Design of a 3-D Printed, Modular Lateral Line Sensory System for Hydrodynamic Force Estimation

September/Octob ize a dipole flow source (Dagamesh et al., 2013). In Abdulsadda and Tan (2013), it is shown that dipole detection and localization can also be performed using ionic polymer-metal composite (IPMC) sensors. Researchers have used microelectromechanical systems (MEMS) pressure sensors manufactured with a liquid crystal polymer (LCP) membrane in attempts to sense flow rate and direction (Kottapalli et al., 2011). An array of these sensors was later used to try to detect background flow velocity and the velocity of passing object (Kottapalli et al., 2012). MEMS sensors inspired by superficial neuromasts were suggested to be capable of detecting flow velocities; when these sensors were embedded in polydimethylsiloxane (PDMS), canal experiments suggest that the static component of the flow was filtered out (Kottapalli et al., er 2017 Volume 51 Number 5 103 2014); however, it also appears to reduce the sensitivity of the sensors. Additionally, researchers have used optical flow sensors mimicking canal neuromasts to detect flow properties (Klein & Bleckmann, 2011). While they have reasonable results, the primary disadvantage of using custommade sensors is added complexity; these sensors add difficulty to the manufacturing process and require additional calibration and testing to ensure every sensor is working. Additionally, sensors mimicking superficial neuromasts generally have a delicate sensing element that is susceptible to being damaged. Due to the complexity and cost of manufacturing custom sensors, many other researchers have used off-theshelf sensors to validate the functionalities of the lateral line. In Fernandez et al. (2011), a linear array of pressure sensors was used to determine the position, shape, and size of various objects in a flow. Chambers et al. (2014) attempted to use absolute pressure sensors to detect the turbulent wake of a cylinder; these sensors suffered from low resolution, requiring amplification and high-precision analog-to-digital converters (ADCs) to achieve moderate sensor resolution. Akanyeti et al. (2013) suggested that an artificial lateral line system can be used to monitor the speed and acceleration of a marine craft; however, the error of the method tends to scale with velocity suggesting that it is not suitable for fast-moving vehicles. In another work, a lateral line system implementing pressure sensors was used to try to detect the angle of attack of a marine craft for use in active yaw control; the system was able to achieve reasonable results after advanced filtering techniques were applied (Gao & Triantafyllou, 2012). Researchers 104 Marine Technology Society Journ attempted to use a multimodal lateral line system using both custom hair cell sensors and off-the-shelf pressure sensors to aid in rheotaxis and station holding (Devries et al., 2015). Many of the aforementioned works used absolute and gauge pressure sensors. The primary disadvantage of these sensors is that they have a low sensitivity when compared to differential pressure sensors (such as those we use in this work), requiring amplifier circuitry and high-precision ADCs to achieve similar results to the differential sensors. This adds additional complexity to the system and requires additional computations to process the data. This paper continues our study of the lateral line sensory system and its application to vehicle control. Previously, in Ren and Mohseni (2012), we developed an airfoil model of a fish in a vortex street and showed that a lateral line is capable of sensing various parameters associated with the vortex street, including the distance from the vortices, the vortex spacing, and the vortex strength. The vortex street can represent the wake produced by a bluff body (i.e., an obstacle) in a flow, or it can represent the wake of a fish or ship (Ren & Mohseni, 2012). In Xu and Mohseni (2017), we designed and built a lateral line system capable of estimating the hydrodynamic forces acting on a body and showed that, if such a system were implemented on a vehicle, it could greatly aid in control strategies (see Figures 3(a) and 3(b) for pictures of these systems). In Ren and Mohseni (2014), we developed a model for the detection of a wall using a lateral line sensory system. We validated the force estimation and wall detection algorithms using lateral line systems with statically placed sensors (Xu & Mohseni, 2017). These systems were specially designed for validation purposes and could not be directly implemented on our group’s vehicle without major modification to its hull, leading us to design a modular lateral line system (as shown in Figure 3(c)). The current work presents a modular lateral line system designed to easily fit around any underwater vehicle. A force estimation algorithm is derived and experimentally validated. Sensor placement methods for differential sensors are also discussed. This paper is organized as follows: first, we present the primary motivation for developing the lateral line system, followed by a brief discussion of our group’s vehicle. Next, we discuss in detail the design and manufacturing process for the modular lateral line. Finally, we discuss a force estimation algorithm and present experimental validation of the said algorithm. Application to Vehicle Control Our primary motivation for developing an artificial lateral line sensory system is the various benefits it could provide for vehicle control. One such application is vehicle modeling. Traditionally, underwater vehicles are modeled by a set of coupled differential equations (Fossen, 2011) representing the governing dynamics of a 6 degree-of-freedom body, given by : η 1⁄4 J Θ η ð Þv; ð1Þ M : v þ C v ð Þv þ D v ð Þv þ g η ð Þ þ g0 1⁄4 τc þ τwave þ τwind; ð2Þ