The cold dark matter (DM) model predicts that every galaxy contains thousands of DM subhalos; almost all other DM models include a physical process that smooths away the subhalos. The subhalos are invisible, but could be detected via strong gravitational lensing, if they lie on the line of sight to a multiply-imaged background source, and perturb its apparent shape. We present an automated strong lens analysis framework, and scan for DM subhalos in Hubble Space Telescope imaging of 54 strong lenses. We identify two compelling DM subhalo candidates (including one previously found in SLACS0946+1006), where a subhalo is favoured after every systematic test we perform. We find that the detectability of subhalos depends upon the assumed parametric form for the lens galaxy's mass distribution. Comparing fits which assume several more complex mass models reveals
additional (generally lower mass) DM subhalo candidates
worthy of further study, and the removal of 11 false positives. We identify 44
non-detections, which are vital to building up enough statistical power to test
DM models. Future work will apply even more flexible models to the results of
this study, to constrain different DM models. Our full analysis results are
available at https://github.com/Jammy2211/autolens_subhalo.