駕駛者之量測與分析。我們使用WTK（World ToolKit）來開發虛擬實境之場景，其中有一些使用上之限制，本文在探討如何有效地運用WTK的一些指令來完成實驗場景。並且在場景中加入推導出的動態模型如：車輛轉彎、引擎加速與懸吊系統，加強模擬駕駛之真實性。在駕駛者開車之警覺性的研究上，我們設計了容易使駕駛者疲勞的實驗，並運用獨立成分分析演算法（ICA）、時頻分析、主要元素分析（PCA）等技術，建構一駕駛員心理負擔之系統，以腦波進行駕駛員心理負擔之估測。

During recent years, public security has become an important issue to societies. Accidents caused by lack of alertness and awareness have a high fatality rate especially in monotonous driving environments, where people can’t always keep continuous alertness and attention. Many researches have reported the training and censoring of operators’ behavior relying on the actual machines, which has high demands in space, time, cost, and safety. To tackle the above dilemma, a virtual-reality (VR) based motion platform is suitable to provide various dynamic driving environments and to reduce the development cost. In this paper, the development of the VR scene and the dynamic model of vehicle are discussed. We efficiently use the WTK (World ToolKit, which is a graphic library and have some limitation) to develop our VR scenes, and we also add some dynamic models to improve the reality of simulation, such as cornering equations, acceleration of engine, and damping condition of suspension. We finally design a highway driving environment by using VR-based motion platform and propose a non-invasive EEG measurement system to demonstrate the feasibility of finding the relationship between operators’ cognitive states and responses to task. Experimental results show that our proposed VR-based simulator can successfully mimicking actual driving situation and the developed EEG-based brain computer interface can achieve a satisfactory result.