Researchers from the National University of Singapore’s (NUS) Tropical Marine Science Institute (TMSI) have created a dolphin-like sonar device with a new echo processing technique that enables clearer underwater images compared to the traditional signal processing method of visualising sound echoes.
The new sonar processing method could have potential benefits in underwater commercial or military sonars. It could be used to scan the seabed to search for features that can be used to aid navigation. The sonar’s compactness also makes it suitable to be mounted on underwater robots for ocean exploration.
The processing method is based on the hypothesis that dolphins use prior information about their environment, apart from broadband sound pulses, to interpret their echoes. The sonar uses information on the sparsity of objects, which allows for a better interpretation of sound echoes.
According to a press statement, the new device provides a better trade-off between sonar-image clarity, the number of sensors, and the size of the sensor array used as compared to current sonars of similar size and purpose. Conventional echo processing techniques tend to fail when sensors are limited in number or widely spaced. The new sonar processing method, however, can extract information and yield image clarity even in these situations.
The researchers noticed that dolphins had the ability to scan underwater objects acoustically and match them visually, indicating that a dolphin’s sound echoes emitted off an object contain information about the object’s shape. They then recorded the echoes emitted by dolphins when scanning an object in the water.
Using their observations as a guide, the team constructed a biomimetic sonar that mimics a dolphin’s sonar system. The device, which is about the size of a dolphin’s head and measures 25 cm in width, is designed to emit sharp, impulsive clicking sounds, similar to those used by dolphins for echolocation.
The team employed three transmitters to send sounds from different directions. They then analysed the echoes produced by both the dolphin and the biomimetic sonar to visualise what information about the object’s shape was revealed in the echoes.
To complement the hardware, the team developed software that improves the visualisation of echoes. The researchers incorporated the concept of sparsity into the sonar’s software. This assumes that out of the space scanned, only a small percentage is occupied by the object. According to Hari Vishnu, Senior Research Fellow at NUS TMSI, “Using prior information, such as the idea of sparsity, is intuitive. It is something humans do all the time – we turn our understanding of reality into expectations that can speed up our inferences and decisions. For example, in the absence of other information, the human brain and vision system tend to assume that in an image, the light on an object will be falling from above.”
The effectiveness of the software was demonstrated when it was able to visualise information from a dolphin’s sonar echoes when scanning an object, as well as sonar signals produced by their compact sonar. A conventional approach to processing both sonar echoes resulted in noisy images. However, the novel processing approach gave better resolution and therefore sharper images. The software is also able to generate visualisations with a mere three clicks from the sonar, thus allowing it to be operationally fast.