I haven't used FMRIprep, and don't know what those files look like.

Also, I could be mistaken, but I thought that FMRIprep by default uses AFNI's 3dvolreg for motion estimation. FMRIprep uses a lot of software tools across existing packages, and I think they picked AFNI's for default motion estimation, for example because of its high-rating when evaluating motion correction tools:
Oakes TR, Johnstone T, Ores Walsh KS, Greischar LL, Alexander AL, Fox AS, Davidson RJ (2005). Comparison of fMRI motion correction software tools. Neuroimage. 28(3):529-543. doi:10.1016/j.neuroimage.2005.05.058
[pubmed.ncbi.nlm.nih.gov]

Can you check that? I'm not sure what you processing you want done in FMRIprep and which in afni_proc.py, but if that is the case, then it might just make sense to do that part in afni_proc.py (which will also use 3dvolreg) rather than import with, say, this afni_proc.py option:

-regress_motion_file FILE.1D  : use FILE.1D for motion parameters

                e.g. -regress_motion_file motion.1D

            Particularly if the user performs motion correction outside of
            afni_proc.py, they may wish to specify a motion parameter file
            other than dfile_rall.1D (the default generated in the volreg
            block).

            Note: such files no longer need to be copied via -copy_files.

            If the motion file is in a remote directory, include the path,
            e.g. -regress_motion_file ../subject17/data/motion.1D .

--pt