In the alignment example above, you would pick the genomic alignment, not transcript, and choose "subject" start and end positions. Note that this example has 3 exons, so your BED file start would be 5186957 and end would be 5188159.
Create a BED file using the chromosome (11 in above example), chromstart (5186957), chromend (5188159), and a name for your gene (HBB). You can then upload this as a custom track in the UCSC genome browser.
You will probably want to add introns to your gene structure, so at this point you can the GenePred format instead of BED. This allows you to specify the strand, as well as introns, exons, and transcription factor binding sites.
Old question, but in the spirit of good science I will post a script that takes any genome fasta file and creates a genome BED file.
Very niche but I don't think there is a good converter out there quite yet.