High precision trajectory determination of an UAS by integrating camera and laser scanner data with generalised object models

提交的项目的目的是综合、厘米级精度、可靠和高频确定无人机系统(UAS)的轨迹，该系统配备了不同的摄像头和激光扫描仪。重点是显示实时潜力的方法。根据细节级别2(LoD2)的建筑模型的信息被集成到轨迹确定中。此外，它还得到了GNSS信号和IMU传感器的支持。这种轨道的可获得性为研究、开发和应用开辟了一系列新的可能性。然而，在目前的方案中不会对这些进行研究。目标确定是在连续的、稳健的参数估计以及使用特定的滤波器设计的情况下进行的，并且扩展了经典的摄影测量捆绑平差以用于实时使用和引入距离信息。该项目的主要重点是调查整合相机和激光扫描仪所带来的准确性和可靠性方面的附加值。科学上的新奇之处在于纳入了广义的物体空间知识，并同时利用对图像坐标和激光扫描仪测量的观测，以高精度估计具有实时潜力的UAS轨迹。此外，该方法可以移植到任何移动地图平台上。我们首次将传感器数据生成的3D平面与通用建筑模型的平面进行匹配，并在另一方面找到图像和激光扫描仪数据之间的对应关系。在这个过程中，物体空间信息有助于长期稳定轨迹。在没有全球导航卫星系统接收的情况下，这一点尤为重要。其他相应的飞机支持图像和激光扫描仪数据的匹配，为弹道的短期和中期稳定提供服务。基于包含独立参考信息的真实数据的模拟和实验对结果进行了验证；开发的方法将根据所获得的结果进行调整和改进。我们为该项目选择的方案是城市地形上方较低的空域，众所周知，那里有许多GNSS信号中断，因此需要一个特别精确的轨迹来防止发生事故，如与障碍物相撞。

The aim of the submitted project is the integrated, cm-precise, reliable and high frequency determination of the trajectory of an unmanned aircraft system (UAS), equipped with different cameras and laser scanners. The focus is on methods showing real-time potential. Information from a building model according to level of detail 2 (LoD2) is integrated into the trajectory determination. In addition, it is supported by GNSS signals and IMU sensors. Availability of such a trajectory opens up a whole range of new possibilities in research, development and application. However, these will not be investigated in the current proposal.Trajectory determination is carried out in a sequential, robust parameter estimation as well as using a specific filter design, and extends the classical photogrammetric bundle adjustment for real-time use and the introduction of range information. The primary focus of the project is on the investigation of the added value in terms of accuracy and reliability brought about by the integration of cameras and laser scanners. The scientific novelty lies in the incorporation of generalised object space knowledge, combined with the simultaneous exploitation of observations for image coordinates and laser scanner measurements, for the high precision estimation of an UAS trajectory with real-time potential. Moreover, the approach can be transferred to any mobile mapping platform. For the first time, we match planes in 3D, generated from the sensor data, with those of the generalised building model on the one hand, and find correspondences between image and laser scanner data on the other hand. In this process, the object space information serves to stabilise the trajectory in the long term. This is particularly important in situations without GNSS reception. Additional corresponding planes, supporting the matching of image and laser scanner data, serve the short and medium term stabilisation of the trajectory. The results are validated based on simulations and experiments with real data involving independent reference information; the developed methods will be adapted and refined according to the results obtained. The scenario we have chosen for the project is the lower air space over urban terrain, which is known to have numerous GNSS signal outages, and for which a particularly precise trajectory is needed to prevent accidents such as a collision with obstacles.