Pose Estimation of Underwater Robots Using Triaxial MI Antenna and Particle Filter



Journal Title

Journal ISSN

Volume Title



Localization is a key ability for robot navigation and collision avoidance.  The advent of GPS has led to enormous improvements in terrestrial navigation. However, localization using Electromagnetic (EM) techniques is challenging underwater.  The well-established, EM-based global position system (GPS) is infeasible underwater since EM signals rapidly attenuate underwater. This limits the achievable communication range to within a few centimeters of the surface. Instead of EM techniques, most underwater communication is performed using acoustic techniques. Although acoustic communication underwater is common, it suffers from multipath fading due to reflections with the sea surface and sea floor, and with nearby obstacles. Other, methods such as using LIDAR and vision sensors give good localization of AUVs (Autonomous Underwater Vehicle) in clear water, but perform poorly due to poor weather conditions and little to no visibility in poor lighting conditions. Magnetic induction (MI) communication has become an attractive communication method as it does not require line of sight. Two triaxial nodes, known as the anchor (transmitter) node and the sensor (receiver) node are used for MI localization tests underwater. This is performed by measuring the voltages in the receiving coils by sequentially exciting the transmitter coils. The recorded magnetic field strengths along with the data from a set of other sensors are applied to a particle filter to generate an estimate of the location of the transmitting antenna with respect to the receiving one. This approach is supported by simulations and validated through experiments in air and underwater. Accurate position information is essential for exploration tasks and localization of AUVs underwater. The deliverables from this thesis are programming code in Mathematica and {\sc Matlab} for (1) calculating the nine voltages at the receiver coils given the position and orientation of the transmitter, (2) calculating the position and orientation of the receiver given the nine voltages at the transmitter, and (3) particle filter code for tracking a robot.



Particle filter, Low variance sampling, underwater robots, Quasi - static magnetic fields