In addition to their utility as tools for observing the natural world, we are interested in radio geophysical instruments as objects of study themselves. Airborne and snowmachine-borne ice penetrating radars offer the flexibility for parameters and surveys to be finely tuned for specific targets, areas, or processes. By contrast, satellite, drone-borne, and stationary surface-based sounders often have resource envelopes that require truly optimized designs. In addition to advancing these more-standard radar architectures, we are actively pursuing the development of heterodox, bespoke, and scalable low-cost radio-glaciological sensors. This includes systems for passive radio sounding using solar emissions and other ambient electromagnetic signals (similar to ambient noise seismology), distributed radar networks for tomographic imaging, and alternative architectures for sensor fusion, machine learning, and autonomy. Developing novel radars also provides a deeper understanding of their specific capabilities and limitations so that they can be put to more creative and productive scientific use.