The objective of this research was to investigate acoustic characterization techniques for seabed targets that are scalable to multiple low-cost autonomous vehicles fitted with simple hydrophone arrays and sources in response to the “detection, classification, and remediation of military munitions underwater” statement of need. Specifically, this research investigated the use of bistatic and multistatic scattering from seabed targets for target localization and characterization using unmanned marine vehicles.

Technical Approach

In the bistatic approach a region of interest is insonified by a single omnidirectional source while multiple low-cost, low-power receiver vehicles record echo data from the surrounding area. This multistatic receiver network is then used to detect and classify seafloor objects. Typical sonar systems use a monostatic approach, where the source and receiver are collocated and the echo data is primarily backscattering. If the receiver and source are not collocated, the recorded echoes are bistatic; the bistatic scattering strength from a target will depend on both the source and receiver positions, as well as frequency, target composition, target position/orientation, environment, and other factors. In the approach, an acoustic data collection payload including a hydrophone array was located on an autonomous surface vehicle (ASV) and a time-synchronized acoustic source was mounted on a separate surface platform. As the receiver vehicle progressed through the environment, the bistatic acoustic reflections from the scene (and associated seabed targets) were captured on the receiving array. Two techniques were considered in this work (Fig. 1): the case where the source position is fixed will be considered the bistatic configuration, while in the multistatic configuration the source platform is also mobile. In both cases, the intent was to feed the resulting source/receiver position dependent scattering data into signal processing algorithms for initial mapping of targets, and then into machine learning algorithms to attempt classification.

Interim Results

The project team encountered significant limitations to low-cost bistatic and multistatic target detection/classification in this project. Previous work was conducted on tightly-integrated autonomous underwater vehicle systems with custom hydrophone arrays, high-quality sound sources, and well-controlled target fields. This project attempted to replicate those results using low-cost sources, receivers, and a less well-defined target field. Simulation work also attempted to extend previous work by looking at the possibility of detecting and classifying targets using multiple vehicles driving in straight lines through a target field, intersecting the scattering radiation pattern but not fully circling the target. Both of these techniques were found to be ineffective due to different limitations: for multistatic scattering, intersecting without fully sampling the aspect-dependent radiation pattern did not provide enough information for classification. The project was re-focused on bistatic imaging and/or classification using low-cost vehicles, arrays, and sources in a concentrated field study in June 2019. Analysis of that data found that the uncertainty introduced by system noise, navigation error, and mechanical vibration on the low-cost system meant that the target detection and classification was not possible within the data set.


The hope was that this technique would be found to be an effective option for low-cost unexploded ordnance (UXO) detection/classification. The conclusion is that, at this time, the low-cost off-the-shelf systems do not provide high enough quality data to make bistatic or multistatic sensing viable for low-cost UXO detection/classification. In addition, there are fundamental challenges to the bistatic/multistatic configurations that may preclude these techniques from ever being effective as compared to the more familiar single platform monostatic configuration.

  • Acoustic,