Starting with the cylinders around an arbitrary minor axis, there are some issues, but as far as the exercise itself is concerned, you've done quite well. Your ellipses/side edges are a bit hesitant at times so just continue pushing yourself on executing those marks with a confident stroke using the ghosting method. You're also checking for minor axis discrepancies quite fastidiously so good job on that front.

Your cylinders fall into two categories. You've got cylinders with a bit of convergence to the side edges, and then there's a few cylinders which have side edges that are effectively running parallel to one another. The former is fine but the latter does pose a few problems.

In this case, we do not control where the vanishing point goes - we control how that set of edges is oriented in 3d space and it is that which determines the vanishing point. There are but only a limited number of circumstances in which the vanishing point runs at infinity - it occurs when the edges of the form run perfectly perpendicular to the viewer's line of sight, going across the field of view without any slanting towards or away from the viewer. Given that this challenge has us rotating our forms freely and randomly in space, it is safe to assume that this perfect of an orientation would not occur and we should be working towards more concrete vanishing points - even if our convergences end up being very slight and gradual. Of course, this isn't an issue that came up a ton in this section but it is enough to warrant an explanation.

The reason why we ask for varied foreshortening comes down do the ways in which foreshortening manifests itself on the forms we create. It does so through the shift in scale (where the back end is smaller than the end closer to the viewer) and the shift in degree where the farther end is relatively wider. This is something students seem to understand not consciously but on a gut-feeling level, and others have trouble grasping the concept that these shifts occur in conjunction with each other.

A dramatic shift in degree with minimal shift in scale tells the viewer two contradictory things: that the length of the cylinder exists in the unseen dimension of depth, and the the length visible on the page is all there is. Similarly, a cylinder with a narrow front face but had dramatic side edges tells us that the front is facing away from us and the side is also facing away from us. Both can't be true, so we must ensure that both shifts exist together.

You seem to understand this as a whole - if not consciously, then subconsciously. Hopefully this explanation helps push this further into you conscious mind.

Moving onto your cylinders in boxes, these are similarly well done. What we're trying to do here is develop our understanding on how we construct our boxes to have proportionately square faces regardless of the box orientation. To do this, we don't actively memorize every single configuration but instead we subconsciously develop that understanding through repetition and analysis.

The box challenge was all about developing a stronger sense of how to achieve more consistent convergences by analyzing the line extensions. Here, we're just adding three more sets of line extensions: the minor axis lines (which also happen to be one of the vanishing points), and the two contact points. We can check how far off these are from the box's vanishing points and this helps us determine whether the ellipse represents a circle in 3d space, and in turn how far off the plane was from representing a square.

As a whole, you've done a solid job completing the challenge. There are a few times where your side edges were running parallel so try to avoid those in the future. Feel free to move onto the next lesson.