Most or possibly all programs that conduct downstream analyses will accept any type of data.matrix or data.frame that has transcripts as rows and samples as columns. I don't believe that BaySeq is any exception.

You can either read your files manually into R with a loop that runs over a file listing (i.e. your Kallisto counts files), or you can use the highly recommended (by the main players in RNA-seq analyses) tximport.

Once you have your count data read in, you can then run it through BaySeq.

Kevin