Adaptive vision system/3D map reconstruction

本研究的目的是设计和开发一种集轨迹、位置估计和控制于一体的自由测距户外导航系统。自由测距导航系统不受特定路径的限制，适用于室外或动态环境。这项工作将结合几个传感器的读数，包括视觉、GPS、航位推算和测距仪，以开发一个混合导航平台。该研究将通过使用场景的光学、物理、几何和统计信息生成场景对象或特征的数学和计算模型来执行。然后，这些模型通过结合特定传感器的读数进行更新。在更新每个模型时，应该包含哪些传感器读数的决定将自动做出。这个决定将根据模型属性、周围环境信息、时间、天气和照明条件做出。该方法可以提高智能导航机的精度和鲁棒性，减少（或限制）长期位置漂移，从而提高智能导航机的可靠性。自由测距导航系统具有广泛的应用范围，从工业，如矿山和港口运输系统，到娱乐，如动物园引导机器人。

The objective of this research is to design and develop a free ranging outdoor navigation system that incorporates trajectory, position estimation and control. The free ranging navigation systems are not bounded to follow specific predefined paths, therefore, suitable for outdoor or dynamic environments. This work will combine the readings of several sensors including vision, GPS, dead reckoning, and rangefinder to develop a hybrid navigation platform. This research will be performed by generating mathematical and computational models of scene objects or features using optical, physical, geometrical, and statistical information of the scene. These models are then updated by incorporating particular sensors' readings. The decision of what sensors' readings should be included in updating each model will be made automatically. This decision will be made according to the model properties, surrounding environmental information, time, and weather and lighting conditions. The suggested approach can improve the reliability of intelligent navigation machines by enhancing their accuracy and robustness, and reducing (or bounding) the long-term positional drift. Free ranging navigation systems have a wide range of applications from industrial such as transporting systems for mines and harbors, to entertainment such as zoo guiding robots.