眼动控制的人机交互系统

摘要目前摄像头广泛存在于小车、飞行器等移动设备当中，是信息采集与获取的最有效的载荷之一。本项目基于移动设备中的这些摄像头，开发了通过人眼来控制移动设备运动，实现采集人眼感兴趣目标的视频信息的系统。项目设计的理念是：移动设备上的摄像头就是人类的第三只眼睛，移动设备就是人类的双脚，我们要基于人类视觉系统的相关知识建立人类肉眼与移动设备之间有效的数据互联，让移动设备朝着人眼感兴趣的目标运动，并采集其清晰的视频信息。人类观察感兴趣目标的通常过程是：远远的看到一个感兴趣目标，例如这个目标就是一个人。大脑会根据人眼看到的视频信息，分析这个人所处的位置，然后我们走到这个人跟前获取他清晰的图像。这个过程中没有手的参与。目前飞速发展的机器人、小型飞行器等设备大大拓展了人类的视野，像火灾现场、深海探测、抗震救灾等人无法亲身到达的一些场景，现在也可以操纵机器进去获取第一手的视频资料。但目前操纵这些设备还都需要人手的介入，完成控制手柄等操作，这即束缚了人的双手与人类的实际观察过程相违背，并且在一些双手不能使用的场景也不适用。针对手动操作的缺点，我们开发只通过眼睛控制即可获得感兴趣目标视频信息的系统。基于眼动的控制方式有以下明显优势：首先，基于眼动的控制是最贴近人类实际感知的一种方式，解放了对双手的束缚。其次，在双手不能使用的一些场景下，本系统也可以完成相应的操作，具有更加明显的优势，例如面向双手残疾人士应用的场景。截至目前本项目开发了针对小车的通过人眼控制其运动从而获取人眼感兴趣目标视频信息的系统，后期会继续开发针对其它运动设备，例如飞行器，以及更加复杂操作，例如控制相机对于感兴趣环境区域的实时自动聚焦，的系统。本项目开发的系统将来可以被应用于特殊抗震救灾，火灾现场等特殊环境勘察、人机交互、残疾人士应用等场景。关键词：眼动控制；移动设备控制；人机交互；人类视觉系统

AbstractRecently, cameras are widely used in small moving devices such as cars, smallaircrafts and so on It is one of the most effective information collection andacquisition methods Based on the cameras on these moving devices, this projectdeveloped a system that the movements of the devices are controlled by the humaneyes to obtain the video information of interest This project is design as: camera onthe moving devices is “the third eye” of human Through establishing an effectivedata association relationship between the human eyes and “the third eye” based on therelevant knowledge of the human visual system, we can control a moving devicemoves towards our interested target, and finally achieve an effective videoinformation acquisitionWhen we human observing an interested target, the general process can bedescribed as: We first see the interested target in a distance, eg the target is a human,our brain achieve the location of this target by analyzing the visual information sentby our eyes, and then our brain sends control command to let us move to this humanand achieve more information of this target It is worth to mention that, our hands arenot involved during this process The rapid development of robot, small aircraft andother equipment has greatly expanded the human vision Scenes like the fire, deep-seaexploration, earthquake relief and others which human couldn’t personally arrived,now can get the first hand of the video data through operating the moving machinesHowever, the manipulation of these devices still need of manpower intervention tocomplete the control of handle, which is not only bound our hands and is against thehuman observation process In some applications which hands cannot be used, thismethod is also not applicable Based on the shortcomings analyzed before, wedeveloped the interested information acquisition system through eyecontrolCompared to the traditional control method, controlling based on eyemovement has the following obvious advantages: first of all, controlling based on theeye movements is the closest way to the actual human perception and releasesshackles to our hands since Secondly, in some circumstances that our hands can’t beused to control moving devices, the established system can also complete thecorresponding operation without using hands With this obvious advantage, oursystem can be used in the applications such as for the disabled person who can’tcontrol their handsUp to now, the project has finished the relevant operations thatcontrol the car by the human eyes to obtain the video information of interest In thefollowing, we will continue to expand the control equipment type from cars to othermoving equipment, such as small aircraft, as well as develop more complex operationsuch as real-time camera auto focusing to regions of interest The system establishedin this project can be used in special earthquake relief, fire rescue scene, specialenvironment reconnaissance, human-computer interaction, or applications fordisabled person and so on