There have only been a few attempts to design an engineering equivalent to the sensory hairs of insects and spiders [16
]. These sensors were used for air flow sensing and for aerial acoustic perception [16
]. Researchers also have built artificial lateral line neuromasts. Campenhausen et al. [19
] built an artificial SN that consisted of a needle that connected a plastic blade (the artificial cupula) with a piece of paper that partly intercepted a light beam. In standing water, this system was able to sense vertically oriented metal bars while passing. Yung et al. [25
] built a monolithically integrated array of microfabricated hot wire anemometers. The hot wires had a length of 400 µm, an elevation of 600 µm and a spacing of 1 mm. This matches the dimensions of many fish lateral lines. The anemometer system exhibited a sensitivity of about 200 μm/s flow velocity, a bandwidth of about 1 kHz, and, due to its small dimensions, a reduced interference with the flow field and with the neighbouring anemometers. Unfortunately, hot wire anemometers cannot discern flow direction. As a result, they only provide a rectified reading of oscillatory flow fields. Nevertheless, arrays of hot wire anemometers were successfully used to determine the location of a moving dipole source up to a distance equalling the length of the sensor array, to access the signature of a wake caused by an upstream object and to determine the general direction to the location of that object [25
]. Recently, Yan et al. [17
] built an ALL that consisted of superficial microfabricated ANs wrapped around a cylinder. Each of the 15 ANs consisted of a horizontal cantilever with a vertical hair (length 500 μm) attached at the distal end. At the fixed end of the cantilever was a piezoresistor. Flow that impinged upon the vertical hair created a bending torque that acted on the cantilever to induce stress concentration at the site of the piezoresistor. The induced change of resistance was converted to a voltage and thus could be used to infer the local flow velocity. Unlike sensors based on hot-wire anemometry, these sensors were directional like lateral line neuromasts, i.e., along their dominant axis they responded to water flow in both directions with equal sensitivity but with opposite polarity. The detection limit for water flow was 100 μm/s, a value comparable to the fluid velocity detection thresholds of SNs (38 to 60 μm/s [26
Some researchers [28
] have also studied the filter properties of ALLCs. To do so they measured optically particle displacements inside ALLCs in relation to particle displacements in the outside medium. According to these studies, particle displacements inside ALLCs are related linearly to those in the medium. Below 80 Hz, the ratio of particle displacement inside the canal to particle displacement in the medium roughly followed the velocity in the medium. Water displacements in the ALLC were proportional to the component of the external velocity parallel to the canal. If a cylinder is placed near to an ALLC, an externally exposed flow induces complicated flow patterns inside the canal that depended on the position of the cylinder relative to the ALLC [30
]. There was little mechanical coupling between neighbouring parts of the canal.
The present study is the first to measure the performance of ALLCs equipped with optical ANs. This has enabled us to illustrate the potential of optical ANs and of biomimetic ALLCs. In general, ALLCs can be used to measure and quantify air and water motions, but due to the different Reynolds numbers, larger flow speeds are needed in air compared to water to obtain a signal. In comparison with a pure SN-system, a canal system has several advantages: 1) Sensors situated in canals are not exposed to the external environment, thus they cause minimal interference to the external flow field; 2) the sensors are protected from the external environment and thus are less prone to physical damage; 3) the sensitivity, frequency response and dynamic amplitude range of ALLCs can easily be altered by changing canal morphology (e.g., the number, size and placement of canal pores, canal diameter, and canal compartmentalization) [31
] and 4) by building artificial canals that have tubuli [6
], the mechanical filter properties of ALLCs can be further altered in a predictable manner [32
With the aid of ALLCs equipped with optical flow sensors, we were able to detect sinusoidal water motions. The responses were phase shifted by about 90°, i.e., as expected [22
] the ALLC acted as a differentiator. According to the flow field equations [21
], at a distance of 4 cm from a vibrating sphere (1 cm diameter, 50 Hz, 237 µm p–p sphere displacement) the tangential flow amplitude is about 4 µm/s (about 12.7 nm displacement). Thus a sinusoidal signal with this amplitude could be detected with our ALLCs. Besides a stationary vibrating sphere, the ALLCs responded to linear object motions and to the vortices shed by a cylinder. A stationary sensor platform equipped with two ALLCs (one on each side) was sufficient to determine the approximate upstream position of a cylinder. And a canal equipped with at least two ANs could determine bulk flow velocity. If the variables bulk flow velocity and vortex-shedding frequency are known, then the size of the cylinder can also be calculated (cf. Eq. 2
According to a recent study an artificial cupula made out of hydrogel can improve the performance of artificial sensory hairs (ANs) by about two orders of magnitude. Minimal thresholds were as low as 2.5 μm/s [33
]. Thus, one should be able to increase further the sensitivity of ALLCs by attaching artificial cupulae to the artificial CNs.
Overall, our study shows that ALLCs equipped with optical flow sensors are effective for the detection and quantification of aerial or aquatic stimuli. ALLCs could potentially provide unprecedented sensing and control functions to underwater vehicles and platforms. If so they may be useful for guiding autonomous vehicles (in both air and water), tracking and identifying wake generators and for the measurement of air and water movements in pipes and canals. To improve the performance of ALLCs, one important step will be to reduce the size of the optical ANs using MEMS technology. This is currently being done in cooperation with the center for advanced studies (caesar) in Bonn. Smaller sensors will allow us to build smaller canal systems and thus to miniaturize the sensor platforms. Finally, the development of ALLs will facilitate fundamental studies on the fish lateral line.