Localization of Subsurface Targets using Optimal Maneuvers of Seismic Sensors

Abstract

The use of seismic waves to detect subsurface targets such as landmines is a very promising technology compared to existing methods like Ground Penetrating Radar (GPR) and Electromagnetic Induction (EMI) sensing. The fact that seismic waves induce resonance in man-made targets, and hence more scattering, gives this method a natural ability to discriminate landmines from common types of clutter like rocks, wood, etc. Reflection and resonance from the targets can be used in imaging to detect the location of targets. However, existing methods require a large number of measurements for imaging and detection, which are expensive and time consuming. To reduce the number of measurements and enable faster detections, a new sensing strategy is proposed based on optimally maneuvering sensors. The system would operate in two main modes. In search mode, the goal would be to move on top of a target using the minimum number of measurements. Once the target is found, the system would switch to a detection mode to make its final decision. The seismic sensor system is an active system, where a seismic source generates the probing pulse. The waves reflected from buried targets are collected by an array of sensors placed on the surface, and then an imaging algorithm is used to estimate the target position. The performance bounds for this position estimate are derived in terms of the Fisher information matrix (FIM). This matrix gives the dependence of the target position estimate on the array position. Based on the FIM, the next optimal array position is determined by using the theory of optimal experiments. The next array position will be the one that reduces the uncertainty of the target position estimate the most. The whole array is moved to this new position, where the same steps are repeated. In this way, the target can be localized in a few iterations.