For tracking the large scale environments, Wireless Sensor Network is an efficient tool. In such environment, restriction of battery life of the sensor networks is done because of the grouping of data, utilization of sensing, computation and communication. To determine this, a mobile robot is designed to recognize the data exist in the separated sensor networks and moved on to the sink. In new data collection technique, the outcome of the data grouping process is minimized as mobile robot can be employed in the restricted range. To improve the data collection in a varying environment, Non Linear Fault Recognizer (NLFR) algorithm has been introduced and designed in this paper to construct the robot using fault models that are non-linear. Experimental evaluation has been performed to calculate the outcome of the proposed NLFR and it has been examined that the proposed NLFR algorithm improves the fault correction rate upto 25% and efficiency upto 57%. © 2015 Institute of Control, Robotics and Systems - ICROS.