I think the crucial difference is the way how the sidechain would be to be implemented, according to the information available until now in the whitepaper.

SatoshiVM is thought to be a rollup, this means it's not a completely independent sidechain but rather a method to bundle transactions so they occupy less space on L1 (but regularly injecting data on L1).

RSK and Liquid are currently federated, but afaik their goal is not to use rollup technology but eventually become completely independent sidechains with 2-way pegs, so apart from the deposit/withdrawal transactions, no information has to be stored on L1. This is of course a much more ambitious goal. RSK was originally aiming for the Drivechain 2-way peg mechanism, but they have recently moved more in the direction of a dynamic federation, as already operational on Nomic. Drivechain seems to be still in testing stage but its big problem is that it requires new Bitcoin opcodes.

I am however also a bit disappointed about the lacking level of detail of the technical explanations for SatoshiVM. While there are some docs available, they don't explain the really critical points (above all: which kind of (Taproot) script is used on BTC to block coins which are currently living on the sidechain until a withdrawal takes place?) This could be seen even as a kind of red flag. Another red flag could be the planned token distribution, which looks very much like an average ICO and thus very centralized. At its very least, it looks like a very early-stage project, even if there seems to be an usable Testnet app.

Anyway it's worthy to follow its evolution over time.