Fusion of Multisensing for Autonomous Navigation

本文的研究是随着多传感器融合技术在自主车辆中应用的发展而进行的。在自主车辆上安装的一些传感器可能存在一定的缺点，如陀螺仪的漂移、转速计的模糊计数等，为了寻求合理的方法来克服这些问题，我们认为作为图像传感器的摄像机的视觉信息对各个传感器起着重要的补充作用。获得了如下重要结果。在目标跟踪问题中，视觉信息给出了目标轨迹的预测，从而实现快速跟踪。在车辆的位置测量中，发现视觉信息弥补了陀螺仪内漂和GPS系统误差的不足，为多传感器融合技术在自主车辆中的应用奠定了基础。

This investigation is considered with the development of the fusion technique of the multisensing be used for the autonomous vehicle. Some sensors installed in the autonomous vehicle may have a certain disadvantage such as the information loss effected by the drift for the gyro, the ambiguity count for the tachometer, and so on.To seek out the reasonable method to overcome the problems, we consider that the vision information from the camera as the image sensor plays the important role as complementing for each sensor.In our works, the important results are obtained as follows. In the object tracking problem, the vision information gives the prediction of the object trajectory ; this leads to achieve the rapid tracking. In the position measurement of the vehicle, it is found that the vision information complements the disadvantage of the internal drift of the gyro and the systematic error of the GPS (Global Positioning System).These results will contribute the development of the fusion technique of the multisensing be used for the autonomous vehicle.