You can create several DMX cues with different X and Y position values. You can then put those on a cuelist which does manual crossfading.

There is an image speed channel built into Catalyst. Make sure to use it as a buffer or you'll end up with jumpy looking transitions.

An idea if you do not have manual xfade for the position values, try building a ten step cuelist where step 1 is 10%, step 2 is 20%, etc. Estimate the total time to move the scenery and then divide that by 10. Put the fraction you come up with for each cue. If you have the ability to use dampened fade times, also do so as it will soften any jerks.

Then when the DMX crossfades, if the movement is slow you just run one cue at a time. As things speed up, you can skip ahead on cues and the movement will speed up.

This is the same trick I used to make automated lights track to a performer. Obviously it's tedious but you'll find it will accommodate your needs.