Solution of rpolicastro is the easiest. Here a custom solution.

library(Biostrings)

#/ Some dummy data, here yeast genome
url <- "http://ftp.ensembl.org/pub/release-106/fasta/saccharomyces_cerevisiae/dna/Saccharomyces_cerevisiae.R64-1-1.dna.toplevel.fa.gz"
fasta <- readDNAStringSet(url)
names(fasta) <- gsub(" .*", "", names(fasta))

make_bins <- function(fasta, binsize){

  intervals <- 
    lapply(names(fasta), function(x){

      current <- fasta[[x]]
      div  <- length(current)/binsize
      chunks <- floor(div)
      rest <- length(current) - chunks*binsize

      if(chunks>0){

        from <- seq(0, chunks*binsize, binsize)
        to <- c(from[2:length(from)])

        if(rest==0){

          from <- from[1:(length(from)-1)]

        } else {
          to <- c(to, to[length(to)]+rest)
        }

      } else {

        from <- 0
        to <- length(current)

      }

      data.frame(chr=x, start=from, end=to)

    })

  intervals <- do.call(rbind, intervals)

  return(intervals) 

}



b <- make_bins(fasta, 50000)

head(b)

  chr  start    end
1   I      0  50000
2   I  50000 100000
3   I 100000 150000
4   I 150000 200000
5   I 200000 230218
6  II      0  50000

Since you said you want to do this in R, you would first use the GenomicRanges::tileGenome function to generate a GRanges object with your genomic tiles. This can then be exported using rtracklayer::export.

As a generic example.

library("GenomicRanges")
library("rtracklayer")

chrm_sizes <- c(I=1e9, II=2e8)
genome_tiles <- tileGenome(chrm_sizes, tilewidth=1e6)

export(genome_tiles, "genome_tiles.bed", format="bed")