The ability to identify natural landmarks on a regional scale could contribute to the navigation skills of echolocating bats and also advance the quest for autonomy in natural environments with man-made systems. However, recognizing natural landmarks based on biosonar echoes has to deal with the unpredictable nature of echoes that are typically superpositions of contributions from many different reflectors with unknown properties. The results presented here show that a deep neural network (ResNet50) was able to classify 10 different field sites and 20 different tracks (2 at each site) distributed over an area about 40 kilometers in diameter. Based on spectrogram representations of single echoes, classification accuracies up to 99.6% for different sites and 94.7% for different tracks have been achieved. Classification performance was found to depend on the used pulse component (constant-frequency - CF vs. frequency-modulated - FM) and the trade-off between time and frequency resolution in the spectrogram representations of the echoes. For the former, classification performance increased monotonically with better time resolution. For the latter, classification performance peaked at an intermediate trade-off point between time and frequency resolution indicating that both dimensions contained relevant information. Future work will be needed to further characterize the quality of the spatial information contained in the echoes, e.g., in terms of spatial resolution and potential ambiguities.

Many bat species navigate in complex, heavily vegetated habitats. To achieve this, the animal relies on a sensory basis that is very different from what is typically done in engineered systems that are designed for outdoor navigation. Whereas the engineered systems rely on data-heavy senses such as lidar, bats make do with echoes triggered by short, ultrasonic pulses. Prior work has shown that "clutter echoes" originating from vegetation can convey information on the environment they were recorded in -- despite their unpredictable nature. The current work has investigated the spatial granularity that these clutter echoes can convey by applying deep-learning location identification to an echo data set that resulted from the dense spatial sampling of a forest environment. The GPS location corresponding to the echo collection events was clustered to break the survey area into the number of spatial patches ranging from two to 100. A convolutional neural network (Resnet 152) was used to identify the patch associated with echo sets ranging from one to ten echoes. The results demonstrate a spatial resolution that is comparable to the accuracy of recreation-grade GPS operating under foliage cover. This demonstrates that fine-grained location identification can be accomplished at very low data rates.