Real-time positioning a pile for accurate pile driving is desirable for modern construction foundation work, but it suffers from the deficiency of the traditional systems because surveying instruments are manually used to mark the pile positions in which the accuracy heavily depends on the worker's experience. The paper confronts this problem by proposing a highly efficient positioning system using a Laser Range Finder (LRF). Over the traditional systems ours is superior to automatically detect the position of the pile or pile driver in real time with high accuracy. To this end, we first develop LRF based surveying system to scan the construction site in real time and gather the 2D laser point data. Then we detect target object such as pile or pile driver by fast fitting a circle-like geometric model to the data based on Maximum Likelihood Estimation (MLE) inference. The performance of the algorithm is validated by both synthesized and real data set. The results demonstrate the potentials on feasibility of our method in future construction field.