Animals such as spiders are known to effectively hunt in total darkness with the aid of arrays of vibration sensors that detect acoustic signals received through the ground. Animals also use neural networks to analyze acoustic signals generated by motion of prey to effectively locate prey. Conventional centralized target location systems suffer from communicational and computational bottlenecks. Thus massively parallel robust decentralized neural network based target location schemes are of interest. In this paper a generic feedforward and radial basis network are trained to successfully learn to locate a target using output from a linear array of acoustic sensors. The neural network based approach is compared to a least square estimation scheme. Recent advances in the development of analog and digital hardware implementations of neural networks allow decentralized neural network based target location schemes to be implemented. A decentralized neural network based target location estimation system that avoids communication bottlenecks posed by a centralized information fusion center is proposed. The goal of this research effort is to explore possible schemes for locating a ground based target solely using acoustic signals transmitted through the ground. Such a scheme has an advantage because it does not need to emit or receive electromagnetic radiation and hence can be operated without being detected. © 2013 Praise Worthy Prize S.r.l. - All rights reserved.