Page 31 - Winter 2020

for this niche because the focus of these systems has been to encode useful sensory information about the outside world in the acoustic domain.
In fact, the entire evolution of the bat robots from its very beginnings to the present day can be understood as a sequence of attempts to extract more and more information from the ultrasonic echo wavefields that a (bio)sonar system elicits. In the process of designing ever more sophisticated bat robots, scientists and engineers have hence pushed the boundary of what can be learned from an ultrasonic echo by mimicking more and more sophisticated features found in bat biosonar. In this article, we attempt to trace this “parallel evolution” in robots and acoustic concepts.
Evolutionary Origins: Autofocus Cameras
The oldest “evolutionary ancestor” of all bat robots actually came into being as part of a camera. In the
early 1980s, Polaroid introduced an autofocus camera with an ultrasonic ranging module that was designed to determine the distance to a subject (Biber et al., 1980). To accomplish this, the ranging module emits a short burst of ultrasound and extracts a single rather simple feature from the returning echo, namely time of flight. This means that the module measures how long it took the ultrasonic pulse to travel out to the target and back (Figure 1). Such a time-of-flight measurement can be readily converted into an estimate of target distance by multiplying with sound velocity. In the Polaroid ranging module, the time of flight is determined by comparing the received echo amplitudes with a threshold value, which is simple but also susceptible to noise.
Because there are obvious shortcomings to using an ultrasonic range finder for a camera, for example, when taking a picture through a glass screen, camera makers soon moved on to optical systems such as active infrared and passive autofocus. But the ultrasonic range-finding modules they discarded did gain an unexpected second lease on life with mobile robots, where they soon became a standard feature of several commercially available systems.
Dinosaurs: Sonar Rings
Roboticists trying to perform useful tasks with the ranging modules quickly discovered that the readings from the ranging modules cannot only be imprecise but are also prone to grave errors; targets can be missed entirely, especially in cases where a flat surface forms an acoustic mirror that is angled in a way so that it reflects the echo away from the receiver. Narrow gaps between targets can be hard to find if their width is small compared with the sonar beam that is defined as the angle over which an above-threshold ultrasonic energy is emitted (about 30° for a Polaroid range finder).
Even if the targets are correctly detected and resolved and a precise enough distance measurement is obtained, all this is still not sufficient to pin down the location of a target. For any fixed distance, a target could be anywhere on a sphere around the ranging module, with a radius that equals the determined distance. To overcome this shortcoming, roboticists working with sonar range finders have come up with “sonar rings” that combine large numbers (up to 48) of ranging modules that are pointed radially outward to take range measurements in
Figure 1. Principle of sonar range finding. A: acoustic time of flight encodes the target distance because it can be used to estimate the distance traveled to the target and back. B: amplitude thresholding operation to estimate the time of flight. The received time waveform is compared with a threshold value (orange line). C: schematic representation of a sonar ring with 24 transducers (emitters/receivers).
Winter 2020 • Acoustics Today 31