This work presents the design and test of an experiment system for studying a novel CT-guided robotic needle biopsy technique. CT-guided needle biopsy is a dominant method of obtaining tissue samples from lung nodules for lung cancer diagnosis. Current practice requires patients to hold their breath during the procedure, which is not applicable to those who have difficulty in holding breath. The CT-guided robotic needle biopsy technique adapts to the patient respiratory motion pattern and uses a robot manipulator to drive the biopsy needle towards a target lung nodule with respiratory motion, which will enable biopsies on patients who have difficulty in holding breath and improve the needle placement accuracy. An offline experiment system has been created to facilitate the development and evaluation of algorithms for planning and controlling the robotic needle placement process. It consists of three subsystems coordinated by a computer, i.e. a moving lung nodule phantom subsystem which mimics a lung nodule with respiratory motion, a robotic needle manipulation subsystem which drives the biopsy needle to hit the moving lung nodule, and a real-time vision feedback subsystem which tracks the moving lung nodule phantom to provide feedback for controlling the needle placement. The results from a sequence of needle placement tests based on clinically-obtained lung nodule motion paths show that the robotic needle placement technique has the potential to provide highly accurate needle delivery on moving nodules and allow biopsy of nodules smaller than usually considered. The bending of the needle when inserted into tissue affects the accuracy of many percutaneous procedures including the lung biopsy. This issue is traditionally tackled by additional medical imaging, which is expensive. A simple method has been derived to accurately predict the position of the needle tip. An algorithm of the new method has been developed and tested with satisfactory accuracy.