Paper accepted for publication in IEEE Access (10th Sep 2025)

Our follow-up work on echolocation-inspired perception is now published in IEEE Access. This work demonstrates key abilities of our method: 1) scalability to learn new shapes sequentially, 2) multi-object identification when more than one object is in the vicinity, and 3) localization of the objects. 

This work, along with the recent study that demonstrated we can synthetically train these CNN models and still process measured echoes, offers an effective sound-based sensing and perception method for potential use in imaging, detection, and navigation.