UNDERWATER GLOBAL LOCALIZATION Sample Clauses

UNDERWATER GLOBAL LOCALIZATION. The current practice in underwater localization is based on sonar, motion sensing, and the occasional GPS fixes when the AUV surfaces. Each AUV uses its own sonar and gyroscope/accelerometer/propeller units to track its position when underwater, possibly aided by one or more surface “anchors” (i.e., emitting sonar beacons). This approach maintains full autonomy of each AUV, yet leaves much to be desired in terms of localization accuracy, early warning capability, fault tolerance and other attributes that are crucial in mission-critical operations. Most of our know-how in localization is based on experience gained from terrestrial localization using radio waves – from GPS to cellular radio, and, more recently, related problems in wireless sensor networks. One of the lessons learned from the latter is the importance of cooperation in ensuring accurate localization in difficult scenarios – and the underwater environment is one of the harshest in terms of multipath, Doppler, and wavefront aberration. We will therefore consider cooperative swarm localization approaches that pool together information from the different AUVs. Suppose that each AUV can measure its distance from a few other AUVs, and possibly also landmarks and beacon nodes. Individual measurements will usually be incomplete, and may not allow obtaining a unique (let alone accurate) position fix. The question is how we can combine these individual measurements to improve localization performance for all, in a way that respects the constraints (e.g., on communication rates and latency) of the underwater environment. Given pair-wise distances between certain entities, the problem of finding points in 2-D or 3-D space that generate those distances is known as Multi-Dimensional Scaling (MDS). MDS has its roots in Psychology [L1, L2, L3], but it has found abundant applications in many areas, including, for example, multimedia search and sensor network localization [L5, L6, L7]. Consider N nodes (e.g., cities, mobile phones, or AUVs), with given pair-wise distances. The problem is to find N points on the plane (or in 3-D space) whose Euclidean distances are (approximately) equal to the given pair-wise distances between the nodes. One concern is that the distance between points is invariant with respect to rotation, reflection, and shift – hence the reconstructed map will not have the customary orientation. One needs at least three (in the 2-D case) anchor nodes whose absolute coordinates are known (e.g.,...