This has been tried. In the 1960s a military portable-reactor programme built and fielded units, and it ended. It did not end because the reactors did not work, and it did not end because of a safety catastrophe in the ordinary sense. It ended on economics and maintenance: the plants demanded skilled attention that was expensive to deliver where the plants were useful, and the arithmetic never closed. That is a precise and uncomfortable finding, because it is exactly the failure mode autonomy claims to fix. So the honest position for anyone entering this field is not that the graveyard is irrelevant now, but that autonomy must be shown to fix the specific thing that killed the last attempt. A demonstration that the reactor works is not a response to that history. A demonstration that it can be run unattended, and audited, is.