That data seems wrong, though I have seen it reported on blogs and such. FBI data on total burglaries is under 1,000,000 since 2019 (857k in 2022, the last reported year) Home invasion should be a somewhat smaller subset of that.
FBI / UCR covers only reported crimes at police agencies that forward data, and is a subset. BJS tend to be better at getting realistic numbers. (E.g. a majority of burglaries aren't reported to the police, and thus don't show up in UCR*)
Also robbery is not a superset of home invasion, due to the difference in definition from burglary. Burglary isn't actually about stealing stuff, it's just breaking in with the intent to commit some crime (which is often theft, but not always). Robbery implies theft, but doesn't require it to be in someone's abode. So there's some overlap, but there's robberies that aren't burglaries (because no breaking in is involved), and there's burglaries (even some of the <10% subset of burglaries that are violent) that aren't robberies, because a theft isn't occurring.
* For things like homicide, UCR tends to be a lot better, because the reporting rate is quite high.
Note that the majority of burglaries are unreported to the police and don't show up in the UCR. You should use the Bureau of Justice Statistics for numbers on crimes with very low reporting rates.
https://cde.ucr.cjis.gov/LATEST/webapp/#/pages/explorer/crim...