Optical Navigation Rigs
Different baselines would pose different resolutions in visual odometry. Placing number of cameras might solve the problem but that is an overkill. Thanks to the variable focal length / Optical zoom feature of the conventional cameras a two camera stereo rig with a different baseline can be realized by having a longer baseline stereo rig with variable focal length.
Using this strategy, one can get most of the work done with a single “rig”. Adding a second stereo rig set with minimum base-line at the centre of previous stereo rig. This smaller stereo rig would take over from the point when the bigger one fails to resolve depth / distance. A deviation by centimetre at 100 km would cause a miss of hundreds of meters from the intended landing target. Even with highly sophisticated terrain-comparison (Ter-Com) techniques, one can do with a broader blob / resolution from greater heights using broad features to ascertain GO / NO-GO zones instead of trying to identify exact terrain features. However as one gets closer to the lunar surface, clearly recognising terrain features becomes relevant. It’s important to understand the resolution required against the altitude so that the software of optical navigation can be optimized for computation for a given height / period of time in the descent.
Multiple comibinations of smaller, larger, overlapping rigs will have to be used to accomplish automated lunar landing using OpNav. A great illustration provided by Masten Space in their video.
Next up is an experimental set up to help us test this software on earth and methods to identify the optimal number of cameras.