Demonstration of large positional drift and some basic navigation principles

To make navigation errors visible, a scenario giving a large positional drift is simulated in NavLab. However, after only one position measurement, the errors leading to the drift are estimated, leading to a very low drift for the rest of the mission.

What is shown?

The white line with the HUGIN model is the true position. The green line with the transparent HUGIN model is the real-time estimated position. The green ellipse is the 3 sigma covariance ellipse of the position estimation error.

Simulated sensors:

Large positional drift due to:

Note that a positional drift of this magnitude is not realistic for a real mission, but is used in the video for the demonstration. In a real mission with the same sensors, the large drift would be simply eliminated by using one or more position measurements during the initial phase.

Note also that the use of optimal smoothing in NavLab would remove almost all drift, also in the start of the mission (smoothed solution is not shown in the video).

What happens?

With the large initial heading error, the position error grows rapidly. The change of pitch gives a slight observability of the heading error. This results in low positional cross track drift during the descent and ripples in the position estimate (due to the combination of observability and a large uncertainty). Why an optimal real-time estimate inevitably has jumps is discussed in the NavLab paper, last bullet at page 4 (Section 3.2).

After the descent HUGIN continues horizontally, and the position error increases further. Then a position measurement is received, indicated by a red flash of the HUGIN-models. The position measurement makes the estimate jump towards the true trajectory, and makes the error ellipse very small (not visible in the video).

The position measurement makes the errors that lead to the drift observable, and the drift is thus small for the rest of the mission.